Description:
Digital breast tomosynthesis (DBT) uses existing digital mammography equipment with specialized software to obtain low dose images acquired in an arc (3D acquisition) which are then reconstructed into slices which can be viewed on a workstation. This allows for visualization of the breast in layers and therefore reduces the issue of tissue overlap. Currently at most sites both a 2D image as well as the 3D acquisition is obtained for each patient. Potential advantages of DBT are similar to those for full field digital mammography; more accurate estimation BI-RADS classification of a lesion (improved conspicuity), reduction of distortions, reduction of false positives associated with glandular clusters, greater security in the study of dense breasts, and the reduction of the number of recalls. Tomosynthesis involves some additional imaging time and doubling of the radiation exposure. To reduce the increased radiation exposure when obtaining both 2D images along with DBT, C-view software has been developed which allows 2D images to be generated as a part of the breast tomosynthesis exam. The 2D images created from C-View software are reviewed together with the tomosynthesis slices to make a clinical decision or diagnosis.

Background     
Conventional mammography produces 2D images of the breast. Overlapping tissue on a 2D image can mask suspicious lesions or make benign tissue appear suspicious, particularly in women with dense breast tissue. As a result, women may be recalled for additional mammographic spot views. Inaccurate results may lead to unnecessary biopsies and emotional stress, or to a potential delay in diagnosis. Spot views often are used to evaluate microcalcifications, opacities or architectural distortions; to distinguish masses from overlapping tissue; and to view possible findings close to the chest wall or in the retroareolar area behind the nipple.¹ The National Cancer Institute reports that approximately 20% of cancers are missed at mammography screening.² Average recall rates are approximately 10%, with an average cancer detection rate of 4.7 per 1,000 screening mammography examinations.³ The Mammography Quality Standards Act audit guidelines anticipate 2 to 10 cancers detected per 1,000 screening mammograms.⁴ Interval cancers, which are detected between screenings, tend to have poorer prognoses.⁵ 

Digital breast tomosynthesis was developed to improve the accuracy of mammography by capturing 3-dimensional (3D) images of the breast, further clarifying areas of overlapping tissue. Developers proposed that its use would result in increased sensitivity and specificity, as well as fewer recalls due to inconclusive results.⁶ Digital breast tomosynthesis produces a 3D image by taking multiple low-dose images per view along an arc over the breast. During breast tomosynthesis, the compressed breast remains stationary while the X-ray tube moves approximately 1º for each image in a 15º to 50º arc, acquiring 11 to 49 images.⁷ These images are projected as cross-sectional “slices” of the breast, with each slice typically 1-mm thick. Adding breast tomosynthesis takes about 10 seconds per view. In 1 study in a research setting, mean time (SD) for interpretation of results was 1.22 (1.15) minutes for digital mammography and 2.39 (1.65) minutes for combined digital mammography and breast tomosynthesis.⁸ 

With conventional 2D mammography, breast compression helps decrease tissue overlap and improve visibility. By reducing problems with overlapping tissue, compression with breast tomosynthesis may be reduced by up to 50%. This change could result in improved patient satisfaction.⁷ 

A machine equipped with breast tomosynthesis can perform 2D digital mammography, 3D digital mammography or a combination of both 2D and 3D mammography during a single compression. Radiation exposure from tomosynthesis is roughly equivalent to mammography. Therefore, adding tomosynthesis to mammography doubles the radiation dose, although it still is below the maximum allowable dose established in the U.S. Mammography Quality Standards Act.

Studies typically compare 1-view (i.e., mediolateral oblique [MLO] view), or more commonly, 2-view (MLO plus craniocaudal view) breast tomosynthesis alone or combined with standard 2D mammography with standard 2D mammography alone. A 2014 TEC Assessment⁹ (updated in 2015¹⁰) focused on 2-view tomosynthesis. The FDA Radiological Devices Panel, which reviewed this new modality in 2011, recommended that 2-view breast tomosynthesis is preferable to 1-view tomosynthesis (both used in combination with full-field digital mammography).¹¹ 

In May 2013, FDA approved new tomosynthesis software that permits creation of 2D images (called CView) from images obtained during tomosynthesis.¹² As a result, 2D mammography may become unnecessary, thereby lowering radiation dose. In other words, only the tomosynthesis procedure will be needed, and both 2D and 3D images will be created. It is too early to gauge how traditional mammography plus tomosynthesis compares with C-View plus tomosynthesis.

