Mammography

Number: 0584

Table Of Contents

Policy
Applicable CPT / HCPCS / ICD-10 Codes
Background
References

---

Policy

Scope of Policy

This Clinical Policy Bulletin addresses mammography.

1. Medical Necessity

   Aetna considers mammography medically necessary for the following indications when criteria are met:

   1. Annual mammography screening as a preventive service for women aged 40 and older. Annual mammography is also considered medically necessary for younger women who are judged to be at high-risk including:
      
      
      1. BRCA1 or BRCA2 mutation carrier; or
      2. Women who meet criteria for BRCA mutation testing in CPB 0227 - BRCA Testing, Prophylactic Mastectomy, and Prophylactic Oophorectomy; or
      3. Women with diagnosis of, or has first-degree relative with, one or more of the following:
         
         
         1. Bannayan-Riley-Ruvalcaba syndrome; or
         2. Cowden syndrome; or
         3. Li-Fraumeni syndrome; or
         4. Personal history of radiation to chest between ages 10 and 30 years;
   2. Annual mammography screening for BRCA positive men with gynecomastia starting at the age of 50 or 10 years before the earliest known male breast cancer in the family (whichever comes first);
   3. Mammography screening for transfeminine (male-to-female) persons who are 40 years of age or older with past or current hormone use equal to or greater than 5 years;
   4. Diagnostic mammography for members (women and men) with signs or symptoms of breast disease or history of breast cancer. Note: Diagnostic mammography is covered regardless of whether the member has preventive services benefits;
   5. Digital mammography is considered an acceptable alternative to film mammography;
   6. Digital breast tomosynthesis ("3D mammography") is considered an acceptable alternative to standard (2D) mammography;
   7. Computer-aided detection (CAD) is considered a medically necessary adjunct to mammography.

2. Experimental, Investigational, or Unproven

   Mammography is considered experimental, investigational, or unproven for the following indications (not an all-inclusive list) because the effectiveness of these approaches has not been established:

   1. Artificial intelligence (deep learning and machine learning)-based mammography for diagnosis or screening of breast cancer;
   2. Contrast-enhanced mammography for breast cancer screening;
   3. Low-dose CT combined mammography for the diagnosis of overflow breast disease (breast nipple discharge);
   4. Screening mammography for men (except for men with gynecomastia and transfeminine persons, see Section I), as the clinical benefits of such screening in men are unproven. Current guidelines from the U.S. Preventive Services Task Force and the American College of Radiology recommend such screening only for women.  Aetna considers mammography medically necessary for surveillance of men with a prior history of breast cancer;
   5. Screening mammography for other women not included in Section I because its benefits in these other women are unproven;
   6. Xeroradiography for breast imaging because this method of radiography is obsolete.

3. Related Policies

   1. CPB 0105 - Magnetic Resonance Imaging (MRI) of the Breast
   2. CPB 0227 - BRCA Testing, Prophylactic Mastectomy, and Prophylactic Oophorectomy
   3. CPB 0269 - Breast Biopsy Procedures
   4. CPB 0386 - Breast Transillumination and Electrical Impedance Scanning (EIS), and Elastography

---

Table:

CPT Codes / HCPCS Codes / ICD-10 Codes

Code	Code Description
CPT codes covered if selection criteria are met:
77061 - 77062	Digital breast tomosynthesis
77063	Screening digital breast tomosynthesis, bilateral (List separately in addition to code for primary procedure)
77065	Diagnostic mammography, including computer-aided detection (CAD) when performed; unilateral
77066	Diagnostic mammography, including computer-aided detection (CAD) when performed; bilateral
77067	Screening mammography, bilateral (2-view study of each breast), including computer-aided detection (CAD) when performed
CPT codes not covered for indications listed in the CPB:
Intelligence (deep learning and machine learning)-based mammography - no specific code
HCPCS codes covered if selection criteria are met:
G0279	Diagnostic digital breast tomosynthesis, unilateral or bilateral (List separately in addition to 77065 or 77066)
ICD-10 codes covered if selection criteria are met (Diagnostic) (not all-inclusive)::
C50.011 - C50.929	Malignant neoplasm of breast
C79.81	Secondary malignant neoplasm of breast
D05.00 - D05.92	Carcinoma in situ of breast
D24.1 - D24.9	Benign neoplasm of breast
D48.60 - D48.62	Neoplasm of uncertain behavior of breast
N60.01 - N65.1	Disorders of breast
Q85.82, Q85.83, Q85.89	Other phakomatoses, not elsewhere classified [Cowden syndrome]
ICD-10 codes covered if selection criteria are met (Screening) (not all-inclusive):
F64.8 – F64.9	Gender identity disorders [Transfeminine persons]
N62	Hypertrophy of breast
Z12.31	Encounter for screening mammogram for malignant neoplasm of breast
Z15.01	Genetic susceptibility to malignant neoplasm of breast [Li-Fraumeni syndrome]
Z80.3	Family history of malignant neoplasm of breast
Z84.81	Family history of carrier of genetic disease
Z85.3	Personal history of malignant neoplasm of breast
Z86.000	Personal history of in-situ neoplasm of breast
Z92.3	Personal history of irradiation
Low-dose combined CT mammography:
CPT codes not covered for indications listed in the CPB:
Low-dose combined CT mammography-no specific code
ICD-10 codes not covered for indications listed in the CPB (not all-inclusive):
N64.52	Nipple discharge

---

Background

A mammogram is an x-ray of the breast.  A screening mammography is one of several tools that are used for early detection of breast cancer in asymptomatic women.  Other screening tools include the clinical breast examination and breast self-examination.  Diagnostic mammography is used to diagnose breast cancer in women who have signs or symptoms of breast disease, or who has a history of breast cancer.

With screen-film mammography, 2D X-ray images of the breasts are recorded onto photographic film. Each breast is positioned and compressed between two clear plates, which are attached to a specialized camera, and pictures are taken from two directions.

With full field digital mammography (FFDM), two-dimensional (2D) X-ray images are recorded onto a computer, rather than directly onto film. The technique is the same as in screen-film mammography. Adjustments can be made during the procedure, thus reducing the need to repeat mammograms and reducing the exposure to radiation. Images of the entire breast can be captured regardless of tissue density.

Screening mammography aims to reduce morbidity and mortality from breast cancer by early detection and treatment of occult malignancies.  There is extensive evidence from a variety of well-conducted, randomized controlled studies that annual or biennial mammography is effective in reducing breast cancer mortality by 30 % in women aged 50 to 69 years.  Data on women under age 50 are less clear.  Results from the Canadian National Breast Screening Study (CNBSS) suggest that the contribution of mammography over good physical examinations to breast cancer mortality reduction may be less than has been assumed.  This observation re-emphasizes a truism of screening – that it is not necessary to detect cancers as early as possible to obtain a benefit – it is only necessary to detect them early enough.  What is early enough in any individual case is uncertain because there are insufficient outcomes data.  This has made it difficult for professional societies to develop specific mammography screening recommendations for high-risk women.

1. screening mammography in women aged 75 years or older,
2. clinical breast examination (CBE) beyond screening mammography in women aged 40 years or older, and
3. either digital mammography or magnetic resonance imaging instead of film mammography as screening modalities for breast cancer.

In addition, the USPSTF recommended against clinicians teaching women how to perform breast self-examination.

The American Medical Association (AMA), the Society of Breast Imaging (SBI), the American College of Radiology (ACR), and the American Cancer Society (ACS), all support screening with mammography and CBE beginning at age 40.  Recent recommendations from the SBI and the ACR (2010) released after the 2009 USPSTF recommendations, which recommended that average-risk women wait until age 50 to undergo screening mammography, continue to support yearly screening mammography beginning at age 40 for women at average-risk for breast cancer.  The American College of Obstetricians and Gynecologists (ACOG, 2000) supports screening with mammography beginning at age 40 and CBE beginning at age 19.  The Canadian Task Force on Preventive Health Care (CTFPHC), the American Academy of Family Physicians (AAFP), and the American College of Preventive Medicine (ACPM) recommend beginning mammography for average-risk women at age 50.  The AAFP and ACPM recommend that mammography in high-risk women begin at age 40, and AAFP recommends that all women aged 40 to 49 be counseled about the risks and benefits of mammography before making decisions about screening.

A 1997 Consensus Development Panel convened by the National Institutes of Health concluded that the evidence was insufficient to determine the benefits of mammography among women aged 40 to 49.  This panel recommended that women aged 40 to 49 should be counseled about potential benefits and harms before making decisions about mammography.  In 2001, the CTFPHC concluded there was insufficient evidence to recommend for or against mammography in women aged 40 to 49.

Organizations differ on their recommendations for the appropriate interval for mammography.  Annual mammography is recommended by AMA, ACR, and ACS.  Mammography every 1 to 2 years is recommended by AAFP, ACPM, and the CTFPHC.  ACOG recommends mammography every 1 to 2 years for women aged 40 to 49 and annually for women aged 50 and older.

On behalf of the USPSTF, Siu (2016) updated the 2009 USPSTF recommendation on screening for breast cancer. The USPSTF reviewed the evidence on the following: effectiveness of breast cancer screening in reducing breast cancer-specific and all-cause mortality, as well as the incidence of advanced breast cancer and treatment-related morbidity; harms of breast cancer screening; test performance characteristics of digital breast tomosynthesis as a primary screening strategy; and adjunctive screening in women with increased breast density.  In addition, the USPSTF reviewed comparative decision models on optimal starting and stopping ages and intervals for screening mammography; how breast density, breast cancer risk, and comorbidity level affect the balance of benefit and harms of screening mammography; and the number of radiation-induced breast cancer cases and deaths associated with different screening mammography strategies over the course of a woman's lifetime.  This recommendation applies to asymptomatic women aged 40 years or older who do not have pre-existing breast cancer or a previously diagnosed high-risk breast lesion and who are not at high risk for breast cancer because of a known underlying genetic mutation (such as a BRCA1 or BRCA2 gene mutation or other familial breast cancer syndrome) or a history of chest radiation at a young age. 

• The USPSTF recommends biennial screening mammography for women aged 50 to 74 years (B recommendation). The decision to start screening mammography in women prior to age 50 years should be an individual one. Women who place a higher value on the potential benefit than the potential harms may choose to begin biennial screening between the ages of 40 and 49 years (C recommendation).
• The USPSTF concludes that the current evidence is insufficient to assess the balance of benefits and harms of screening mammography in women aged 75 years or older (I statement).
• The USPSTF concludes that the current evidence is insufficient to assess the benefits and harms of digital breast tomosynthesis (DBT) as a primary screening method for breast cancer (I statement).
• The USPSTF concludes that the current evidence is insufficient to assess the balance of benefits and harms of adjunctive screening for breast cancer using breast ultrasonography, magnetic resonance imaging (MRI), DBT, or other methods in women identified to have dense breasts on an otherwise negative screening mammogram. (I statement).

Currently, there are no guideline recommendations from leading medical professional organizations to screen men at risk for hereditary breast cancer with mammography.  Robson (2002) explained "There are no established guidelines for screening of men at risk for hereditary breast cancer.  It is reasonable to suggest periodic self-examination and evaluation by a provider experienced in clinical breast examination.  The utility of screening mammography in men is unknown, although it is technically possible in at least some individuals."  However, diagnostic mammography may be indicated in men with a breast mass on clinical examination (American Cancer Society, 2003).

Berg et al (2008) compared the diagnostic yield, defined as the proportion of women with positive screen test results and positive reference standard, and performance of screening with ultrasound plus mammography versus mammography alone in women at elevated risk of breast cancer.  A total of 2,809 women, with at least heterogeneously dense breast tissue in at least 1 quadrant, were recruited from 21 sites to undergo mammographic and physician-performed US examinations in randomized order by a radiologist masked to the other examination results.  Reference standard was defined as a combination of pathology and 12-month follow-up and was available for 2,637 (96.8 %) of the 2,725 eligible participants.  Main outcome measures included diagnostic yield, sensitivity, specificity, and diagnostic accuracy (assessed by the area under the receiver operating characteristic curve) of combined mammography plus ultrasound versus mammography alone and the positive predictive value of biopsy recommendations for mammography plus ultrasound versus mammography alone.  A total of 40 participants (41 breasts) were diagnosed with cancer: 8 suspicious on both ultrasound and mammography, 12 on ultrasound alone, 12 on mammography alone, and 8 participants (9 breasts) on neither.  The diagnostic yield for mammography was 7.6 per 1,000 women screened (20 of 2,637) and increased to 11.8 per 1,000 (31 of 2,637) for combined mammography plus ultrasound; the supplemental yield was 4.2 per 1,000 women screened (95 % confidence interval [CI]: 1.1 to 7.2 per 1,000; p = 0.003 that supplemental yield is 0).  The diagnostic accuracy for mammography was 0.78 (95 % CI: 0.67 to 0.87) and increased to 0.91 (95 % CI: 0.84  to 0.96) for mammography plus ultrasound (p = 0.003 that difference is 0).  Of 12 supplemental cancers detected by ultrasound alone, 11 (92 %) were invasive with a median size of 10 mm (range of 5 to 40 mm; mean [SE], 12.6 [3.0] mm) and 8 of the 9 lesions (89 %) reported had negative nodes.  The positive predictive value of biopsy recommendation after full diagnostic workup was 19 of 84 for mammography (22.6 %; 95 % CI: 14.2 % to 33 %), 21 of 235 for ultrasound (8.9 %, 95 % CI: 5.6 %  to 13.3 %), and 31 of 276 for combined mammography plus ultrasound (11.2 %; 95 % CI: 7.8 % to 15.6 %).  The authors concluded that adding a single screening ultrasound to mammography will yield an additional 1.1 to 7.2 cancers per 1,000 high-risk women, but it will also substantially increase the number of false positives.

Poller et al (2012) stated that annual ultrasound screening may detect small, node-negative breast cancers that are not seen on mammography; and magnetic resonance imaging (MRI) may reveal additional breast cancers missed by both mammography and ultrasound screening.  These researchers examined supplemental cancer detection yield of ultrasound and MRI in women at elevated risk for breast cancer.  From April 2004 to February 2006, a total of 2,809 women at 21 sites with elevated cancer risk and dense breasts consented to 3 annual independent screens with mammography and ultrasound in randomized order.  After 3 rounds of both screenings, 612 of 703 women who chose to undergo an MRI had complete data.  The reference standard was defined as a combination of pathology (biopsy results that showed in-situ or infiltrating ductal carcinoma or infiltrating lobular carcinoma in the breast or axillary lymph nodes) and 12-month follow-up.  Main outcome measures included cancer detection rate (yield), sensitivity, specificity, positive-predictive value (PPV) of biopsies performed and interval cancer rate.  A total of 2,662 women underwent 7,473 mammogram and ultrasound screenings, 110 of whom had 111 breast cancer events: 33 detected by mammography only, 32 by ultrasound only, 26 by both, and 9 by MRI after mammography plus ultrasound; 11 were not detected by any imaging screen.  Among 4,814 incidence screens in the second and third years combined, 75 women were diagnosed with cancer.  Supplemental incidence-screening ultrasound identified 3.7 cancers per 1,000 screens (95 % CI: 2.1 to 5.8; p < 0.001).  Sensitivity for mammography plus ultrasound was 0.76 (95 % CI: 0.65 to 0.85); specificity, 0.84 (95 % CI: 0.83 to 0.85); and PPV, 0.16 (95 % CI: 0.12 to 0.21).  For mammography alone, sensitivity was 0.52 (95 % CI: 0.40 to 0.64); specificity, 0.91 (95 % CI: 0.90 to 0.92); and PPV, 0.38 (95 % CI: 0.28 to 0.49; p < 0.001 all comparisons).  Of the MRI participants, 16 women (2.6 %) had breast cancer diagnosed.  The supplemental yield of MRI was 14.7 per 1,000 (95 % CI: 3.5 to 25.9; p = 0.004).  Sensitivity for MRI and mammography plus ultrasound was 1.00 (95 % CI: 0.79 to 1.00); specificity, 0.65 (95 % CI: 0.61 to 0.69); and PPV, 0.19 (95 % CI: 0.11 to 0.29).  For mammography and ultrasound, sensitivity was 0.44 (95 % CI: 0.20 to 0.70, p = 0.004); specificity 0.84 (95 % CI: 0.81 to 0.87; p < 0.001); and PPV, 0.18 (95 % CI: 0.08 to 0.34; p = 0.98).  The number of screens needed to detect 1 cancer was 127 (95 % CI: 99 to 167) for mammography; 234 (95 % CI: 173 to 345) for supplemental ultrasound; and 68 (95 % CI: 39 to 286) for MRI after negative mammography and ultrasound results.  The authors concluded that the addition of screening ultrasound or MRI to mammography in women at increased risk of breast cancer resulted in not only a higher cancer detection yield but also an increase in false-positive findings.

On behalf of the American College of Physicians, Qaseem and colleagues (2019) provided advice to clinicians on breast cancer screening in average-risk women based on a review of existing guidelines; and the target patient population is all asymptomatic women with average risk for breast cancer.

• Guidance statement 1: In average-risk women aged 40 to 49 years, clinicians should discuss whether to screen for breast cancer with mammography before age 50 years.  Discussion should include the potential benefits and harms and a woman's preferences.  The potential harms outweigh the benefits in most women aged 40 to 49 years.
• Guidance statement 2: In average-risk women aged 50 to 74 years, clinicians should offer screening for breast cancer with biennial mammography
• Guidance statement 3: In average-risk women aged 75 years or older or in women with a life expectancy of 10 years or less, clinicians should discontinue screening for breast cancer
• Guidance statement 4: In average-risk women of all ages, clinicians should not use clinical breast examination to screen for breast cancer.

García-Albeniz and associates (2020) noted that randomized trials have shown that initiating breast cancer screening between ages 50 and 69 years and continuing it for 10 years decreases breast cancer mortality.  However, no trials have studied whether or when women can safely stop screening mammography.  An estimated 52 % of women aged 75 years or older undergo screening mammography in the U.S.  In a large-scale, population-based, observational study, these researchers estimated the effect of breast cancer screening on breast cancer mortality in Medicare beneficiaries aged 70 to 84 years; 2 screening strategies were used: continuing annual mammography, and stopping screening.  A total of 1,058,013 Medicare beneficiaries aged 70 to 84 years who had a life expectancy of at least 10 years, had no previous breast cancer diagnosis, and underwent screening mammography were included in this analysis.  Outcome measures included 8-year breast cancer mortality, incidence, and treatments, plus the PPV of screening mammography by age group.  In women aged 70 to 74 years, the estimated difference in 8-year risk for breast cancer death between continuing and stopping screening was -1.0 (95 % CI: -2.3 to 0.1) death per 1,000 women (hazard ratio [HR], 0.78 [CI: 0.63 to 0.95]) (a negative risk difference favors continuing).  In those aged 75 to 84 years, the corresponding risk difference was 0.07 (CI: -0.93 to 1.3) death per 1,000 women (HR, 1.00 [CI: 0.83 to 1.19]).  The authors concluded that continuing annual breast cancer screening past age 75 years did not result in substantial reductions in 8-year breast cancer mortality compared with stopping screening.

Digital Mammography

Although digital mammography has not shown greater accuracy than film mammography, it has become standard of care.  The Food and Drug Administration (FDA)'s initial market approval of digital mammography technology was based, in part, on studies that demonstrated its effectiveness in patients referred for further testing after an initial suspicious mammogram (FDA, 2004).  In that setting, the FDA found the accuracy of digital technology was similar to that of screen film.  Studies are currently underway to evaluate the effectiveness of digital mammography in screening of the general population.

Hendrick and colleagues (2008) retrospectively compared the accuracy for cancer diagnosis of digital mammography with soft-copy interpretation with that of screen-film mammography for each digital equipment manufacturer, by using results of biopsy and follow-up as the reference standard.  The American College of Radiology Imaging Network Digital Mammographic Imaging Screening Trial (DMIST) collected screening mammography studies performed by using both digital and screen-film mammography in 49,528 women (mean age of 54.6 years; range of 19 to 92 years).  Digital mammography systems from 4 manufacturers (Fischer, Fuji, GE, and Hologic) were used.  For each digital manufacturer, a cancer-enriched reader set of women screened with both digital and screen-film mammography in DMIST was constructed.  Each reader set contained all cancer-containing studies known for each digital manufacturer at the time of reader set selection, together with a subset of negative and benign studies.  For each reader set, 6 or 12 experienced radiologists attended 2 randomly ordered reading sessions 6 weeks apart.  Each radiologist identified suspicious findings and rated suspicion of breast cancer in identified lesions by using a 7-point scale.  Results were analyzed according to digital manufacturer by using areas under the receiver operating characteristic curve (AUCs), sensitivity, and specificity for soft-copy digital and screen-film mammography.  Results for Hologic digital are not presented owing to the fact that few cancer cases were available.  The implemented design provided 80 % power to detect average AUC differences of 0.09, 0.08, and 0.06 for Fischer, Fuji, and GE, respectively.  No significant difference in AUC, sensitivity, or specificity was found between Fischer, Fuji, and GE soft-copy digital and screen-film mammography.  Large reader variations occurred with each modality.  The authors concluded that no statistically significant differences were found between soft-copy digital and screen-film mammography for Fischer, Fuji, and GE digital mammography equipment.

Pisano and associates (2008) retrospectively compared the accuracy of digital versus film mammography in population subgroups of the DMIST defined by combinations of age, menopausal status, and breast density, by using either biopsy results or follow-up information as the reference standard.  For analysis, AUCs for each modality were compared within each subgroup evaluated (age less than 50 versus 50 to 64 versus greater than or equal to 65 years; dense versus non-dense breasts at mammography; and pre- or peri-menopausal versus post-menopausal status for the 2 younger age cohorts) while controlling for multiple comparisons (p < 0.002 indicated a significant difference).  All DMIST cancers were evaluated with respect to mammographic detection method (digital versus film versus both versus neither), mammographic lesion type (mass, calcifications, or other), digital machine type, mammographic and pathologic size and diagnosis, existence of prior mammographic study at time of interpretation, months since prior mammographic study, and compressed breast thickness.  A total of 33 centers enrolled 49,528 women.  Breast cancer status was determined for 42,760 women, the group included in this study.  Pre- or peri-menopausal women younger than 50 years who had dense breasts at film mammography comprised the only subgroup for which digital mammography was significantly better than film (AUCs, 0.79 versus 0.54; p = 0.0015).  Breast Imaging Reporting and Data System-based sensitivity in this subgroup was 0.59 for digital and 0.27 for film mammography.  AUCs were not significantly different in any of the other subgroups.  For women aged 65 years or older with fatty breasts, the AUC showed a non-significant tendency toward film being better than digital mammography (AUCs, 0.88 versus 0.70; p = 0.0025).  The authors concluded that digital mammography performed significantly better than film for pre- and peri-menopausal women younger than 50 years with dense breasts, but film tended non-significantly to perform better for women aged 65 years or older with fatty breasts.

Screening film mammography has been shown to reduce the mortality rate from breast cancer; however, conventional mammography does not detect all breast cancers.  A significant factor contributing to the limitations of mammography is the structure overlap that results on a 2-dimensional mammogram.  Structure overlap not only obscures lesions, but can mimic abnormalities, thus contributing to reductions in both the sensitivity and specificity of mammography.  A number of new imaging techniques and enhancements for digital mammography have recently become available or are likely to become available in the near future.

Computer-Aided Detection and Diagnosis (CAD)

Computer-Aided Detection (CAD) involves computer software used by radiologists to assist in the interpretation and identification of suspicious findings on mammogram MRI, or ultrasound of the breast. CAD is not intended to be used in place of a radiologist but as a second set of eyes when examining the images.

1. to improve radiologists' ability to identify suspicious areas that may otherwise be overlooked on screening mammograms (detection), and
2. to distinguish between benign and malignant lesions (diagnosis). 

The radiologist remains the reader and interpreter of the mammogram.  CAD assists the radiologist by identifying areas warranting further review.

The American Cancer Society breast cancer screening guidelines (Smith et al, 2003) indicate that for digital mammography and CAD there is "some clinical evidence for effectiveness or equivalence to screen-film mammography for screening" (Evidence Level B).

Bennett et al (2006) assessed current evidence to ascertain if the accuracy of single reading with CAD compares with that of double reading.  These researchers performed a literature review to identify studies where both protocols had been investigated and compared.  They identified 8 studies that compared single reading with CAD against double reading, of which 6 reported on comparisons of both sensitivity and specificity.  Of the 6 studies identified, 3 showed no differences in either sensitivity or specificity; 1 showed single reading with CAD had a higher sensitivity at the same specificity; 1 showed that single reading with CAD had a higher specificity at the same sensitivity.  However, 1 study, in a real-life setting, showed that single reading with CAD had a higher sensitivity but a lower specificity.  The authors concluded that as the majority of the studies were not in a real-life setting, used test sets, lacked sufficient training in the use of CAD and simulated double reading (using a protocol of recall if one suggests), current evidence is therefore limited as to the accuracy, in terms of sensitivity and specificity, of single reading with CAD in comparison with the most common practice in the United Kingdom of double reading using a protocol of consensus or arbitration.

Fenton and colleagues (2007) stated that CAD identifies suspicious findings on mammograms to assist radiologists.  Since the FDA approved the technology in 1998, it has been disseminated into practice, but its effect on the accuracy of interpretation is unclear.  These investigators determined the association between the use of CAD at mammography facilities and the performance of screening mammography from 1998 through 2002 at 43 facilities in 3 states.  They had complete data for 222,135 women (a total of 429,345 mammograms), including 2,351 women who received a diagnosis of breast cancer within 1 year after screening.  They calculated the specificity, sensitivity, and positive predictive value of screening mammography with and without CAD, as well as the rates of biopsy and breast-cancer detection and the overall accuracy, measured as the area under the receiver-operating-characteristic (ROC) curve.  A total of 7 facilities (16 %) implemented CAD during the study period.  Specificity decreased from 90.2 % before implementation to 87.2 % following implementation (p < 0.001), the positive predictive value decreased from 4.1 % to 3.2 % (p = 0.01), and the rate of biopsy increased by 19.7 % (p < 0.001).  The increase in sensitivity from 80.4 % before implementation of CAD to 84.0 % after implementation was insignificant (p = 0.32).  The change in the cancer-detection rate (including invasive breast cancers and ductal carcinomas in situ) was insignificant (4.15 cases per 1,000 screening mammograms before implementation and 4.20 cases after implementation, p = 0.90).  Analyses of data from all 43 facilities showed that the use of CAD was associated with significantly lower overall accuracy than was non-use (area under the ROC curve, 0.871 versus 0.919; p = 0.005).  The authors concluded that the use of CAD is associated with reduced accuracy of interpretation of screening mammograms.  The increased rate of biopsy with the use of CAD is not clearly associated with improved detection of invasive breast cancer.  This is in agreement with Bazzocchi et al (2007) who noted that there is still considerable variation among different studies in the level of benefit deriving from CAD.  Thus, the role of these systems in clinical practice is still debated, and their real contribution to the overall management of the diagnostic process is still unclear.

An assessment of CAD in mammography screening by the Swedish Council on Health Technology Assessment (SBU, 2011) concluded that the scientific evidence is insufficient to determine whether CAD plus single reading by one breast radiologist would yield results that are at least equivalent to those obtained in standard practice, i.e. double reading where two breast radiologists independently read the x-ray images.

Xeroradiography

Xeroradiography (Xerox Corporation, Stamford, CT) is an outmoded X-ray imaging method that had been used especially in mammographic screening for breast cancer.  With Xeroradiography, X-rays pass through the body to an X-ray-sensitive metal plate.  The plate is then processed through a unique photocopying-type machine, and the X-ray image transferred to paper rather than X-ray film.  Unlike photographically recorded X-ray images, Xeroradiographs produce a "positive" image in which the denser elements appear darker.  Unlike X-ray films, the Xeroradiographic image is a mirror image of the object.  The primary advantage of Xeroradiography over conventional plain film mammography is that the former produces instant radiographs.  However, Xeroradiography has become outmoded because the radiation exposure required is much higher than with conventional radiographs, and the Xeroradiographic image has a number of defects, such as excessive contrast and edge enhancement.

Breast Tomosynthesis (3D Mammography)

Breast tomosynthesis is a 3-dimensional (3D) imaging technique based on FFDM that involves acquiring images of a stationary compressed breast at multiple angles during a short scan.  During an examination, the individual’s breast is positioned and compressed as with a standard mammogram. An X-ray tube moves along an arc around the breast to acquire multiple image slices of approximately one milliliter or less in about 10 seconds. A computer processes the series of slices and displays the data on a workstation. The individual images are then reconstructed into a series of thin, high-resolution slices that can be displayed individually or in a dynamic cine mode.  While holding the breast stationary, images are acquired at a number of different x-ray source angles.  Objects at different heights in the breast project differently for each angle.  The data are then reconstructed to generate images that enhance objects from a given height by appropriate shifting of the projections relative to one another. 

1. detector efficiency and dose,
2. field of view, and
3. equipment geometry. 