Regulatory Status
Table 1 provides a summary of DBT systems approved by the FDA through the premarket approval process. FDA product code: OTE. The tomosynthesis portion of the mammography unit is considered a separate mammographic module, and for a facility to use this module, the facility must apply to the FDA for certification that extends to the tomosynthesis module. The Mammography Quality Standards Act requires interpreting physicians, radiologic technologists, and medical physicists to complete eight hours of DBT training, and mandates a detailed mammography equipment evaluation before use.

Table 1. FDA-Approved DBT Systems

DBT: digital breast tomosynthesis; FDA: Food and Drug Administration; FFDM: full-field digital mammography; PMA: premarket approval; 2D: 2-dimensional; 3D: 3- dimensional.

Policy:
Digital breast tomosynthesis is considered MEDICALLY NECESSARY in the screening or diagnosis of breast cancer.

Policy Guidelines
77061 Digital breast tomosynthesis; unilateral
77062 bilateral
77063 Screening digital breast tomosynthesis, bilateral (list separately in addition to code for primary procedure)

Also effective Jan. 1, 2015, Medicare established an add-on HCPCS G code specific to diagnostic breast tomosynthesis:

G0279 Diagnostic digital breast tomosynthesis, unilateral or bilateral (list separately in addition to G0204 or G0206).

Prior to 2015, there were no specific CPT codes for this testing. The testing would have been reported with the appropriate breast mammography code (77055 – 77057 or G0202 – G0206) along with an unlisted code (e.g., 76499) for the additional views.

Benefit Application
BlueCard^®/National Account Issues
State or federal mandates (e.g., FEP) may dictate that all FDA-approved devices, drugs or biologics may not be considered investigational, and, thus, these devices may be assessed only on the basis of their medical necessity.

Rationale
Evidence reviews assess whether a medical test is clinically useful. A useful test provides information to make a clinical management decision that improves the net health outcome. That is, the balance of benefits and harms is better when the test is used to manage the condition than when another test or no test is used to manage the condition.

The first step in assessing a medical test is to formulate the clinical context and purpose of the test. The test must be technically reliable, clinically valid, and clinically useful for that purpose. Evidence reviews assess the evidence on whether a test is clinically valid and clinically useful. Technical reliability is outside the scope of these reviews, and credible information on technical reliability is available from other sources.

Digital Breast Tomosynthesis for Screening
Clinical Context and Test Purpose
The purpose of 3-dimensional (3D) DBT in patients who are being screened for breast cancer is to inform a decision whether to recall women for further diagnostic testing.

The question addressed in this portion of the review is whether there is sufficient evidence that 3D DBT, used to screen for breast cancer, improves the net health outcome compared with standard techniques. Specifically, is 3D DBT as an adjunct to 2-dimensional (2D) mammography or 3D DBT plus synthesized 2D mammography superior to mammography alone, and is 3D DBT instead of mammography at least as beneficial as mammography? For both interventions, are differences in accuracy likely to improve health outcomes via earlier diagnosis and treatment?

The following PICOs were used to select literature to inform this review.

Patients
The relevant population of interest are asymptomatic individuals being screened for breast cancer.

Interventions
The intervention of interest is 3D DBT screening as an adjunct to 2D mammography and 3D DBT plus synthesized 2D mammography. DBT devices approved in the U.S. are summarized in Table 1.

Comparators
The primary comparator of interest is mammography alone.

Outcomes
The reference standard is histopathology or at least one-year follow-up for women with negative findings.

The health outcomes of interest are:

• Overall and breast cancer-specific survival.
• Quality of life.
• Recall rates, which may lead to unnecessary follow-up testing and possibly unnecessary biopsies and treatment.
• Cancer risk from radiation exposure.