While holding the breast stationary, an x-ray tube is rotated over a limited angular range and a series of low-dose exposures are made every few degrees, creating a series of digital images.  The x-ray tube is rotated about +/-15 degrees, and 11 exposures are made every 3 degrees during a total scan of a few seconds.  These individual images are then reconstructed into slices.  There are 2 basic tomosynthesis system designs that differ in the motion of the detector during acquisition.  One method moves the detector in concert with the x-ray tube so as to maintain the shadow of the breast on the detector.  An altenate method is to keep the detector stationary relative to the breast platform. The tomosynthesis reconstruction process consists of computing high-resolution images whose planes are parallel to the breast support plates.  These images are reconstructed with slice separation of 1 mm; thus, a 5-cm compressed breast tomosynthesis study will have 50 reconstructed slices.  The reconstructed tomosynthesis slices can be displayed similarly to computed tomography reconstructed slices.  Proponents of breast tomosynthesis hope it will resolve many of the tissue overlap reading problems that are a major source of recalls and additional imaging in 2-D mammography examinations (Smith, 2005). Studies suggest that breast tomosynthesis has comparable or superior image quality to that of film-screen mammography and has the potential to decrease the recall rate when used adjunctively with digital screening mammography (Good et al, 2008; Chen et al, 2007; Poplack et al, 2007). 

In February 2011, the FDA approved the Selenia Dimensions 3D System (Digital Breast Tomosynthesis), a mammography device that provides digital 2D and 3D images for the screening and diagnosis of breast cancer.

Spangler and colleagues (2011) compared the ability of digital breast tomosynthesis and full field digital mammography (FFDM) to detect and characterize calcifications.  A total of 100 paired examinations were performed utilizing FFDM and digital breast tomosynthesis: 20 biopsy-proven cancers, 40 biopsy-proven benign calcifications, and 40 randomly selected negative screening studies were retrospectively reviewed by 5 radiologists in a crossed multi-reader multi-modal observer performance study.  Data collected included the presence of calcifications and forced Breast Imaging, Reporting and Data System (BI-RADS) scores.  Receiver operator curve analysis using BI-RADS was performed.  Overall calcification detection sensitivity was higher for FFDM (84 % [95 % CI: 79 % to 88 %]) than for digital breast tomosynthesis (75 % [95 % CI: 70 % to 80 %]).  In the cancer cohort, 75 (76 %) of 99 interpretations identified calcification in both modes.  Of those, a BI-RADS score less than or equal to 2 was rendered in 3 (4 %) and 9 (12 %) cases with FFDM and digital breast tomosynthesis, respectively.  In the benign cohort, 123 (62 %) of 200 interpretations identified calcifications in both modes.  Of those, a BI-RADS score greater than or equal to 3 was assigned in 105 (85 %) and 93 (76 %) cases with FFDM and digital breast tomosynthesis, respectively.  There was no significant difference in the non-parametric computed AUC using the BI-RADS scores (FFDM, AUC = 0.76 and SD = 0.03; digital breast tomosynthesis, AUC = 0.72 and SD = 0.04 [p = 0.1277]).  The authors concluded that in this small data set, FFDM appears to be slightly more sensitive than digital breast tomosynthesis for the detection of calcification.  However, diagnostic performance as measured by AUC using BI-RADS was not significantly different. With improvements in processing algorithms and display, digital breast tomosynthesis could potentially be improved for this purpose.

Sahiner et al (2012) designed a computer-aided detection (CADe) system for clustered micro-calcifications in reconstructed DBT volumes and performed a preliminary evaluation of the CADe system.  Institutional review board approval and informed consent were obtained in this study.  A data set of 2-view DBT of 72 breasts containing micro-calcification clusters was collected from 72 subjects who were scheduled to undergo breast biopsy.  Based on tissue sampling results, 17 cases had breast cancer and 55 were benign.  A separate data set of 2-view DBT of 38 breasts free of clustered micro-calcifications from 38 subjects was collected to independently estimate the number of false-positives (FPs) generated by the CADe system.  A radiologist experienced in breast imaging marked the biopsied cluster of micro-calcifications with a 3D bounding box using all available clinical and imaging information.  A CADe system was designed to detect micro-calcification clusters in the reconstructed volume.  The system consisted of pre-screening, clustering, and FP reduction stages.  In the pre-screening stage, the conspicuity of micro-calcification-like objects was increased by an enhancement-modulated 3D calcification response function.  An iterative thresholding and 3D object growing method was used to detect cluster seed objects, which were used as potential centers of micro-calcification clusters.  In the cluster detection stage, micro-calcification candidates were identified using a second iterative thresholding procedure, which was applied to the signal-to-noise ratio (SNR) enhanced image voxels with a positive calcification response.  Starting with each cluster seed object as the initial cluster center, a dynamic clustering algorithm formed a cluster candidate by including micro-calcification candidates within a 3D neighborhood of the cluster seed object that satisfied the clustering criteria.  The number, size, and SNR of the micro-calcifications in a cluster candidate and the cluster shape were used to reduce the number of FPs.  The pre-screening stage detected a cluster seed object in 94 % of the biopsied micro-calcification clusters at a threshold of 100 cluster seed objects per DBT volume.  After clustering, the detection sensitivity was 90 % at 15 marks per DBT volume.  After FP reduction, at 85 % sensitivity, the average number of FPs estimated using the data set containing micro-calcification clusters was 3.8 per DBT volume, and that estimated using the data set free of micro-calcification clusters was 3.4.  The detection performance for malignant micro-calcification clusters was superior to that for benign clusters.  The authors concluded that these findings indicated the feasibility of the 3D approach to the detection of clustered micro-calcifications in DBT and that the newly designed enhancement-modulated 3D calcification response function is promising for pre-screening.  They stated that further work is needed to assess the generalizability of this approach and to improve its performance.

Skaane and colleagues (2012) compared DM and DBT in a side-by-side feature analysis for cancer conspicuity, and examined if there is a potential additional value of DBT to standard state-of-the-art conventional imaging work-up with respect to detection of additional malignancies.  The study had ethics committee approval.  A total of 129 women underwent 2D DM including supplementary cone-down and magnification views and breast ultrasonography if indicated, as well as DBT.  The indication for conventional imaging in the clinical setting included a palpable lump in 30 (23 %), abnormal mammographic screening findings in 54 (42 %), and surveillance in 45 (35 %) of the women.  The women were examined according to present guidelines, including spot-magnification views, ultrasonography, and needle biopsies, if indicated.  The DBT examinations were interpreted several weeks after the conventional imaging without knowledge of the conventional imaging findings.  In a later session, 3 radiologists performed a side-by-side feature analysis for cancer conspicuity in a sample of 50 cases.  State-of-the-art conventional imaging resulted in needle biopsy of 45 breasts, of which 20 lesions were benign and a total of 25 cancers were diagnosed.  The remaining 84 women were dismissed with a normal/definitely benign finding and without indication for needle biopsy.  The subsequent DBT interpretation found suspicious findings in 4 of these 84 women, and these 4 women had to be called back for repeated work-up with knowledge of the tomosynthesis findings.  These delayed work-ups resulted in 2 cancers (increasing the cancer detection by 8 %) and 2 FP findings.  The side-by-side feature analysis showed higher conspicuity scores for tomosynthesis compared to conventional 2D for cancers presenting as spiculated masses and distortions.  The authors concluded that DBT is a promising new technique.  The authors’ preliminary clinical experience showed that there is a potential for increasing the sensitivity using this new technique, especially for cancers manifesting as spiculated masses and distortions.

Bernardi et al (2012a) prospectively evaluated the effect of integrating 3D mammography as a triage to assessment in 158 consecutive recalls to assessment (recalled in standard 2D-mammographic screening) in asymptomatic subjects.  Radiologists provided 3D mammography-based opinion as to whether recall/assessment was warranted or unnecessary, and all subjects proceeded to assessment.  3D triage was positive (confirmed the need for assessment) in all 21 subjects with breast cancer (there were no false-negatives), and would have avoided recall in 102 of 137 (74.4 %) subjects with a negative/benign final outcome in whom 3D triage did not recommend recall.  Proportion of true negative 3D triage (as a proxy for potential reduction in recalls) was slightly higher in dense than non-dense breasts, did not differ across age-groups, but was significantly associated with the type of lesion seen on imaging (being highest for distortions, asymmetric densities, and lesions with ill-defined margins).  While the simulation design may have over-estimated the potential for 3D mammography triage to reduce recalls, this study clearly demonstrates its capability to improve breast screening specificity and to reduce recall rates.  The authors stated that future studies of 3D mammography should further assess its role as a recall-reducing strategy in screening practice and should include formal cost-analysis.

Bernardi et al (2012b) supplemented the information available on logistical aspects of the application of 3D mammography in breast screening.  These investigators prospectively examined the effect on radiographers' and radiologists' workload of implementing 3D mammography in screening by comparing image acquisition time and screen-reading time for 2D mammography with that of combined 2D+3D mammography.  Radiologists' accuracy was also calculated.  Average acquisition time (measured from start of first-view breast positioning to compression release at completion of last view) for 7 radiographers, based on 20 screening examinations, was longer for 2D+3D (4 min 3 s; range of 3 min 53 s to 4 min 18 s) than 2D mammography (3 min 13 s; range of 3 min 0 s to 3 min 26 s; p < 0.01).  Average radiologists' reading time per screening examination (3 radiologists reading case-mix of 100 screens: 10 cancers, 90 controls) was longer for 2D+3D (77 s; range of 60 to 90 s) than for 2D mammography (33 s; range of 25 to 46 s; p < 0.01).  2D+3D screen-reading was associated with detection of more cancers and with substantially fewer recalls than 2D mammography alone.  The authors concluded that relative to standard 2D mammography, combined 2D+3D mammography prolongs image acquisition time and screen-reading time (at initial implementation), and appears to be associated with improved screening accuracy.  They stated that these findings provided relevant information to guide larger trials of integrated 3D mammography (2D+3D) and its potential implementation into screening practice.

Houssami and Skaane (2013) stated that DBT, a 3D derivative of DM, reduces the effect of tissue superimposition and may improve mammographic interpretation.  These investigators examined the evidence on the accuracy of DBT in clinical studies.  Published studies of DBT were relatively small studies, mostly test-set observer (reader) studies or clinical series that included symptomatic and screen-recalled cases, and were generally enriched with cancers.  With these limitations in mind, the evidence showed some consistent findings, summarized as follows: 2-view DBT has at least equal or better accuracy than standard 2-view DM, whereas 1-view DBT does not have better accuracy than standard DM; the addition of DBT to standard mammography (for mammographic interpretation or for assessment or triage of screen-recalled abnormalities) increases accuracy; improved accuracy from using DBT (relative to, or added to, DM) may be due to increased cancer detection or due to reduced false positive recalls, or both; and subjective interpretation of cancer conspicuity consistently found that cancers were equally or more conspicuous on DBT relative to DM.  Preliminary data from population screening trials suggested that the integration of DBT with conventional DM (screen-reading using combined 2D + 3D mammography) may substantially improve breast cancer detection, although final results are not yet available, and many logistical issues need further evaluation to determine the potential implications and cost of combined 2D + 3D mammographic screening.  At present, there is insufficient evidence to justify a change from standard DM to DBT however the available data strongly support investment in new large-scale population screening trials.  These trials need to avoid the 'double' acquisitions required for 2D + 3D mammograms, and should therefore focus on evaluating integrated 2Dsynthetic + 3D mammography (where 2D-images are reconstructed from the DBT acquisition), and should consider using a randomized design.

Friedewald et al (2014) determined if mammography combined with tomosynthesis is associated with better performance of breast screening programs in the United States.  The authors concluded that addition of tomosynthesis to digital mammography was associated with a decrease in recall rate and an increase in cancer detection rate.  Moreover, they stated that further studies are needed to assess the relationship to clinical outcomes.

Gartner et al (2014) discussed clinical applications of digital breast tomosynthesis (DBT) in both screening and diagnostic settings.  The authors concluded that DBT is a promising new technology that has shown improved accuracy for screening and diagnostic breast imaging.  Their early clinical experience supported these findings.  One year after implementing DBT for all screening patients, these researchers demonstrated a substantial reduction in their overall callback rate and a trend toward increased cancer detection.  They stated that "As with any new technology, several issues must be considered when implementing DBT into daily practice.  Ongoing large-scale prospective trials will help guide the evidence-based utilization of this new technology".

Garcia-Leon et al (2015) estimated and compared the diagnostic validity of tomosynthesis and digital mammography for screening and diagnosing breast cancer.  These investigators systematically searched MedLine, EMBASE, and Web of Science for the terms breast cancer, screening, tomosynthesis, mammography, sensitivity, and specificity in publications in the period comprising June 2010 through February 2013.  They included studies on diagnostic tests and systematic reviews.  Two reviewers selected and evaluated the articles.  They used QUADAS 2 to evaluate the risk of bias and the NICE criteria to determine the level of evidence.  They compiled a narrative synthesis.  Of the 151 original studies identified, these investigators selected 11 that included a total of 2,475 women.  The overall quality was low, with a risk of bias and follow-up and limitations regarding the applicability of the results.  The level of evidence was not greater than level II.  The sensitivity of tomosynthesis ranged from 69 % to 100 % and the specificity ranged from 54 % to 100 %.  The negative likelihood ratio was good, and this makes tomosynthesis useful as a test to confirm a diagnosis.  One-view tomosynthesis was no better than 2-view digital mammography, and the evidence for the superiority of 2-view tomosynthesis was inconclusive.  The authors concluded that the results for the diagnostic validity of tomosynthesis in the diagnosis of breast cancer were inconclusive and there were no results for its use in screening.

1. 2D or
2. 2D + DBT or
3. synthetic 2D + DBT images for each case without access to original screening mammograms or prior examinations. 

Sensitivities and specificities were calculated for each reading arm and by subgroup analyses.  Data were available for 7,060 subjects comprising 6,020 (1,158 cancers) assessment cases and 1,040 (2 cancers) family history screening cases.  Overall sensitivity was 87 % [95 % confidence interval (CI): 85 % to 89 %] for 2D only, 89 % (95 % CI: 87 % to 91 %) for 2D + DBT and 88 % (95 % CI: 86 % to 90 %) for synthetic 2D + DBT.  The difference in sensitivity between 2D and 2D + DBT was of borderline significance (p = 0.07) and for synthetic 2D + DBT there was no significant difference (p = 0.6).  Specificity was 58 % (95 % CI: 56 % to 60 %) for 2D, 69 % (95 % CI 67 % to 71 %) for 2D + DBT and 71 % (95 % CI: 69 % to 73 %) for synthetic 2D + DBT.  Specificity was significantly higher in both DBT reading arms for all subgroups of age, density and dominant radiological feature (p < 0.001 all cases).  In all reading arms, specificity tended to be lower for micro-calcifications and higher for distortion/asymmetry.  Comparing 2D + DBT to 2D alone, sensitivity was significantly higher: 93 % versus 86 % (p < 0.001) for invasive tumors of size 11 to 20 mm.  Similarly, for breast density 50 % or more, sensitivities were 93 % versus 86 % (p = 0.03); for grade 2 invasive tumors, sensitivities were 91 % versus 87 % (p = 0.01); where the dominant radiological feature was a mass, sensitivities were 92 % and 89 % (p = 0.04).  For synthetic 2D + DBT, there was significantly (p = 0.006) higher sensitivity than 2D alone in invasive cancers of size 11 to 20 mm, with a sensitivity of 91 %.  The authors concluded that the specificity of DBT and 2D was better than 2D alone; but there was only marginal improvement in sensitivity.  The performance of synthetic 2D appeared to be comparable to standard 2D.  If these results were observed with screening cases, DBT and 2D mammography could benefit to the screening program by reducing the number of women recalled unnecessarily, especially if a synthetic 2D mammogram were used to minimize radiation exposure.  They stated that further research is required into the feasibility of implementing DBT in a screening setting, prognostic modeling on outcomes and mortality, and comparison of 2D and synthetic 2D for different lesion types.