For breast cancer, the most important health outcome is an overall survival from the disease. DBT, as any breast screening test, may not directly improve breast cancer-specific survival; however, the higher sensitivity of breast DBT could lead to earlier cancer detection, which may, in turn, lead to improved health outcomes if earlier treatment is more effective. Although there is indirect evidence that earlier detection improves health outcomes, possible overdetection also needs to be taken into account. Overdetection would subject women to testing and treatment that does not improve health outcomes. When screening leads to diagnosis at an early stage, it may also affect the quality of life by permitting the use of less invasive or otherwise less difficult to tolerate treatments for breast cancer. If using breast DBT reduces the false positive
rate, it would reduce recalls for a diagnostic workup or for biopsy. Fewer unnecessary recalls would, in turn, have a positive impact on patient's quality of life by avoiding the anxiety and additional imaging associated with recalls. Finally, adding breast DBT to traditional mammography doubles the radiation dose, even though the combined dose remains below the limit set in the Mammography Quality Standards Act of 1992. The increased dose might be offset in part by fewer diagnostic tests if the recall rate falls. If synthesized mammography permits the use of tomosynthesis to create both 2D and 3D images, then the dose would be roughly equivalent to a single mammogram and increased radiation exposure would no longer be an issue.

Study Selection Criteria
For the evaluation of the clinical validity of DBT, studies that met the following eligibility criteria were preferred:

• Prospective studies (preferably in a U.S. setting)
• Comparing DBT plus mammography with mammography alone
• Including asymptomatic individuals being screened for breast cancer
• Including performance characteristics such as screening sensitivity and specificity (i.e., follow-up of negative findings and interval
• cancers for at least one year)

Several studies did not meet the preferred selection criteria, in particular, most lacked data on the follow-up of negative findings and interval cancers. The prospective studies without sufficient follow-up of negative findings are summarized briefly in tabular form following the discussion of studies with follow-up of negative findings.

Technically Reliable
Assessment of technical reliability focuses on specific tests and operators and requires a review of unpublished and often proprietary information. Review of specific tests, operators and unpublished data are outside the scope of this evidence review and alternative sources exist. This evidence review focuses on the clinical validity and clinical utility.

3D DBT as an Adjunct to 2D Mammography
Clinically Valid
A test must detect the presence or absence of a condition, the risk of developing a condition in the future, or treatment response (beneficial or adverse).

Prospective Studies With Long-Term Follow-Up of Negative Findings
Characteristics of prospective studies with follow-up of negative findings are shown in Table 2. The table includes three publications of previously reported prospective studies that have provided additional data including follow-up for interval cancers.

Houssami et al. (2018) reported on the results from STORM (Screening with Tomosynthesis OR standard Mammography) study, which assessed interval breast cancers, based on ascertainment at 2-year follow-up from screening examinations.¹¹ STORM examined comparative cancer detection for traditional mammography with or without DBT in a general population of 7292 asymptomatic Italian women being screened for breast cancer. In the initial screening of STORM, women were recalled if either of two independent readers recorded a positive result at either mammography alone or mammography plus DBT. Previous reports of STORM have summarized initial findings of one round of screening and partial follow-up of the cohort (summarized in the following section). The 2018 report focused on screening measures requiring completed ascertainment of interval cancers, i.e., interval cancer rates and screening sensitivity, including 2 years of follow-up. Interval cancers were identified using a combination of checking local hospital and pathology databases; and checking with the local cancer registry for cancer notifications. Interval cancer rates for concurrent Italian cohorts screened with 2D-mammography alone were provided for descriptive purposes. The study was not powered for formal comparisons of mammography alone to mammography plus DBT.

Similarly, Skaane et al. (2018) reported performance indicators and characteristics of screen-detected and interval cancers from 24,301 women in the Oslo Tomosynthesis Screening Trial (OTST) administered by the Norwegian Cancer Registry.^12,13 The OTST was designed to compare four different reading modes for mammography with or without DBT. The results reported in the 2018 publication include the double reading mammography plus DBT and double-reading mammography alone arms. Decisions regarding recalls from the initial screens were made by consensus conference review of images that were rated by any reader as any score other than negative or definitely benign. Previous reports from the OTST included initial results of one round of screening without the follow-up of negative results (summarized in the following section). The 2018 publication reported a comparison of mammography plus DBT in women from the OTST who had 2 years of follow-up with 2 previous mammography screening rounds in Oslo using data from the Norwegian Cancer Registry. The OTST (2019) publication reports final results of the OTST, including the sensitivity and specificity of all 4 arms.