Tagliafico et al (2016) noted that debate on adjunct screening in women with dense breasts has followed legislation requiring that women be informed about their mammographic density and related adjunct imaging. Ultrasound or tomosynthesis can detect breast cancer (BC) in mammography-negative dense breasts, but these modalities have not been directly compared in prospective trials.  These researchers conducted a trial of adjunct screening to compare, within the same participants, incremental BC detection by tomosynthesis and ultrasound in mammography-negative dense breasts.  This trial is a prospective multi-center study recruiting asymptomatic women with mammography-negative screens and dense breasts.  Eligible women had tomosynthesis and physician-performed ultrasound with independent interpretation of adjunct imaging.  Outcome measures included cancer detection rate (CDR), number of FP recalls, and incremental CDR for each modality; these were compared using McNemar's test for paired binary data in a pre-planned interim analysis.  Among 3,231 mammography-negative screening participants (median age of 51 years; interquartile range [IQR] of 44 to 78 years) with dense breasts, 24 additional BCs were detected (23 invasive): 13 tomosynthesis-detected BCs (incremental CDR, 4.0 per 1,000 screens; 95 % CI: 1.8 to 6.2) versus 23 ultrasound-detected BCs (incremental CDR, 7.1 per 1,000 screens; 95 % CI: 4.2 to 10.0), p = 0.006.  Incremental FP recall occurred in 107 participants (3.33 %; 95 % CI: 2.72 % to 3.96 %); FP recall (any testing) did not differ between tomosynthesis (FP = 53) and ultrasound (FP = 65), p = 0.26; FP recall (biopsy) also did not differ between tomosynthesis (FP = 22) and ultrasound (FP = 24), p = 0.86.  The authors concluded that the Adjunct Screening With Tomosynthesis or Ultrasound in Women With Mammography-Negative Dense Breasts' interim analysis showed that ultrasound has better incremental BC detection than tomosynthesis in mammography-negative dense breasts at a similar FP-recall rate.  However, they stated that future application of adjunct screening should consider that tomosynthesis detected more than 50 % of the additional BCs in these women and could potentially be the primary screening modality.

In an editorial that accompanied the afore-mentioned study by Tagliafico et al, Berg (2016) stated that "Because the primary goal of screening is detection of early breast cancer, US would seem the clear choice compared with tomosynthesis …. On the basis of the results from ASTOUND, tomosynthesis still misses a substantial number of invasive cancers in women with dense breasts: supplemental US after tomosynthesis would still be reasonable, although further study is warranted …. For women with dense breasts given the choice of US or tomosynthesis, US shows more cancers. Further validation of these results is critically needed, as is longer-term follow-up to compare incidence screening results for tomosynthesis and US".

Houssami and Turner (2016) stated that high breast tissue density increases BC risk, and the risk of an interval BC in mammography screening.  Density-tailored screening has mostly used adjunct imaging to screen women with dense breasts, however, the emergence of DBT provides an opportunity to steer density-tailored screening in new directions potentially obviating the need for adjunct imaging.  These investigators summarized data on DBT screening in women with heterogeneously dense or extremely dense breasts, with the aim of estimating incremental (additional) BC detection attributed to DBT in comparison with standard 2D-mammography.  Meta-analyzed data from prospective trials comparing these mammography modalities in the same women (n = 10,188) in predominantly biennial screening showed significant incremental BC detection of 3.9/1,000 screens attributable to DBT (p < 0.001).  Studies comparing different groups of women screened with DBT (n = 103,230) or with 2D-mammography (n = 177,814) yielded a pooled difference in BC detection of 1.4/1,000 screens representing significantly higher BC detection in DBT-screened women (p < 0.001), and a pooled difference for recall of -23.3/1,000 screens representing significantly lower recall in DBT-screened groups (p < 0.001), than for 2D-mammography.  The authors noted that enhanced BC detection from DBT, in comparison with mammography alone, or indeed from adjunctive screening in dense breasts, has not been examined in studies evaluating screening benefit in terms of mortality reduction; therefore, increased BC detection can only be taken as indicative of the potential to improve screening outcomes and/or to reduce the risk of interval BC.  They stated that although evidence of improved screening benefit is currently lacking for density-tailored screening, what makes the issue of DBT screening for dense breasts immediately relevant in breast imaging and screening practice is the implementation of density-specific legislation in some settings, plus the reality of using DBT alone for primary screening.  Since DBT acquisitions can now also be used to reconstruct high-definition 2D images, yielding similar BC detection as that from DBT plus 2D-mammography acquisitions, it seems reasonable that the next phase of DBT research and practice will increasingly adopt DBT-only acquisitions (with reconstructed 2D images).  Therefore, a new direction in research is needed to evaluate DBT alone as the primary screening modality for women known to have mammography-dense breasts, potentially obviating the need to add other imaging tests to screen dense breasts.  The estimates provided in this review can be used to plan estimates of effect for future studies and trials of DBT for density-tailored BC screening.

In a retrospective study, Aujero and colleagues (2017) compared the clinical performance of synthesized two-dimensional (s2D) mammography combined with DBT with that of FFDM alone and FFDM combined with DBT in a large community-based screening population by analyzing recall rate, PPV, and cancer detection rate.  A total of 78,810 screening mammograms from October 11, 2011, to June 30, 2016, were retrospectively collected.  Of these, 32,076 were FFDM, 30,561 were DBT-FFDM, and 16,173 were DBT-s2D mammograms.  Diagnostic performance of FFDM, DBT-FFDM, and DBT-s2D mammography was compared.  Statistical significance was determined by using the Pearson χ2 test and was expressed as odds ratios (ORs)and related CIs determined by means of logistic regression analysis with pair-wise comparisons.  Recall rates were significantly lower with DBT-s2D mammography (4.3 %, 687 of 16,173) when compared with DBT-FFDM (5.8 %, 1,785 of 30,561; OR, 0.72; 95 % CI: 0.65 to 0.78; p < 0.0001) and when compared with FFDM alone (8.7 %, 2,799 of 32,076; OR, 0.46; 95 % CI: 0.43 to 0.51).  The cancer detection rate was similar among FFDM alone (5.3 of 1,000 screening examinations), DBT-FFDM (6.4 of 1,000 screening examinations), and DBT-s2D mammography (6.1 of 1,000 screening examinations) with no significant difference (FFDM versus DBT-FFDM, p = 0.08; FFDM versus DBT-s2D, p = 0.27).  The percentage of invasive cancers detected was significantly higher with DBT-s2D mammography (76.5 %) than with DBT-FFDM (61.3 %, p = 0.01), and PPVs with DBT-s2D mammography (40.8 %) were significantly higher than those with DBT-FFDM (28.5 %, p < 0.0001).  The authors concluded that screening with DBT-s2D mammography in a large community-based practice improved recall rate and PPVs without loss of cancer detection rate when compared with DBT-FFDM and FFDM alone.

1. this was a single-institution, single-vendor study, and PPV varied with BC prevalence, which differs per geographic location and limits the generalizability of these results,
2. this was a retrospective study, and given the free text dictation system used at the authors’ institution for reporting, complete data regarding availability of prior studies, risk profile, or reason for recall were not available for full analysis,
3. it is possible that the learning-curve effect influenced the results of our study, because the readers had nearly 3 years of experience with DBT, including a trial period, before s2D mammography was fully implemented,
4. since this was a retrospective review, intra-patient comparisons could not be avoided.  Thus, women with more than 1 screening examination during the study time-period might have been more or less likely to have obtained screening with the same or different technologies after a previous negative or positive result, and
5. the last accrual date was June 30, 2016, and thus information on true- or false-negative results was not available for complete assessment. 

Although the results showed that DBT-s2D mammography detected more invasive cancers without a statistically significant loss of in-situ cancer detection compared with DBT-FFDM, further research with adequate follow-up time and sample size is needed to elucidate the reasons behind this result.  Theoretical considerations include the theory that subtle features of invasive cancers may have been more conspicuous on the DBT-s2D images, and this may have contributed to the increased detection of invasive cancers.  It is also possible that in-situ cancers are being under-diagnosed.  Additional research is necessary to clarify the underlying factors.

Guidelines on breast cancer screening from the American College of Obstetricians and Gynecologists (ACOG, 2011) considered, but did not recommend, breast tomosynthesis. The guidelines concluded that "[c]olor Doppler ultrasonography, computer-aided detection, positron emission tomography, scintimammography, and digital breast tomosynthesis have shown promise in selected clinical situations or as adjuncts to mammography for breast cancer diagnosis. However, these technologies are not considered alternatives to routine mammography."

An American College of Obstetrician and Gynecologist technology assessment on "Digital breast tomosynthesis" (ACOG, 2013) noted that "Clinical data suggest that digital mammography with tomosynthesis produces a better image, improved accuracy, and lower recall rates compared with digital mammography alone.  Further study will be necessary to confirm whether digital mammography with tomosynthesis is a cost-effective approach, capable of replacing digital mammography alone as the first-line screening modality of choice for breast cancer screening". The ACOG technology assessment on DBT (2013) stated that this strategy of 2D plus 3D mammography would need to be compared to other potential strategies for comparative cost-effectiveness, such as the use of computer aided detection.  Additional studies will be needed to confirm whether digital mammography with tomosynthesis is a cost-effective approach capable of replacing digital mammography alone as the first-line screening modality of choice for breast cancer screening. Furthermore, the ACOG technology assessment also noted the concern with the high-dose of radiation with breast tomosynthesis.

An assessment by the Standing Committee on Screening of the Screening Section of the Australian^° Department of Health and Ageing (2013) stated that: "Whilst there is some evidence that DBT is at least as sensitive and specific as DM, it is not clear whether the use of DBT in large populations outside of a research setting would deliver the same results or whether the additional cost is justified. The issue of radiation dose also needs to be considered in terms of safety for women being screened".

A BlueCross BlueShield Association TEC Assessment, "Use of digital breast tomosynthesis with mammography for breast cancer screening or diagnosis" (BCNSA, 2014), concluded that "Improved health outcomes following the addition of breast tomosynthesis to screening or diagnostic mammography [has] not been demonstrated in the investigational setting.  Based on the above, the addition of digital breast tomosynthesis to screening or diagnostic mammography does not meet the TEC criteria". The BCBS Medical Advisory Panel reaffirmed their position on digital breast tomosynthesis in 2015.

The USPSTF’s most recent (2016) review concludes that there is insufficient evidence to assess the benefits and harms of DBT as a primary screening method or adjunctive screening method. The USPSTF found insufficient evidence to assess the balance of benefits and harms of DBT as a primary screening method for breast cancer. The Task Force stated that the evidence for DBT is limited. In support, they cited an  assessment of DBT prepared for the USPSTF by the Federal Agency for Healthcare Research and Quality (AHRQ). The AHRQ review found only a single study on the test characteristics of DBT as a primary screening strategy for breast cancer that met the inclusion criteria of the systematic evidence review.

Digital breast tomosynthesis decreases the recall rate by 15 percent (Friedewald, et al., 2014) and increases the number of cancers detected. The extent to which the increased cancers detected are the consequence of over-diagnosis is unknown. The USPSTF found that, "from the limited data available, DBT seems to reduce recall rates (that is, follow-up for additional imaging or testing) and increase cancer detection rates compared with conventional digital mammography alone. However, current study designs cannot determine whether all of the additional cases of cancer detected would have become clinically significant (that is, the degree of overdiagnosis) or whether there is an incremental clinical benefit to detecting these cancers earlier than with conventional digital mammography. In addition, no studies of DBT looked at clinical outcomes, such as breast cancer morbidity or mortality or quality of life."

The USPSTF found that, "as currently practiced in most settings, DBT exposes women to approximately twice the amount of radiation as conventional digital mammography.  In 2013, the U.S. Food and Drug Administration approved a method to generate synthetic reconstruction of 2-dimensional images from 3-dimensional views, which reduces the total radiation dose associated with DBT. Although the extent to which this new software technology has been implemented in mammography screening centers is not precisely known, it is currently thought to be low. In women with abnormal findings, DBT may also increase the rate of breast biopsy compared with conventional digital mammography." 

National Comprehensive Cancer Network (NCCN) guidelines on breast cancer screening (NCCN, 2017) include a recommendation to "consider tomosynthesis." Although the NCCN guidelines were published in July 2016, the accompanying discussion providing the rationale for their decision was published in May 2017. The NCCN guidelines state that digital breast tomosynthesis "appears to improve cancer detection rates and reduce false-positive call-back rates. . . The use of 2D and DBT results in double the radiation exposure compared with mammography alone… The radiation dose can be minimized by newer tomosynthesis techniques that create a synthetic 3D image … " 

The ongoing TOMMY trial: (A comparison of TOMosynthesis with digital MammographY in the UK NHS Breast Screening Programme) will compare the accuracy of digital breast tomosynthesis (DBT) with standard digital full field mammography( FFDM) in the diagnosis of breast cancer. The aim of the trial is to assess whether DBT could improve upon digital mammography as a screening tool, particularly in certain groups of women, e.g. those with a family history of breast cancer, or women with dense breasts. The diagnostic accuracy of DBT, FFDM and DBT+FFDM will be evaluated in an independent retrospective reading study and compared to the final clinical outcome for each case.

Contrast-Enhanced Spectral Mammography (CEM/CESM)

With contrast enhanced spectral mammography (CEM/CESM), an X-ray contrast agent is used to produce contrast enhanced images to purportedly facilitate localization of a lesion. Two sets of images are produced (pre/post contrast) for comparison purposes. 

Allec et al (2011) noted that the accumulation of injected contrast agents allows the image enhancement of lesions through the use of contrast-enhanced mammography.  In this technique, the combination of 2 acquired images is used to create an enhanced image.  There exist several methods to acquire the images to be combined, which include dual energy subtraction using a single detection layer that suffers from motion artifacts due to patient motion between image acquisition.  To mitigate motion artifacts, a detector composed of 2 layers may be used to simultaneously acquire the low and high energy images.  In this work, these researchers evaluated both of these methods using amorphous selenium as the detection material to find the system parameters (tube voltage, filtration, photoconductor thickness and relative intensity ratio) leading to the optimal performance.  They then compared the performance of the 2 detectors under the variation of contrast agent concentration, tumor size and dose.  The detectability was found to be most comparable at the lower end of the evaluated factors.  The single-layer detector not only led to better contrast, due to its greater spectral separation capabilities, but also had lower quantum noise.  The single-layer detector was found to have a greater detectability by a factor of 2.4 for a 2.5 mm radius tumor having a contrast agent concentration of 1.5 mg ml(-1) in a 4.5 cm thick 50 % glandular breast.  The authors stated that the inclusion of motion artifacts in the comparison is part of ongoing research efforts.

Zhao et al (2012) noted that mammography is the primary imaging tool for screening and diagnosis of human breast cancers, but approximately 10 to 20 % of palpable tumors are not detectable on mammograms and only about 40 % of biopsied lesions are malignant.  These researchers reported a high-resolution, low-dose phase contrast X-ray tomographic method for 3D diagnosis of human breast cancers.  By combining phase contrast X-ray imaging with an image reconstruction method known as equally sloped tomography, these investigators imaged a human breast in 3D and identified a malignant cancer with a pixel size of 92 μm and a radiation dose less than that of dual-view mammography.  According to a blind evaluation by 5 independent radiologists, this method can reduce the radiation dose and acquisition time by approximately 74 % relative to conventional phase contrast X-ray tomography, while maintaining high image resolution and image contrast.  The authors concluded that these results demonstrated that high-resolution 3D diagnostic imaging of human breast cancers can, in principle, be performed at clinical compatible doses.

Furthermore, an UpToDate review on "MRI of the breast and emerging technologies" (Slanetz, 2013) states that "Emerging Imaging Technology for Breasr Cancer Detection – Recognition of the limitations of mammography, ultrasound, and breast MRI has led to investigation of other breast imaging techniques including contrast-enhanced dual energy digital mammography, high-field strength MRI, magnetic resonance spectroscopy, diffusion weighted imaging, breast specific gamma imaging, and positron emission mammography".  Contrast enhanced spectral mammography is not mentioned in this review.

Luczyiska et al (2015) compared contrast-enhanced spectral mammography (CESM) and breast MRI with histopathological results and to compare the sensitivity, accuracy, and positive and negative predictive values for both imaging modalities.  After ethics approval, CESM and MRI examinations were performed in 102 patients who had suspicious lesions described in conventional mammography.  All visible lesions were evaluated independently by 2 experienced radiologists using BI-RADS classifications (scale 1 to 5).  Dimensions of lesions measured with each modality were compared to post-operative histopathology results.  There were 102 patients entered into CESM/MRI studies and 118 lesions were identified by the combination of CESM and breast MRI.  Histopathology confirmed that 81 of 118 lesions were malignant and 37 were benign.  Of the 81 malignant lesions, 72 were invasive cancers and 9 were in-situ cancers.  Sensitivity was 100 % with CESM and 93 % with breast MRI.  Accuracy was 79 % with CESM and 73 % with breast MRI.  Receiver-operating-characteristic curve areas based on BI-RADS were 0.83 for CESM and 0.84 for breast MRI.  Lesion size estimates on CESM and breast MRI were similar, both slightly larger than those from histopathology.  The authors concluded that these findings indicated that CESM has the potential to be a valuable diagnostic method that enables accurate detection of malignant breast lesions, has high negative predictive value, and a false-positive rate similar to that of breast MRI.

Hobbs et al (2015) stated that CESM may have similar diagnostic performance to contrast-enhanced MRI (CEMRI) in the diagnosis and staging of breast cancer.  To date, research has focused exclusively on diagnostic performance when comparing these 2 techniques.  Patient experience is also an important factor when comparing and deciding on which of these modalities is preferable.  These researchers compared patient experience of CESM against CEMRI during pre-operative breast cancer staging.  A total of 49 participants who underwent both CESM and CEMRI, as part of a larger trial, completed a Likert questionnaire about their preference for each modality according to the following criteria: comfort of breast compression, comfort of intravenous (IV) contrast injection, anxiety and overall preference.  Participants also reported reasons for preferring one modality to the other.  Quantitative data were analyzed using a Wilcoxon sign-rank test and chi-squared test.  Qualitative data were reported descriptively.  A significantly higher overall preference towards CESM was demonstrated (n = 49, p < 0.001), with faster procedure time, greater comfort and lower noise level cited as the commonest reasons.  Participants also reported significantly lower rates of anxiety during CESM compared with CEMRI (n = 36, p = 0.009).  A significantly higher rate of comfort was reported during CEMRI for measures of breast compression (n = 49, p = 0.001) and the sensation of IV contrast injection (n = 49, p = 0.003).  The authors concluded that these data suggested that patients preferred the experience of CESM to CEMRI, adding support for the role of CESM as a possible alternative to CEMRI for breast cancer staging.

Fredenberg and colleagues (2010) noted that spectral imaging is a method in medical x-ray imaging to extract information about the object constituents by the material-specific energy dependence of x-ray attenuation.  The authors have investigated a photon-counting spectral imaging system with 2 energy bins for contrast-enhanced mammography.  System optimization and the potential benefit compared to conventional non-energy-resolved absorption imaging were studied.  A framework for system characterization was set up that included quantum and anatomical noise and a theoretical model of the system was benchmarked to phantom measurements.  Optimal combination of the energy-resolved images corresponded approximately to minimization of the anatomical noise, which is commonly referred to as energy subtraction. In that case, an ideal-observer detectability index could be improved close to 50% compared to absorption imaging in the phantom study.  Optimization with respect to the signal-to-quantum-noise ratio, commonly referred to as energy weighting, yielded only a minute improvement.  In a simulation of a clinically more realistic case, spectral imaging was predicted to perform approximately 30 % better than absorption imaging for an average glandularity breast with an average level of anatomical noise.  For dense breast tissue and a high level of anatomical noise, however, a rise in detectability by a factor of 6 was predicted.  Another approximately 70 to 90 % improvement was found to be within reach for an optimized system.  The authors concluded that contrast-enhanced spectral mammography is feasible and beneficial with the current system, and there is room for additional improvements.  Inclusion of anatomical noise is essential for optimizing spectral imaging systems.

Schmitzberger et al (2011) demonstrated the feasibility of low-dose photon-counting tomosynthesis in combination with a contrast agent (contrast material-enhanced tomographic mammography) for the differentiation of breast cancer.  All studies were approved by the institutional review board, and all patients provided written informed consent.  A phantom model with wells of iodinated contrast material (3 mg of iodine per milliliter) 1, 2, 5, 10, and 15 mm in diameter was assessed.  A total of 9 patients with malignant lesions and 1 with a high-risk lesion (atypical papilloma) were included (all women; mean age of 60.7 years).  A multi-slit photon-counting tomosynthesis system was utilized (spectral imaging) to produce both low- and high-energy tomographic data (below and above the k edge of iodine, respectively) in a single scan, which allowed for dual-energy visualization of iodine.  Images were obtained prior to contrast material administration and 120 and 480 seconds after contrast material administration.  Four readers independently assessed the images along with conventional mammograms, ultrasonographic images, and magnetic resonance images. Glandular dose was estimated.  Contrast agent was visible in the phantom model with simulated spherical tumor diameters as small as 5 mm.  The average glandular dose was measured as 0.42 mGy per complete spectral imaging tomosynthesis scan of one breast.  Because there were 3 time-points (prior to contrast medium administration and 120 and 480 seconds after contrast medium administration), this resulted in a total dose of 1.26 mGy for the whole procedure in the breast with the abnormality.  Seven of 10 cases were categorized as Breast Imaging Reporting and Data System score of 4 or higher by all four readers when reviewing spectral images in combination with mammograms.  One lesion near the chest wall was not captured on the spectral image because of a positioning problem.  The authors concluded that the use of contrast-enhanced tomographic mammography has been demonstrated successfully in patients with promising diagnostic benefit.  They stated that further studies are necessary to fully assess diagnostic sensitivity and specificity.

1. their diagnostic usefulness, and
2. the relation between parameters assessed using these techniques and well-known diagnostic/prognostic/predictive markers (histological, clinical, and molecular). 

Thus, these researchers studied the relationship between the tumor margin assessed on CESM and

1. tumor borders defined on the basis of macroscopic and microscopic examination,
2. pT,
3. pN, and
4. tumor grade in a group of 82 breast cancer patients. 

Based on CESM, the tumor border was defined as sharp, indistinct or spiculated, whereas in the case of lesions showing weak or medium enhancement on CESM the borders were classified as unspecified.  These investigators found a statistically significant relationship between tumor margin on CESM and

1. macroscopic border (a spiculated margin on CESM was found only in carcinomas with an invasive border on histological examination; p = 0.004),
2. pT (p = 0.016), and
3. pN (nodal involvement was observed most frequently in carcinomas with a spiculated or indistinct margin on CESM; p = 0.045). 

Moreover, in cases with an undefined margin on CESM (cases showing weak or medium enhancement on CESM), both invasive and pushing borders were found on histological examination.  The authors concluded that the findings of this preliminary study suggested that it is possible to assess macroscopic borders of examined lesions on the basis of CESM imaging.  This might be useful in planning the extent of surgical excision.  On the other hand, the assessment of the tumor margin on CESM might not be precise in cases showing weak enhancement.

Kariyappa and colleagues (2016) the role of contrast-enhanced dual-energy spectral mammogram (CEDM) as a problem-solving tool in equivocal cases.  A total of 44 consenting females with equivocal findings on full-field digital mammogram underwent CEDM.  All the images were interpreted by 2 radiologists independently.  Confidence of presence was plotted on a 3-point Likert scale and probability of cancer was assigned on Breast Imaging Reporting and Data System scoring.  Histopathology was taken as the gold standard.  Statistical analyses of all variables were performed.  A total of 44 breast lesions were included in the study, among which 77.3 % lesions were malignant or pre-cancerous and 22.7 % lesions were benign or inconclusive; 20 % of lesions were identified only on CEDM.  True extent of the lesion was made out in 15.9 % of cases, multi-focality was established in 9.1 % of cases and ductal extension was demonstrated in 6.8 % of cases.  Statistical significance for CEDM was p < 0.05; inter-observer kappa value was 0.837.  The authors noted that CEDM has a useful role in identifying occult lesions in dense breasts and in triaging lesions.  In a mammographically visible lesion, CEDM characterizes the lesion, affirmed the finding and better demonstrated response to treatment.  They concluded that CEDM is a useful complementary tool to standard mammogram; it has the potential to be a screening modality with need for further studies and validation.

Phillips and associates (2017) evaluated patient preferences toward screening CESM versus MRI.  As part of a prospective study, high-risk patients had breast MRI and CESM.  Patients completed an anonymous survey to evaluate preferences regarding the 2 modalities; 88 % of participants completed the survey; 79 % preferred CESM over MRI if the exams had equal sensitivity; 89 % would be comfortable receiving contrast as part of an annual screening test.  The authors concluded that high-risk populations may accept CESM as a screening exam and may prefer it over screening MRI if ongoing trials demonstrate screening CESM to be clinically non-inferior MRI.

Li and co-workers (2017) retrospectively compared the diagnostic performance of CESM with that of breast MRI (BMRI) in breast cancer detection using parameters, including sensitivity, PPV, lesion size, morphology, lesion and background enhancement, and examination time.  A total of 48 women (mean age of 56 years ± 10.6 [SD]) with breast lesions detected between October 2012 and March 2014 were included.  Both CESM and BMRI were performed for each patient within 30 days.  The enhancement intensity of lesions and breast background parenchyma was subjectively assessed for both modalities and was quantified for comparison.  Statistical significance was analyzed using paired t-test for mean size of index lesions in all malignant breasts (an index lesion defined as the largest lesion in each breast), and a mean score of enhancement intensity for index lesions and breast background; PPV, sensitivity, and accuracy were calculated for both CESM and BMRI.  The average duration time of CESM and MRI examinations was also compared.  A total of 66 lesions were identified, including 62 malignant and 4 benign lesions.  Both CESM and BMRI demonstrated a sensitivity of 100 % for detection of breast cancer.  There was no statistically significant difference between the mean size of index lesions (p = 0.108).  The enhancement intensity of breast background was significantly lower for CESM than for BMRI (p < 0.01).  The mean score of enhancement intensity of index lesions on CESM was significantly less than that for BMRI (p < 0.01).  The smallest lesion that was detected by both modalities measured 4 mm; CESM had a higher PPV than BMRI (p > 0.05).  The average examination time for CESM was significantly shorter than that of BMRI (p < 0.01).  The authors concluded that CESM had similar sensitivity than BMRI in breast cancer detection, with higher PPV and less background enhancement.  They stated that CESM is associated with significantly shorter examination time thus a more accessible alternative to BMRI, and has the potential to play an important tool in breast cancer detection and staging.

Barra and colleagues (2017) evaluated the feasibility of CESM of the breast for assessing the size of residual tumors after neoadjuvant chemotherapy (NAC).  In BC patients who underwent NAC between 2011 and 2013, these researchers evaluated residual tumor measurements obtained with CESM and FFDM.  They determined the concordance between the methods, as well as their level of agreement with the pathology.  A total of 3 radiologists analyzed 8 CESM and FFDM measurements separately, considering the size of the residual tumor at its largest diameter and correlating it with that determined in the pathological analysis; inter-observer agreement was also evaluated.  The sensitivity, specificity, PPV, and negative predictive value (NPV) were higher for CESM than for FFDM (83.33 %, 100 %, 100 %, and 66 % versus 50 %, 50 %, 50 %, and 25 %, respectively).  The CESM measurements showed a strong, consistent correlation with the pathological findings (correlation coefficient = 0.76 to 0.92; intra-class correlation coefficient = 0.692 to 0.886).  The correlation between the FFDM measurements and the pathological findings was not statistically significant, with questionable consistency (intra-class correlation coefficient = 0.488 to 0.598).  Agreement with the pathological findings was narrower for CESM measurements than for FFDM measurements.  Inter-observer agreement was higher for CESM than for FFDM (0.94 versus 0.88).  The authors concluded that CESM is a feasible means of evaluating residual tumor size after NAC, showing a good correlation and good agreement with pathological findings.  They stated that these findings might be used as reference data for future prospective studies designed to evaluate the impact of CESM on surgical decision-making.

This study had several drawbacks.  The main one being its small sample size (12 lesions in 11 patients).  The small sample size precluded an analysis of diverse cancer types and the collection of data regarding the influence of molecular subtypes.  In addition, because CESM was not performed before or during NAC, these researchers were unable to predict the response or estimate tumor size reduction. Finally, readers were not totally blinded, because the laterality of the tumor was known to them.