Zackrisson et al. (2018) reported final results of the Malmö Breast Tomosynthesis Screening Trial (MBTST;NCT01091545), a population based screening study of 14,851 women ages 40 – 74 years attending a national breast cancer screening in Malmö, Sweden between 2010 and 2015.¹⁴ Preliminary results have been reported previously and are discussed in the following section. The 2018 publication included follow-up of participants for at least 2 years. MBTST was designed to compare one-view DBT to standard two-view digital mammography (DM). Decisions regarding recalls were made by consensus conference review of images that were rated by reading groups as any score other than negative or definitely benign.

Table 2. Characteristics of Prospective Studies With Long-Term Follow-Up of Negatives

OTST: Oslo Tomosynthesis Screening Trial; MBTST: Malmö Breast Tomosynthesis Screening Trial; STORM: Screening with Tomosynthesis OR standard Mammography.

Results of prospective studies meeting with sufficient follow-up are shown in Table 3. Nine interval cancers were detected in STORM; three were diagnosed within one year of screening and the remaining six were diagnosed between one and two years after screening. STORM reported an interval breast cancer rate in mammography plus DBT screening participants that were numerically lower (and screening sensitivity numerically higher) than the rate in 2D-screened women although confidence intervals (CIs) overlapped. These findings should be interpreted with caution given that STORM was not designed to compare interval cancer data and there were a small number of interval cases. Specificity was not reported in the publication; however, based on the information provided and the data on mammography plus DBT test results in the previous publications, it appears that the specificity was 96.6% (95%CI, 96.2% to 97.0%) in the STORM participants.

Interval cancer rates were similar in women who received mammography alone and DBT plus mammography in the report including the OTST participants. The OTST also reported numerically but not statistically higher sensitivity while also reporting statistically higher specificity of mammography plus DBT compared with mammography alone. Most of the additional DBT-detected cancers in the OTST were reported to be small node-negative invasive cancers of molecular subtypes known to have a good prognosis.

MBTST was a paired design and the 2018 results did not compare to a non-MBTST cohort (as in STORM and OTST reports). Therefore interval cancers were reported overall and not by screening modality; 139 breast cancers were detected in 137 (less than 1%) of 14,848 women; 89 were detected by both DBT and DM, 42 were detected only by DBT, and 8 were detected only DM. Most of the cancers detected by DBT and DM were invasive, and nearly all cancers detected by DBT only were invasive. DBT detected a higher number of invasive lobular cancers.

Table 3. Results of Prospective Studies With Long-Term Follow-Up of Negatives

CI: confidence interval; DBT: digital breast tomosynthesis;mammo: mammography; MBTST: Malmö Breast Tomosynthesis Screening Trial; NR: not reported; NPV:
negative predictive value; OTST: Oslo Tomosynthesis Screening Trial; PPV: positive predictive value; STORM: Screening with Tomosynthesis OR standard
Mammography.
^aSTORM participants were screened with both mammography and digital breast tomosynthesis. Women were recalled if either of two independent readers recorded a
positive result at either mammography alone or mammography plus digital breast tomosynthesis

The purpose of the limitations tables (see Tables 4 and 5) is to display notable limitations identified in each study. This information is
synthesized as a summary of the body of evidence and provides the conclusions on the sufficiency of the evidence supporting the position
statement. STORM, OTST, and MBTSTwere not conducted in a U.S. setting and screening practices differ in European countries. While
STORM and OTST included a prospective cohort of women receiving DBT plus mammography, the comparison group in the OTST study for
the purposes of the 2018 publication was a cohort previously screened with mammography alone (ie, not concurrent) and few details were
provided on selection of the women included in that cohort.