In a prospective study, Iotti and associates (2017) compared CESM and contrast-enhanced-MRI (MRI) in the evaluation of tumor response to NAC.  A total of 54 consenting women with BC and indication of NAC were consecutively enrolled between October 2012 and December 2014.  Patients underwent both CESM and MRI before, during and after NAC; MRI was performed first, followed by CESM within 3 days.  Response to therapy was evaluated for each patient, comparing the size of the residual lesion measured on CESM and MRI performed after NAC to the pathological response on surgical specimens (gold standard), independently of and blinded to the results of the other test.  The agreement between measurements was evaluated using Lin's coefficient.  The agreement between measurements using CESM and MRI was tested at each step of the study, before, during and after NAC.  And last of all, the variation in the largest dimension of the tumor on CESM and MRI was assessed according to the parameters set in RECIST 1.1 criteria, focusing on pathological complete response (pCR).  A total of 46 patients (85 %) completed the study; CESM predicted pCR better than MRI (Lin's coefficient 0.81 and 0.59, respectively).  Both methods tend to under-estimate the real extent of residual tumor (mean of 4.1 mm in CESM, 7.5 mm in MRI).  The agreement between measurements using CESM and MRI was 0.96, 0.94 and 0.76 before, during and after NAC respectively.  The distinction between responders and non-responders with CESM and MRI was identical for 45/46 patients.  In the assessment of CR, sensitivity and specificity were 100 % and 84 %, respectively, for CESM, and 87 % and 60 % for MRI.  The authors concluded that CESM and MRI lesion size measurements were highly correlated; CESM appeared at least as reliable as MRI in assessing the response to NAC, and may be an alternative if MRI is contraindicated or its availability is limited.

This study had several drawbacks.  First of all, the small number of patients enrolled who were not randomized, which limits the statistical significance of the results.  Nevertheless, the promising findings encouraged further studies.  Second, MRI was performed using different units, which may have led to heterogeneous findings despite following the same protocol.  Also, the participation of 7 independent radiologists might be considered a limitation because of a supposed non-homogeneous interpretation of images.  Finally, no multi-centric or multi-focal BCs were included in the study.

Deng and co-workers (2018) retrospectively analyzed the quantitative measurement and kinetic enhancement among pathologically proven benign and malignant lesions using CESM.  These investigators examined the differences in enhancement between 44 benign and 108 malignant breast lesions in CESM, quantifying the extent of enhancements and the relative enhancements between early (between 2 to 3 mins after contrast medium injection) and late (3 to 6 mins) phases.  The enhancement was statistically stronger in malignancies compared to benign lesions, with good performance by the receiver operating characteristic curve (0.877, 95 % CI: 0.813 to 0.941).  Using optimal cut-off value at 220.94 according to Youden index, the sensitivity was 75.9 %, specificity 88.6 %, positive likelihood ratio 6.681, negative likelihood ratio 0.272 and accuracy 82.3 %.  The relative enhancement patterns of benign and malignant lesions, showing 29.92 versus 73.08 % in the elevated pattern, 7.14 versus 92.86 % in the steady pattern, 5.71 versus 94.29 % in the depressed pattern, and 80.00 versus 20.00 % in non-enhanced lesions (p < 0.0001), respectively.  The authors concluded that despite variations in the degree of tumor angiogenesis, quantitative analysis of the breast lesions on CESM documented the malignancies had distinctive stronger enhancement and depressed relative enhancement patterns than benign lesions.  To the authors’ knowledge, this was the first study evaluating the feasibility of quantifying lesion enhancement on CESM.  The quantities of enhancement were informative for assessing breast lesions in which the malignancies had stronger enhancement and more relative depressed enhancement than the benign lesions.

Richter and associates (2018) noted that CESM is a novel breast imaging technique providing comparable diagnostic accuracy to breast MRI.  These investigators attempted to show that CESM in patients with MRI contraindications is feasible, accurate, and useful as a problem-solving tool, and to highlight its limitations.  A total of 118 patients with MRI contraindications were examined by CESM.  Histology was obtained in 94 lesions and used as gold standard for diagnostic accuracy calculations.  Imaging data were reviewed retrospectively for feasibility, accuracy, and technical problems.  The diagnostic yield of CESM as a problem-solving tool and for therapy response evaluation was reviewed separately.  CESM was more accurate than mammography (MG) for lesion categorization (r = 0.731, p < 0.0001 versus r = 0.279, p = 0.006) and for lesion size estimation (r = 0.738 versus r = 0.689, p < 0.0001); NPV of CESM was significantly higher than of MG (85.71 % versus 30.77 %, p < 0.0001).  When used for problem-solving, CESM changed patient management in 2/8 (25 %) cases.  Super-position artifacts and timing problems affected diagnostic utility in 3/118 (2.5 %) patients.  The authors concluded that CESM was a feasible and accurate alternative for patients with MRI contraindications, however, it is necessary to be aware of the method's technical limitations.

Xing and colleagues (2019) evaluated the diagnostic value between CESM and breast MRI in breast disease.  A total of 235 patients who were suspected of having breast abnormalities by clinical examination or mammography underwent CESM and MRI examination.  Using histopathologic results as the criterion standard, the diagnostic performance of CESM and MRI was examined.  The AUC-ROC were applied to analyze diagnostic efficiency.  The Pearson correlation coefficients between CESM versus pathology and MRI versus pathology were calculated.  A total of 263 breast lesions were found in 235 patients, in which 177 were malignant and 86 were benign.  By evaluating the diagnostic value, sensitivity, PPV, NPV, and false-negative rate from CESM examination were comparable to those from MRI (91.5 %, 94.7 %, 83.7 %, and 8.5 % versus 91.5 %, 90.5 %, 82.1 %, and 8.5 %).  More importantly, the accuracy and the specificity were higher for CESM than those for MRI (81 % and 89.5 % versus 80.2 % and 71.7 %), whereas the false-positive rate was lower (10.5 % versus 19.8 %).  The AUC-ROC of CESM and MRI were 0.950 and 0.939, displaying the equivalent diagnostic efficiency (p = 0.48).  For the agreement between measurements, mean tumor sizes were 3.1 cm for CESM and 3.4 cm for MRI compared with 3.2 cm on histopathologic results.  The Pearson correlation coefficient of CESM versus histopathology (r = 0.774, p = 0.000) was consistent with MRI versus histopathology (r = 0.771, p = 0.000).  The authors concluded that these findings showed better accuracy, specificity, and false-positive rate of CESM in breast cancer detection than MRI.  These researchers stated that CESM displayed a good correlation with histopathology in assessing the lesion size of breast cancer, which is consistent with MRI.

Perry and colleagues (2019) discussed the essential steps involved in performing, interpreting, managing, and reporting findings on contrast-enhanced mammography (CEM) for successful implementation into clinical practice.  The authors concluded  that CEM is a promising tool for breast cancer detection that uses contrast agent to identify areas of increased blood flow due to angiogenesis.  In CEM, the use of a dual-energy technique results in a low-energy image providing information comparable to that of a conventional digital mammogram, with the added benefit of recombined images highlighting areas of angiogenesis.  Understanding imaging acquisition and the spectrum of negative, benign, and malignant findings for CEM is critical for interpretation, management, and reporting.  No formal CEM lexicon exists, but reporting should address both morphologic findings on the low-energy images and the enhancement findings on the recombined images.

National Comprehensive Cancer Network’s clinical practice guideline on "Breast Cancer Screening and Diagnosis" (Version 1.2022) states that "Contrast-enhanced mammography is also an option for higher risk breast cancer screening … The use of contrast-enhanced breast MRI during pregnancy is generally considered to be contraindicated because gadolinium in all forms crosses the placenta and enters the fetal circulation …".

The National Institute for Health and Care Excellence’ guideline on "Contrast-enhanced spectral mammography for breast cancer"(NICE, 2022) discussed 7 diagnostic accuracy studies, and 3 cohort studies, including a total of 1,809 women with known or suspected breast cancer.  Indirect comparisons with histopathology showed that the performance of CESM was comparable with other imaging techniques in women with known or suspected breast cancer.  Moreover, this NICE guideline noted that key uncertainties around the evidence or technology are that diagnostic accuracy studies compared imaging with histopathology, and not directly with other imaging techniques.  There is also a lack of prospective, comparative studies reporting on longer-term patient outcomes.

Potsch et al (2022) stated that CEM is a more accessible alternative to contrast-enhanced MRI (CE-MRI) in breast imaging; however, a summary comparison of published studies is lacking.  These researchers compared the performance of CEM and CE-MRI regarding sensitivity, specificity, and NPV in the detection of breast cancer, involving all publicly available studies in the English language.  Two investigators extracted characteristics of studies examining the comparative diagnostic performance of CEM and CE-MRI in detecting breast cancer.  Studies published until April 2021 were eligible.  Sensitivity, specificity, NPV, and positive and negative likelihood ratios (LR+ and LR-) were calculated using bivariate random effects models.  A Fagan nomogram was used to identify the maximum pre-test probability at which post-test probabilities of a negative CEM or CE-MRI examination were in line with the 2 % malignancy rate benchmark for down-grading a Breast Imaging Reporting and Data System (BI-RADS) category 4 to a BI-RADS category 3 result.  I2 statistics, Deeks funnel plot asymmetry test for publication bias, and meta-regression were employed.  A total of 7 studies entailing 1,137 lesions (654 malignant, 483 benign) with an average cancer prevalence of 65.3 % (range of 47.3 % to 82.2 %) were included.  No publication bias was found (p = 0.57).  While the LH+ was equal at a value of 3.1 for CE-MRI and 3.6 for CEM, the LR- of CE-MRI (0.04) was lower than that with CEM (0.12).  CE-MRI had higher sensitivity for breast cancer than CEM (97 % [95 % CI: 86 % to 99 %] versus 91 % [95 % CI: 77 % to 97 %], respectively; p < 0.001) but lower specificity (69 % [95 % CI: 46 % to 85 %] versus 74 % [95 % CI: 52 % to 89 %]; p = 0.09).  A Fagan nomogram showed that the maximum pre-test probability at which both tests could rule out breast cancer was 33 % for CE-MRI and 14 % for CEM.  In addition, iodine concentration was positively associated with CEM sensitivity and negatively associated with its specificity (p = 0.04 and p < 0.001, respectively).  The authors concluded that contrast-enhanced MRI had superior sensitivity and LH- with higher pre-test probabilities to rule out malignancy compared with contrast-enhanced mammography.

In a systematic review and meta-analysis, Neeter et al (2023) examined the diagnostic accuracy of contrast-enhanced mammography (CEM) compared to standard contrast-enhanced breast MRI.  Like breast MRI, CEM enables tumor visualization by contrast accumulation.  CEM appeared to be a viable substitute for breast MRI.  This systematic search examined the diagnostic accuracy of these techniques in women with suspicious breast lesions on previous imaging or physical examination, who have undergone both breast MRI and CEM.  CEM had to be performed on a commercially available system.  The MRI sequence parameters had to be described sufficiently to ensure that standard breast MRI sequence protocols were used.  Pooled values of sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR), were estimated using bi-variate mixed-effects logistic regression modeling.  Hierarchical summary ROC curves for CEM and breast MRI were also constructed.  A total of 6 studies (607 patients with 775 lesions) met the pre-defined inclusion criteria.  Pooled sensitivity was 96 % for CEM and 97 % for breast MRI.  Pooled specificity was 77 % for both modalities.  DOR was 79.5 for CEM and 122.9 for breast MRI.  Between-study heterogeneity expressed as the I2-index was substantial with values over 80 %.  The authors concluded that pooled sensitivity was high for both CEM and breast MRI, with moderate specificity.  The pooled DOR estimates, however, indicated higher overall diagnostic performance of breast MRI compared to CEM.  Nonetheless, current scientific evidence is too limited to prematurely discard CEM as an alternative for breast MRI.  These researchers stated that future studies that directly compare CEM and breast MRI for various indications are much needed.

In a systematic review and meta-analysis, Gelardi et al (2023) examined the available evidence regarding the role of CE-MRI and CEM in the early detection, diagnosis, and pre-operative assessment of breast cancer.  The search was carried out on PubMed, Google Scholar, and Web of Science on July 28, 2021 using the following terms "breast cancer", "preoperative staging", "contrast-enhanced mammography", "contrast-enhanced spectral mammography", "contrast enhanced digital mammography", "contrast-enhanced breast magnetic resonance imaging" "CEM", "CESM", "CEDM", and "CE-MRI".  These investigators selected only those studies comparing the effectiveness of CEM and CE-MRI.  The study quality was assessed using the QUADAS-2 criteria.  The pooled sensitivities and specificity of CEM and CE-MRI were calculated using a random-effects model directly from the STATA "metaprop" command.  The between-study statistical heterogeneity was tested (I2-statistics).  A total of 19 studies were selected for this systematic review; and 15 studies (1,315 patients) were included in the meta-analysis.  Both CEM and CE-MRI detected breast lesions with a high sensitivity, without a significant difference in performance (97 % and 96 %, respectively).  The authors concluded that these findings confirmed the potential of CEM as a supplemental screening imaging modality, even for intermediate-risk women, including women with dense breasts and a history of breast cancer.

The authors stated that this study has several drawbacks.  First, the small number of the eligible studies, especially in the sub-analyses in different clinical settings, might have affected results.  Second, the study designs and sample sizes varied widely between the different included studies, as evidenced by the relatively high heterogeneity.  These researchers harmonized the studies by conducting different sub-analyses; however, this strategy furtherly reduced the sample size.  Third, most studies were carried out by examining patients with already known primary lesions or suspected lesions at screening, which might have impacted the sensitivity and specificity.

An UpToDate review on "Clinical features, diagnosis, and staging of newly diagnosed breast cancer" (Joe, 2023) states that "Contrast enhanced mammography is a new technology that approaches accuracy of MRI for pre-operative staging and may be considered in patients unable to obtain breast MRI".

An UpToDate review on "Breast density and screening for breast cancer" (Freer and Slanetz, 2023) states that "Emerging technologies -- Multiple other supplemental screening methods are being investigated, including the use of abbreviated MRI, molecular breast imaging, and contrast-enhanced mammography.  Although some centers are now routinely performing these examinations, they have yet to be broadly adopted into routine clinical practice as supplemental screening tools in women with dense breast tissue … Contrast-enhanced mammography is a dual-energy technique that allows for acquisition of a low-energy image comparable to a mammogram and a subtracted image displaying contrast uptake following injection of an intravenous iodinated contrast agent.  In the diagnostic setting, the sensitivity has been shown to be equivalent to MRI among women at increased risk of breast cancer, although it has not been compared with MRI in women with dense breasts.  However, it may be superior to mammography alone.  As an example, in a 2017 study from Japan, this imaging technique detected substantially more cancers in women with dense breasts than mammography alone; this technique may have a role in supplemental screening".

Furthermore, an UpToDate review on "MRI of the breast and emerging technologies" (Slanetz, 2023) lists contrast-enhanced dual-energy digital mammography as an emerging imaging technology for breast cancer detection.  It notes that "Bilateral dual-energy (DE) contrast-agent-enhanced (CE) digital mammography is a technique for detecting breast cancer that consists of high-energy and low-energy digital mammography following an intravenous injection of an iodinated contrast material.  This technology may be a useful adjunct for patients who cannot undergo MRI evaluation (e.g., claustrophobic, pacemaker), as studies have shown comparable sensitivity".

Digital Breast Tomosynthesis Plus Digital Mammography for Use in Breast Cancer Screening

The Norwegian Institute of Public Health’s technology assessment on "Digital breast tomosynthesis with Hologic 3D mammography Selenia Dimensions System for use in breast cancer screening" (Movik et al, 2017) evaluated the safety, efficacy, and cost-effectiveness of DBT in BC screening in Norway.  There are several manufacturers of DBT systems, but only Hologic Inc., has to-date (June 2017) submitted a documentation pack.  These investigators performed a single technology assessment (TA) of the use of Hologic Selenia Dimensions digital mammography system for BC screening, based on the submission from Hologic Inc.  These researchers did not cover the use of the system in the diagnosis of BC in clinical practice by this TA.  The documentation submitted by the company consisted of 12 studies identified by a systematic literature search; 4 publications met inclusion criteria and were included for assessment in this TA.  They assessed the present documentation using a pre-defined PICOS (Population, Intervention, Comparator, Outcomes and Study design), risk of bias assessment of data provided by the submission file, data extraction, and graded the certainty of the evidence for the estimates using the GRADE (Grades of Recommendation, Assessment, Development and Evaluation) assessment.  The authors also reviewed the cost-effectiveness analysis and budget impact analysis described in the submission.  The submitter provided no documentation assessing the risk associated with the radiation dose with DBT.  Thus, these researchers conducted a separate assessment of the potential risks associated with radiation exposure with DBT.  Hologic submitted a health economic analysis based on an American discrete event analysis model, from which they had drawn results in terms of quality-adjusted life years gained.  Hologic compared the effects of DBT+DM (synthetic 2D) for a hypothetical cohort of women that was followed through 10 rounds of screening over a 20-year time horizon.  The model was based on data (sensitivity and specificity) from an interim analysis of the Oslo Tomosynthesis Screening Trial.  Hologic did not have access to the model, and carried out the costing calculations separately.  The main cost components were screening costs and BC treatment costs broken down by disease stage.  Costs were applied to the model results and varied in a number of 1-way sensitivity analyses.

The main findings were as follows:  The authors were uncertain whether Hologic DBT in combination with DM or synthesized DM decreased or increased recall rates compared to DM alone (very low confidence due to conflicting evidence from observational studies).  The intervention may increase the rate of screening-detected cancer (cancer detection rate (CDR) according to all studies (very low confidence due to sparse evidence from 1 observational study).  These investigators were uncertain whether Hologic DBT in combination with DM or synthesized DM made any difference with regard to the detection of interval cancer compared to DM alone (very low confidence in the evidence due to sparse evidence from 1 observational study).  The authors were uncertain whether Hologic DBT in combination with DM or synthesized DM decreased or increased false positive rates compared to DM alone (very low confidence due to conflicting evidence from observational studies).  The intervention may provide similar sensitivity rates, but may increase specificity rates (low confidence due to evidence from observational studies).  The authors were uncertain whether Hologic DBT in combination with DM or synthesized DM decreased or increased false negative rates compared to DM alone (very low confidence due to sparse evidence from 1 observational study).  Information on death and quality of life (QOL) was not reported.  Uncertainty regarding the effect estimates means that new research may alter the results and the conclusion.  When compared to the current practice with DM, introducing the Hologic Selenia Dimensions DBT-system into the Norwegian Breast Cancer Screening Program (NBCSP) will result in an increased radiation dose followed by an increased risk of radiation-induced cancer for all the evaluated interventions defined by the PICO.  Summary of findings based on doses reported in the OTST and STORM-2 trial:DBT only: The dose and risk will increase by 23 % to 38 %, resulting in a total absorbed dose to granular tissue (AGD) of 3.7 to 3.9 mGy and an estimated incidence of radiation-induced BC of 15 to 16 per 100,000 women and mortality of 1.2 per 100,000 women.  DBT + DM: The dose and risk will increase by a factor of between 2.23 and 2.37, resulting in a total AGD of 6.4 to 7.0 mGy and an estimated incidence of radiation-induced BC of 26 to 29 per 100,000 women and mortality of 2.1 to 2.3 per 100,000 women.  DBT + S2D: The dose and risk will be increased by 23 % to 38 %, but reduced by 42 % t0 45 % compared to DBT + DM, resulting in the same dose and risk as DBT alone.  The estimated values for incidence of radiation-induced BC and mortality must be interpreted with caution as there was a high level of uncertainty associated with them.  However, the ratio between doses and risks for the different interventions provided valid input to the total risk-benefit evaluation to be done for the screening program.  The base case results of the submitted economic analysis of DBT+DM (S2D) versus DM alone were 0.007 quality adjusted life years (QALY) gained per woman screened.  The incremental cost per QALY gained was approximately NOK 144,000.  This result was estimated for a population of women with dense breasts.  Hologic based the budget impact analysis on 3 components: relative costs of equipment procurement, screening costs, and BC treatment costs.  The base case estimate was a net increase in expenditure of 77.5 million NOK in year 5 after implementation.  Hologic also included sensitivity analysis in the budget impact analysis to determine the effect of varying the price, which has yet to be determined, of the DBT equipment, and to examine how changes in important assumptions would influence the results of the budget impact analysis.  The net increase in expenditure reported varied significantly in the sensitivity analyses. 

Compared to DM alone, the use of Hologic DBT in combination with standard DM or synthesized DM may increase the rate of screening-detected cancer (cancer detection rate or CDR) according to all studies.  The studies provided evidence on the first screening round using DM+DBT, which could partly account for the substantial increased cancer detection rate, compared with standard screening with DM alone.  Estimates of cancer detection rates for repeated DBT screening of the same populations are needed to quantify the effect of adjunct DBT on both cancer detection and false positive recalls at repeated screening rounds.  RCTs assessing the impact of adjunct DBT on interval cancer rates as a surrogate for screening benefit would provide critical evidence to underpin future population screening policy and practice.  RCTs should be designed to simultaneously address additional evidence gaps such as DBT’s incremental cost-effectiveness, and detection measures at repeat screening with adjunct DBT.  Using both DBT and standard DM (dual acquisition) caused an increase in the radiation dose.  DBT-systems with the possibility to generate synthetic 2D images was highly favorable compared to DBT in combination with full field DM, due to its reduction in dose and associated risk.  Information on radiation doses should be included in future clinical trials.  The results from the submitter’s health economic analysis indicated that adjunct DBT compared to current screening practice could lead to earlier detection of BC and a lower recall rate, though potential cost reductions resulting from the latter were not actually modeled.  The results suggested therefore that adjunct DBT could be cost-effective if adopted by the Norwegian Breast Cancer Screening Program.  However, there were a number of issues that contribute to uncertainty regarding the results.  First, the uncertainty described above with regard to the clinical effectiveness, particularly with regard to sensitivity, over repeated screening visits and across different populations (e.g., with respect to breast density).  Second, the authors did not know to what extent the potential increase in BC detection may lead to increased over-diagnosis and unnecessary treatment.  Third, since a coherent, adapted health economic model could not be supplied, it was difficult to ascertain the impact of various assumptions in the analysis and evaluate the total uncertainty regarding the health economic results. 

The authors concluded that there is too little evidence to conclude regarding the effects of the use of Hologic DBT in combination with DM or synthesized DM compared to DM alone for the outcomes assessed in their report (recall rates, CDR, interval cancer rate, false positive and false negative rate, sensitivity, specificity, mortality and QOL).  Moreover, they stated that preparation of a full health technology assessment should be considered when sufficient evidence is available.

In a retrospective, cross-sectional study, Mohindra and associates (2018) examined if addition of DBT to DM would aid in better characterization of mammographic abnormalities in BC patients in general and in different breast compositions.  Mammographic findings in 164 patients with 170 pathologically proven lesions were evaluated by using first DM alone and thereafter with addition of DBT to DM.  The perceived utility of adjunct DBT was scored using a rating of 0 to 2. A score of 0 indicating that DM plus DBT was comparable to DM alone, 1 indicating that DM plus DBT was slightly better, and 2 indicating that DM plus DBT was definitely better.  Statistical analysis was performed using the McNemar Chi-squares test, and Fisher's exact test.  On DM, 149 lesions were characterized mass with or without calcifications, 18 asymmetries with or without calcifications, 2 as architectural distortion, and 1 as micro-calcification alone.  Adjunct DBT helped in better morphological characterization of 17 lesions, with revelation of underlying masses in 16 asymmetries and 1 architectural distortion.  Adjunct DBT was perceived to be slightly better than DM alone in 44.7 % lesions, and definitely better in 22.9 % lesions.  Lesions showing score 1 or 2 improvement were significantly higher in heterogeneously and extremely dense breasts (p < 0.001).  The authors concluded that adjunctive DBT improved morphological characterization of lesions in patients with BC.  It highlighted more suspicious features of lesions that indicated the presence of cancer, particularly in dense breasts.

The main drawback of this study was that sensitivity or specificity of adjunctive DBT could not be evaluated since only pathologically proven malignancies were included.  These researchers evaluated the role of adjunctive DBT in the diagnostic environment; however, the "diagnostic environment"  of this study was not comparable to developed countries where most diagnostic mammograms were performed as recalls from abnormal screening.

Low-Dose CT Combined Mammography in the Diagnosis of Overflow Breast Disease

Tian and colleagues (2021) stated that overflow breast disease (OBD), also known as breast nipple discharge, refers fluid or liquid that comes out of nipple.  Many patients with breast cancer experience such condition; however, it is not easy to detect it at early stage, especially for pathological OBD.  Previous study found low-dose CT combined mammography (LDCTMG) could aid in the diagnosis of OBD.  However, there is no systematic review examining this issue.  This study will examine the accuracy of LDCTMG in the diagnosis of OBD.  This study protocol will search literature sources in electronic databases and other sources.  The electronic databases will be retrieved in the Cochrane Library, the Cochrane Register of Diagnostic Test Accuracy Studies, PubMed, Embase, Web of Science, CINAHL, CNKI, and WangFang from inception to the present.  These investigators will also search other sources.  All literature sources will be sought without restrictions to the language and publication status.  Two researchers will independently perform study selection, data extraction, and study quality assessment.  Statistical analysis will be conducted using RevMan 5.3.  This study will exert a high-quality synthesis of eligible studies on the analysis of LDCTMG in the diagnosis of OBD.  The authors concluded that the findings of this study may provide evidence to aid in deciding if LDCTMG is accurate in the diagnosis of OBD.

Mammography Screening for Men with Gynecomastia

National Comprehensive Cancer Network’s clinical practice guideline on "Genetic/familial high-risk assessment: Breast, ovarian, and pancreatic" (Version 2.2022) states that "Consider annual mammogram screening in men with gynecomastia starting at the age of 50 or 10 years before the earliest known male breast cancer in the family (whichever comes first)".

Mammography Screening for Transfeminine Persons

Parikh et al (2020) stated that "Current screening recommendations in the transgender female population vary according to the age of the individual and other risk factors including the duration of hormone therapy.  Generally, advocates recommend annual or biennial screening starting at age 50 years

Sterling and Garcia (2020) stated that over the last 50 years cancer mortality has decreased, the biggest contributor to this decrease has been the widespread adoption of cancer screening protocols.  These guidelines are based on large population studies, which often do not capture the non-gender conforming portion of the population.  These investigators examined current guidelines and practice patterns of cancer screening in transgender patients, and, where evidence-based data is lacking, to draw from cis-gender screening guidelines to suggest best-practice screening approaches for transgender patients.  They carried out a systematic search of PubMed, Google Scholar and Medline, using all iterations of the follow search terms: transgender, gender non-conforming, gender non-binary, cancer screening, breast cancer, ovarian cancer, uterine cancer, cervical cancer, prostate cancer, colorectal cancer, anal cancer, and all acceptable abbreviations.  Given the limited amount of existing literature, inclusion was broad.  After eliminating duplicates and abstract, all queries yielded 85 unique publications.  There are currently very few transgender specific cancer screening recommendations.  All the guidelines discussed in this manuscript were designed for cis-gender patients and applied to the transgender community based on small case series.  Currently, there is insufficient to evidence to determine the long-term effects of gender-affirming hormone therapy on an individual’s cancer risk.  Established guidelines for cisgender individuals and can reasonably be followed for transgender patients based on what organs remain in-situ.  In the future comprehensive cancer screening and prevention initiatives centered on relevant anatomy and high-risk behaviors specific for transgender men and women are needed.

The American College of Radiology (ACR)’s Appropriateness Criteria for "Transgender Breast Cancer Screening" (2021) notes that mammography screening may be appropriate for transfeminine (male-to-female) patient, 40 years of age or older with past or current hormone use equal to or greater than 5 year.  

Artificial Intelligence (Deep Learning and Machine Learning)-Based Mammography for Diagnosis or Screening of Breast Cancer

Liu et al (2023) stated that breast cancer was the 4th leading cause of cancer-related death globally, and early mammography screening could decrease the breast cancer mortality.  Artificial intelligence (AI)-assisted system based on machine learning (ML) methods could aid in improving the screening accuracy and effectiveness.  In a systematic review and meta-analysis, these investigators examined the diagnostic accuracy of mammography diagnosis of breast cancer via various ML methods.  They searched the Springer Link, Science Direct (Elsevier), IEEE Xplore, PubMed and Web of Science for relevant studies published from January 2000 to September 2021.  A Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) was used to evaluate the included studies, and reporting was assessed using the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines.  The pooled summary estimates for sensitivity, specificity, the area under the receiver operating characteristic curve (AUC) for 3 ML methods (convolutional neural network [CNN], artificial neural network [ANN], support vector machine [SVM]) were calculated.  A total of 32 studies with 23,804 images were included in the meta-analysis. The overall pooled estimate for sensitivity, specificity and AUC was 0.914 [95% CI 0.868-0.945], 0.916 [95% CI 0.873-0.945] and 0.945 for mammography diagnosis of breast cancer through 3 ML methods.  The pooled sensitivity, specificity and AUC of CNN were 0.961 [95 % CI: 0.886 to 0.988], 0.950 [95 % CI: 0.924 to 0.967] and 0.974.  The pooled sensitivity, specificity and AUC of ANN were 0.837 [95 % CI: 0.772 to 0.886], 0.894 [95 % CI: 0.764 to 0.957] and 0.881.  The pooled sensitivity, specificity and AUC of SVM were 0.889 [95 % CI: 0.807 to 0.939], 0.843 [95 % CI: 0.724 to 0.916] and 0.913.  Machine learning methods (especially CNN) showed excellent performance in mammography diagnosis of breast cancer screening based on retrospective studies.  Moreover, the authors concluded that additional rigorous, prospective studies are needed to examine the longitudinal performance of AI.