Table 4. Relevance Limitations of Prospective Studies With Long-Term Follow-Up of Negatives

The study limitations stated in this table are those notable in the current review; this is not a comprehensive limitations assessment.
OTST: Oslo Tomosynthesis Screening Trial; MBTST: Malmö Breast Tomosynthesis Screening Trial; STORM: Screening with Tomosynthesis OR standard Mammography.
a Population key: 1. Intended use population unclear; 2. Clinical context is unclear; 3. Study population is unclear; 4. Study population not representative of intended use.
bIntervention key: 1. Classification thresholds not defined; 2. Version used unclear; 3. Not intervention of interest.
c Comparator key: 1. Classification thresholds not defined; 2. Not compared to credible reference standard; 3. Not compared to other tests in use for same purpose.
d Outcomes key: 1. Study does not directly assess a key health outcome; 2. Evidence chain or decision model not explicated; 3. Key clinical validity outcomes not
reported (sensitivity, specificity and predictive values); 4. Reclassification of diagnostic or risk categories not reported; 5. Adverse events of the test not described
(excluding minor discomforts and inconvenience of venipuncture or noninvasive tests).
e Follow-Up key: 1. Follow-up duration not sufficient with respect tonatural history of disease (true-positives, true-negatives, false-positives, false-negatives cannot be
determined).

Table 5. Study Design and Conduct Limitations of Prospective Studies with Long-Term Follow-Up of Negatives

The e study limitations stated in this table are those notable in the current review; this is not a comprehensive limitations assessment.
OTST: Oslo Tomosynthesis Screening Trial; MBTST: Malmö Breast Tomosynthesis Screening Trial; STORM: Screening with Tomosynthesis OR standard Mammography.
a Selection key: 1. Selection not described; 2. Selection not random or consecutive (ie, convenience).
bBlinding key: 1. Not blinded to results of reference or other comparator tests.
cTest Delivery key: 1. Timing of delivery of index or reference test not described; 2. Timing of index and comparator tests not same; 3. Procedure for interpreting tests not
described; 4. Expertise of evaluators not described.
d Selective Reporting key: 1. Not registered; 2. Evidence of selective reporting; 3. Evidence of selective publication.
e Data Completeness key: 1. Inadequate description of indeterminate and missing samples; 2. High number of samples excluded; 3. High loss to follow-up or missing
data.
f Statistical key: 1. Confidence intervals and/or p values not reported; 2. Comparison with other tests not reported.

Prospective Studies Without Long-Term Follow-Up of Negative Results
Other prospective studies assessing the diagnostic accuracy of DBT for screening are summarized in Table 6. The table is subdivided by the characteristics of study designs. Select studies are summarized briefly following the table. In general, these studies do not have follow-up sufficient to capture interval cancers and therefore traditional measures of sensitivity and specificity are not provided.

Table 6. Prospective Studies of DBT for Breast Cancer Screening Without Long-Term Follow-Up of Negatives

CI: confidence interval; DBT: digital breast tomosynthesis; Mammo: mammography; MBTST: Malmö Breast Tomosynthesis Screening Trial; NR: not reported; OTST: Oslo
Tomosynthesis Screening Trial; PPV: positive predictive value; s2D: synthesized 2D mammography; STORM: Screening with Tomosynthesis OR standard
Mammography.
^aData from Ciatto et al. (2013) and Houssami et al. (2014).
^bU.S. population; high-risk preferentially included.
^cTwenty-seven women with no follow-up not included in results.
^dU.S. population; sample enriched with women referred for biopsy (22%).
^eRange across 12 radiologists in study 1 and 15 radiologists in study 2.

Randomized Controlled Trials
Two RCTs have compared screening with mammography alone with mammography plus DBT. Pattacini et al. (2018) reported on the preliminary results from the Reggio Emilia Tomosynthesis trial, which compare mammography plus DBT with mammography alone in women in Italy ages 45 to 74 who had previously been screened with mammography.¹⁶ The trial is designed to enroll 40000 women and compare interval cancers with a cumulative incidence of advanced cancer and had 4.5 years of follow-up. The 2018 publication focuses on the preliminary results for the baseline screen of 19560 women recruited from 2014 to 2016, including cancers diagnosed within 9 months from recruitment and, as such, cannot yet provide data on interval cancers and confirmation of negative findings. Results are shown in Table 6.

Maxwell et al. (2017) reported on the results of a trial of asymptomatic women from 2 centers in the U.K. ages 40 to 49 years who had previously undergone mammography for an increased risk of breast cancer.¹⁷ Participants were randomized in a crossover design to screening with 2D mammography followed by 2D mammography plus DBT a year later, or vice versa. The trial was designed to compare recall rates. Results are shown in Table 6. The crossover design limits the utility of collecting long-term results.

In summary, recall rates did not differ for mammography alone vs mammography plus DBT in either RCT. Maxwell et al. (2017) also reported no statistically significant difference in cancer detection rate. However, preliminary results from Reggio Emilia Tomosynthesis trial would suggest an almost 90% increase in detection rate for mammography plus DBT compared with mammography (relative risk [RR], 1.89; 95% CI, 1.31 to 2.72) and an increase in the PPV for recalls from 13.0% to 24.1%. The gain in cancer detection was observed for all classes of cancers except for very large or late cancers. There were more instances of ductal carcinoma in situ (DCIS) with mammography plus DBT (+1 per 1,000), benign lesions (+1 per 1,000), and invasive cancers (+3 per 1,000). There was also an increase in the risk of surgery for mammography plus DBT (RR=1.90; 95% CI, 1.35, 2.68; risk difference, 5 per 1000; 95% CI, 2 to 7).

Prospective Observational Studies
Lång et al. (2016) reported exploratory results from the first half of the Malmö Breast Tomosynthesis Screening Trial, comparing 1-view (mediolateral oblique) DBT (a lower radiation dose than DM) with 2-view DM.¹⁶ The Malmö Breast Tomosynthesis Screening Trial is a 1-arm, single institution, prospective study. Randomly selected women in Sweden (age range, 40 – 74 years) were offered 1-view DBT and 2-view DM. A sample size of 15,000 was specified to detect an improvement in cancer detection sensitivity from 63% to 88% (power, 80%); 7,500 were included in the exploratory analysis. In Sweden, breast cancer screening is offered to women between ages 40 and 55 every 18 months and every 24 months after that to age 74. Six experienced readers interpreted images (mean experience, 26 years; range, 8 – 41 years).

Blinded double-reading was carried out for DBT and DM with rule-based arbitration of disagreements women in this exploratory analysis were followed at least one year for the development of cancer ascertained through the South Swedish Cancer Registry. Of 10,547 women invited, 71.1% participated with 20% undergoing their first screening test. Results are shown in Table 6. DCIS detection rates were similar between both modalities. Following arbitration, the recall rate was lower for DM (2.6%; 95% CI, 2.3% to 3.0%) than for DBT (3.8%; 95% CI, 3.3% to 4.2%; p < 0.001).

The results of the analysis of a cohort from a large trial, the OTST comparing 4 different reading modes, was published by Skaane et al. (2013) in Norway.^20,25 The Skaane et al. (2013) analysis was a preplanned interim analysis of 2 arms in a larger 4-arm trial; findings of the other 2 arms are not relevant to this topic. The sample included 12,621 women with 121 cancers detected during routine screening.²⁶ Results are shown in Table 6. After adjusting for reader differences, the ratio of cancer detection rates for mammography plus DBT vs mammography alone was 1.27 (98.5% CI, 1.06 to 1.53; p = 0.001). The trialists did not ascertain any increase in detecting DCIS by adding breast tomosynthesis (i.e., additional cancers detected were mostly invasive). In Norway, as in much of Europe, women are screened every other year, and two readers independently interpret the images, which differs from usual practice in the U.S. After adjusting for differences across readers, the ratio of false-positive rates for mammography plus DBT vs mammography alone was 0.85 (98.5% CI, 0.76 to 0.96; p < 0.001).