In a retrospective, single-center study, Chen et al (2023) proposed a deep learning (DL) model based on C2FTrans for diagnosing breast mass using mammographic density.  This trial included patients who underwent mammographic and pathological examination.  Two physicians manually depicted the lesion edges and used a computer to automatically extend and segment the peripheral areas of the lesion (0, 1, 3, and 5 mm, including the lesion).  They then obtained the mammary glands' density and the different regions of interest (ROI).  A diagnostic model for breast mass lesions based on C2FTrans was constructed based on a 7: 3 ratio between the training and testing sets.  Finally, receiver operating characteristic (ROC) curves were plotted.  Model performance was examined using the AUC with 95 % CI, sensitivity, and specificity.  A total of 401 lesions (158 benign and 243 malignant) were included in this study.  The probability of breast cancer in women was positively correlated with age and mass density and negatively correlated with breast gland classification. The largest correlation was observed for age (r = 0.47).  Among all models, the single mass ROI model had the highest specificity (91.8 %) with an AUC = 0.823, and the peri-focal 5 mm ROI model had the highest sensitivity (86.9 %) with an AUC = 0.855.  Furthermore, by combining the cephalon-caudal and medio-lateral oblique views of the peri-focal 5 mm ROI model, these researchers obtained the highest AUC (AUC = 0.877; p < 0.001).  The authors concluded that DL models may improve the accuracy of breast disease diagnosis in future practice, reduce the mis-diagnosis of benign masses to some extent, and become an important auxiliary diagnostic tool for radiologists.  Moreover, these researchers noted that this trial was only focused on the diagnosis of breast lesion classification based on masses and peri-masses and did not examine the correlation between the area around malignant masses and the invasive extent of cancer components, predicted breast cancer prognosis and lymph node metastasis.  They stated that AI and DL have not been used to their full potential for breast cancer diagnosis, staging, and prognosis; therefore, further investigations and development are still needed.

The authors stated that this study had several drawbacks.  First, this trial was a single-center study with samples from the same hospital, which may have resulted in selection bias.  For future studies, it would be desirable to reduce study selection bias via the collaboration of multiple research centers.  Second, the sample size of this retrospective study was relatively small.  Although the performance of the constructed model was stable and the obtained results were promising, a larger, prospective trial is needed to validate the predictive efficiency of the model.  Third, these investigators acknowledged the high proportion of malignant breast lesions (60.6 %), implying that there might have been a potential patient selection bias; thus, balanced datasets were also important for developing DL classification models.  Fourth, due to the characteristics and inherent limitations of CNN and its algorithms, these researchers extracted 2D breast density and other image features on full field digital mammography (FFDM) images.  Compared with 3D density and features, it might lose some lesion information.  However, the results showed that the 2D-based features also displayed good performance in the classification of breast mass lesions.  In the future, the model could be tried to be applied to DBT to obtain the bulk density and to detect more realistic lump density.

In a retrospective study, Liao et al (2023) examined the effectiveness of a DL system based on the DenseNet CNN in diagnosing benign and malignant asymmetric lesions in mammography.  Clinical and image data from 460 women aged 23 to 82 years (47.57 ± 8.73 years) with asymmetric lesions who underwent mammography at Shenzhen People's Hospital, Shenzhen Luohu District People's Hospital, and Shenzhen Hospital of Peking University from December 2019 to December 2020 were analyzed.  Two senior radiologists, 2 junior radiologists, and the DL system read the mammographic images of 460 patients, respectively, and finally recorded the BI-RADS classification of asymmetric lesions.  They then used the AUC of the ROC to assess the diagnostic effectiveness and the difference between AUCs by the Delong method.  Specificity (0.909 versus 0.835, 0.790, χ2 = 8.21 and 17.22, p＜0.05) and precision (0.872 versus 0.763, 0.726, χ2 = 9.23 and 5.22, p＜0.05) of the DL system in the diagnosis of benign and malignant asymmetric lesions were higher than those of junior radiologist A and B, and there was a statistically significant difference between AUCs (0.778 versus 0.579, 0.564, Z = 4.033 and 4.460, p＜0.05)  In addition, the AUC (0.778 versus 0.904, 0.862, Z = 3.191, and 2.167, p＜0.05) of benign and malignant asymmetric lesions diagnosed by the DL system was lower than that of senior radiologist A and senior radiologist B.  The authors concluded that the DL system based on the DenseNet convolution neural network has high diagnostic efficiency, which can help junior radiologists evaluate benign and malignant asymmetric lesions more accurately. It can also improve diagnostic accuracy and reduce missed diagnoses caused by inexperienced junior radiologists.

The authors stated that this study had several drawbacks.  First, this trial was retrospective, and the effectiveness of asymmetric benign and malignant diagnosis requires validation via prospective studies.  Second, the number of asymmetric cases was small, so it is necessary to continue to increase the sample size to improve the model’s accuracy, effectiveness, and stability.  Third, there should be more data on the influence of radiologists on the diagnostic efficiency of asymmetric benign and malignant masses assisted by the DL system.  The follow-up research should further count on the diagnostic results of radiologists combined with the DL system to clarify the diagnostic value of the DL system assisted by radiologists.

Li et al (2023) noted that the detection of masses on mammography represents one of the earliest signs of a malignant breast cancer; however, masses may be hard to detect due to dense breast tissue, resulting in false negative results.  In a retrospective, single-center study, these researchers examined the clinical application of the CNN-based DL system constructed in their previous work as an objective and accurate tool for breast cancer screening and diagnosis in Asian women.  This analysis included 324 patients with masses detected on mammography at Shenzhen People's Hospital between April and December 2019. Detection: images were independently analyzed by 2 junior radiologists who were blinded to relative results.  Then, a senior radiologist analyzed the images after reviewing all the relevant information as the reference.  Classification: masses were classified by the same 2 junior radiologists and in consensus by 2 other seniors.  Images were also input into the DL system.  The sensitivity of detection by junior radiologists and the DL system, effects of different factors [breast density; patient age; morphology, margin, size, breast imaging reporting and data system (BI-RADS) category of the mass] on detection, the accuracy, sensitivity, and specificity of classification, and the AUC-ROC curve were evaluated.  A total of 618 masses were detected.  The detection sensitivity of the 2 junior radiologists [78.0 % (482/618) and 84.0 % (519/618), respectively] was lower than that of the DL system [86.2 % (533/618)].  Breast density significantly affected the detection by 2 junior radiologists (both p = 0.030), but not by the DL system (p = 0.385).  The AUC for classifying masses as negative (BI-RADS 1, 2, 3) or positive (BI-RADS 4A, 4B, 4C, 5) for the DL system was significantly higher compared to those of the 2 junior radiologists, but not significantly different compared to seniors [DL system, 0.697; junior, 0.612 and 0.620 (p = 0.021, 0.019); senior in consensus, 0.748 (p = 0.071)].  The authors concluded that the CNN-based DL system could assist junior radiologists in improving mass detection and is not affected by breast density.  This DL system may have clinical utility in women with dense breasts, including reducing the impact caused by inexperienced radiologists and the potential for missed diagnoses.

The authors stated that this study had 2 main drawbacks.  First, this was a single-center, retrospective study with a small sample size; thus, findings may not be generalizable to clinical practice.  Second, the diagnostic performance of the radiologists combined with the DL system was not examined.

Michel et al (2023) stated that DL techniques, including CNN, have the potential to improve breast cancer risk prediction compared to traditional risk models.  In a retrospective, cohort study, these investigators examined if combining a CNN-based mammographic evaluation with clinical factors in the Breast Cancer Surveillance Consortium (BCSC) model would improve risk prediction.  This trial included a total of 23,467 women, aged 35 to 74 years, undergoing screening mammography (2014 to 2018).  These researchers extracted electronic health record (EHR) data on risk factors.  They identified 121 women who subsequently developed invasive breast cancer at least 1 year after the baseline mammogram.  Mammograms were analyzed with a pixel-wise mammographic evaluation using CNN architecture.  These investigators used logistic regression models with breast cancer incidence as the outcome and predictors including clinical factors only (BCSC model) or combined with CNN risk score (hybrid model).  They compared model prediction performance via AUCs.  Mean age was 55.9 years (SD, 9.5) with 9.3 % non-Hispanic Black, and 36 % Hispanic.  The hybrid model did not significantly improve risk prediction compared to the BCSC model (AUC of 0.654 versus 0.624, respectively, p = 0.063).  In subgroup analyses, the hybrid model out-performed the BCSC model among non-Hispanic Blacks (AUC 0.845 versus 0.589; p = 0.026) and Hispanics (AUC 0.650 versus 0.595; p = 0.049).  The authors concluded that they developed an efficient breast cancer risk assessment method using CNN risk score and clinical factors from the EHR.  Moreover, these researchers stated that with future validation in a larger cohort, their CNN model combined with clinical factors may aid in predicting breast cancer risk in a cohort of racially/ethnically diverse women undergoing screening.

Gurmessa and Jimma (2024) noted that breast cancer is the most common disease in women.  Recently, explainable AI (XAI) approaches have been dedicated to examining breast cancer.  In a systematic review, these investigators examined XAI for breast cancer diagnosis from mammography and ultrasound (US) images.  They examined how XAI methods for breast cancer diagnosis have been evaluated, the existing ethical challenges, research gaps, the XAI used, and the relation between the accuracy and explainability of algorithms.  In this study, the PRISMA check-list and diagram were used.  Peer-reviewed studies and conference proceedings from PubMed, IEEE Explore, ScienceDirect, Scopus and Google Scholar databases were searched.  There was no stated date limit to filter the studies.  The studies were searched on September 19, 2023, using various combinations of the search terms “breast cancer”, “explainable”, “interpretable”, “machine learning”, ”artificial intelligence”, and “XAI”.  Rayyan online platform detected duplicates, inclusion and exclusion of studies.  This review identified 14 primary studies using XAI for breast cancer diagnosis from mammography and US images.  Out of the selected 14 studies, only 1 evaluated humans' confidence in using the XAI system-additionally, 92.86 % of identified studies identified dataset and dataset-related issues as research gaps and future direction.  The result showed that further research and evaluation are needed to determine the most effective XAI method for breast cancer.  The authors concluded that XAI was not conceded to increase users' and doctors' trust in the system.   For the real-world application, effective and systematic evaluation of its trustworthiness in this scenario is lacking.  Moreover, these researchers stated that further investigations are needed to enhance the interpretability of deep learning algorithms through overcoming explainable to accuracy trade-offs, as well as to examine the potential insights they can provide for clinicians’ decision-making.

In a systematic review, Schopf et al (2024) examined the literature regarding the performance of mammography-image based AI algorithms, with and without additional clinical data, for future breast cancer risk prediction.  These investigators carried out a systematic literature review using 6 databases (medRixiv, bioRxiv, Embase, Engineer Village, IEEE Xplore, and PubMed) from 2012 through September 30, 2022.  Studies were included if they used real-world screening mammography examinations to validate AI algorithms for future risk prediction based on images alone or in combination with clinical risk factors.  The quality of studies was assessed, and predictive accuracy was recorded as the AUC.  A total of 16 studies met inclusion and exclusion criteria, of which 14 studies provided AUC values.  The median AUC performance of AI image-only models was 0.72 (range of 0.62 to 0.90) compared with 0.61 for breast density or clinical risk factor-based tools (range of 0.54 to 0.69).  Of the 7 studies that compared AI image-only performance directly to combined image + clinical risk factor performance, 6 showed no significant improvement, and 1 showed increased improvement.  The authors concluded that early efforts for predicting future breast cancer risk based on mammography images alone revealed comparable or better accuracy to traditional risk tools with little or no improvement when adding clinical risk factor data.  These investigators stated that transitioning from clinical risk factor-based to AI image-based risk models may result in more accurate, personalized risk-based screening approaches.

Hussain et al (2024) stated that breast cancer is the leading cause of cancer-related fatalities among women globally.  Conventional screening and risk prediction models primarily rely on demographic and patient clinical history to devise policies and estimate likelihood.  However, recent advancements in AI techniques, especially DL, have shown promise in the development of personalized risk models.  These models leverage individual patient information obtained from medical imaging and associated reports.  In a systematic review, these investigators examined the available evidence on the use of DL to digital mammography, radiomics, genomics, as well as clinical information for breast cancer risk assessment.  They analyzed these studies and discussed their findings, highlighting the promising prospects of DL techniques for breast cancer risk prediction.  Furthermore, these investigators examined ongoing research initiatives and potential future applications of AI-driven approaches to further improve breast cancer risk prediction; thus, facilitating more effective screening and personalized risk management strategies.  They stated that as ongoing studies and future applications continue to evolve, the implementation of DL techniques in breast cancer risk modeling could revolutionize screening strategies and facilitate tailored risk management in women globally.  Given the substantial impact of breast cancer on women’s health, harnessing the power of AI in risk assessment represents a very important step toward early detection and improved patient outcomes.

The authors stated that there were several drawbacks in the studies discussed in this systematic review, along with their potential, such as limited data, lack of explainability of methods, and less attention to the ethical aspects of data usage and data privacy considerations.

• Data: The most common constraints in these studies included the use of homogeneous, uni-modal, and single-vendor data.  Many studies have relied on data from a single center.  Furthermore, these studies often employed a single modality, such as imaging (mammography or breast MRI), genomics, demographic, or clinical data.  It has also been observed that imaging data often originate from a single vendor.
• Short-term risk: Short-term risk assessments do not reflect the long-term risk of breast cancer.  This can limit women’s intention to take preventive measures such as chemo-prevention or prophylactic surgery.  Short-term risk does not encompass all factors such as genetic or environmental factors (BRCA mutations, lifestyle, and hormonal factors), which can influence breast cancer risk.  Women with rare risk profiles require personalized care because complex risk profiles are not adequately evaluated by short-term risk prediction models.
• Explainability: The explainability and interpretability of breast cancer risk prediction methods are among the most under-explored research aspects.  Presently, the focus of AI models is on improving accuracy rather than making models explainable.  Better explainable models are essential, as they will aid in enhancing trust in using AI for breast cancer risk prediction and ultimately improve patient care.
• Ethical Consideration and Data Privacy: AI-driven breast cancer risk prediction presents both moral and ethical challenges.  Previous studies of breast cancer risk prediction studies have focused less on the responsible use of data.  It is important to ensure that patient data is used responsibly and for the public good.  Studies suggested that patients are willing to consent to the use of their data in AI-based breast cancer risk research if they are used confidentially and responsibly with effective governance.

---

References