The STORM study examined comparative cancer detection for traditional mammography with or without DBT in a general population of 7,292 asymptomatic Italian women being screened for breast cancer.^21,22 The reference standard was pathology results for women undergoing biopsies; women with negative results on both mammography and DBT were not followed so neither sensitivity nor specificity could be calculated. Results are shown in Table 6. Mammography plus DBT revealed all 59 cancers; 20 (34%) were missed by traditional mammography (p < 0.001). In the original report, incremental cancer detection by using both modalities was 2.7 cancers per 1,000 screens (95% CI, 1.7 to 4.2). There were 395 false-positive results: 181 were false-positive using either mammography or both imaging modalities together; an additional 141 occurred using mammography only, and 73 occurred using mammography and DBT combined (p < 0.001). In preplanned analyses, combined results of mammography and DBT yielded more cancers in both age groups (less than 60 vs ≥ 60 years) and breast density categories (1 [least dense] and 2 vs 3 and 4 [most dense]). In a follow-up report including available data on interval cancers diagnosed in the first year of follow-up (note, screening was repeated at two years), six additional interval cancers had been diagnosed. The cancer detection rates including the 6 additional cancers were 4.8 (95% CI, 3.3 to 6.7) vs 7.5 (95% CI, 5.7 to 9.8) for mammography vs mammography plus DBT, for an incremental cancer detection rate of 2.7 (95% CI, 1.6 to 4.2; p < 0.01).

Retrospective Studies
Several retrospective studies have also been performed, many of which included several thousand patients and 3 of which included more than 100,000 patients. Many of the retrospective studies have included mixed populations or unclear indications for screening and inadequate reference standards such as historical controls and are therefore not discussed in detail. Results are summarized briefly in Appendix Table 1.

Retrospective studies have, in general, suggested increases in the rates of cancer detection and decreases in recall and false-positive rates.

Systematic Reviews
Marinovich et al. (2018) reported results of a systematic review that included prospective and retrospective studies published through 2017.²⁷ Given that the review does not include the prospective studies with long-term follow-up, it will not be discussed further.

Characteristics of Detected Cancers
Yun et al. (2017) published a meta-analysis assessing the characteristics of cancers detected with DM alone vs DM plus DBT during routine breast cancer screening.²⁸ Eleven studies were included in the meta-analysis, four prospective and seven retrospective observational studies, all of which are described in Table 2 (above). Reviewers evaluated study quality using the Quality Assessment of Diagnostic Accuracy Studies tool and found an overall satisfactory risk of bias, but all studies had a high risk of bias concerning the reference standard as well as flow and timing because patients who were not recalled did not have a reference standard test (i.e., did not have biopsy-confirmed negative findings).

In a pooled analysis, the overall cancer detection rate was significantly higher with DM plus DBT than with DM alone (RR = 1.29; 95% CI, 1.16 to 1.43; I2 = 0%). Moreover, the detection of invasive cancer was significantly higher in the DM plus DBT group compared with DM alone group (RR = 1.33; 95% CI, 1.17 to 1.51; I2 = 7%). The rate of carcinoma in situ detection did not differ significantly between the DM plus DBT group and the DM alone group (RR = 1.20; 95% CI, 0.94 to 1.52; I2 = 29%). Fewer studies reported on cancer detection by T and/or N stage. In a pooled analysis of 5 studies, there was a significantly higher rate of detecting T1 cancers with DM plus DBT than with DM alone (RR = 1.39; 95% CI, 1.14 to 1.70; I2 = 0%), but no significant difference for detecting stage T2 or larger cancer (RR = 1.39; 95% CI, 0.90 to 2.16; I2 = 0%).Similarly, there was a significantly higher rate of detection of stage N0 cancers with DM plus DBT than with DM alone (RR = 1.45; 95% CI, 1.21 to 1.74; I2 = 0%) and no significant difference in the detection of stage N1 or higher cancers (RR = 1.34; 95% CI, 0.92 to 1.99; I2 = 0%). The numbers of more advanced cancers were relatively small, and the pooled analyses of T2 or higher and N1 or higher cancers might have been underpowered. The findings of this meta-analysis were limited by the potential biases of the included studies (e.g., many were retrospective and studies had insufficient confirmatory data on negative imaging results).

Clinically Useful
A test is clinically useful if the use of the results informs management decisions that improve the net health outcome of care. The net health outcome can be improved if patients receive correct therapy, or more effective therapy, or avoid unnecessary therapy, or avoid unnecessary testing.

Direct Evidence
Direct evidence of clinical utility is provided by studies that have compared health outcomes for patients managed with and without the test. Because these are intervention studies, the preferred evidence would be from RCTs.

There is no direct evidence from trials comparing health outcomes in patients screened for breast cancer using DBT and mammography.

Chain of Evidence
Indirect evidence on clinical utility rests on clinical validity. If the evidence is insufficient to demonstrate test performance, no inferences can be made about clinical utility.

A chain of evidence should demonstrate that DBT used as an adjunct to screening improves screening performance compared with standard mammography alone. Available studies have reported that adding DBT to mammography may increase cancer detection and reduce unnecessary recalls. Even if adding breast tomosynthesis simply maintained the same sensitivity as mammography, a decline in the false positive rate would reduce the substantial number of unnecessary diagnostic workups in the U. S.

Three prospective studies (STORM, OTST, MBTST) with a two-year follow-up for interval cancers have been published although none wereconducted in the U.S. The OTST had prospective data on the mammography plus DBT cohort but compared outcomes with previously screened cohorts from a cancer registry. None were powered to compare interval cancer rates. STORM reported an interval breast cancer rate in mammography plus DBT screening participants that were numerically lower (and screening sensitivity numerically higher) than the rate in 2D-screened women although CIs overlapped. The OTST also reported numerically but not statistically higher sensitivity. However, the OTST did report statistically significantly higher specificity of mammography plus DBT compared with DBT alone. MBTST reported statistically higher sensitivity but not specificity.

• Two RCTs without sufficient follow-up to detect interval cancers have reported no difference in recall rates between DBT plus mammography and mammography alone. However, 1 RCT reported approximately a 90% increase in detection rate for DBT plus mammography compared with mammography with more instances of DCIS with mammography plus DBT (+1 per 1,000), benign lesions (+1 per 1,000), and invasive cancers (+3 per 1,000) and an increase in the PPV for recalls from 13.0% to 24.1%. This RCT is ongoing and is designed to compare interval cancers and cumulative incidence of advanced cancer with 4.5 years of follow-up at completion.
• While the incremental radiation per individual is not large, the aggregate impact of that increased radiation dose over a large group can raise greater concern. Although any elevated dose, related to DBT may be offset by fewer diagnostic images required for women who are recalled for further evaluation, it needs to be considered. Synthesized mammography may resolve this issue (discussed in the following section).
• There has been widespread debate over the value of mammography that hinges in large part on beliefs about whether there is substantial overdetection of breast cancer during screening. An argument in favor of tomosynthesis is that the probability of overdetection is lower because most of the additional cancers detected are invasive. On the other hand, mammography is included with tomosynthesis in part because of concern that readers of tomosynthesis images may miss microcalcifications, some of which are malignant.

In summary, estimates of sensitivity and specificity of DBT plus mammography from studies with adequate follow-up of negative results are available from three studies. The sensitivity of DBT plus mammography is likely to be at least as high as mammography alone. One study with limitations reported the specificity of DBT plus mammography was significantly higher than mammography alone but another reported no difference in specificity. An increase in specificity (corresponding to a decrease in the false positives) would reduce unnecessary diagnostic workups and their consequences. Two RCTs with short follow-up reported similar recall rates for DBT plus mammography and mammography alone but one of the RCTs reported a significant increase in cancer detection rate, including invasive cancer and DCIS.

Subsection Summary: Screening With 3D DBT as an Adjunct to 2D Mammography
There is a lack of direct evidence on the clinical utility of 3D DBT from screening trials comparing health outcomes in patients screened for breast cancer with 3D DBT vs 2D mammography. Current evidence would suggest that the use of mammography plus breast tomosynthesis may modestly increase the number of cancers detected, with a potential decrease in the number of women who undergo unnecessary recalls or biopsies. A 2017 meta-analysis including a pooled analysis of 11 screening studies found a significantly higher rate of invasive cancer detection with 3D DBT plus 2D DM than with 2D DM alone. Preliminary data from an RCT also found higher rates of invasive cancer with 3D DBT plus DM.

3D DBT Plus Synthesized 2D Mammography
Clinically Valid
A test must detect the presence or absence of a condition, the risk of developing a condition in the future, or treatment response (beneficial or adverse).

No prospective studies with sufficient follow-up for interval cancers and negative findings were identified.

One systematic review of 3D DBT plus s2D vs DM plus DBT for breast cancer screening has been published. Characteristics are shown in Table 7. Houssami et al. (2018) included studies that evaluated s2D plus DBT compared with DM plus DBT for population screening and provided quantitative data on screening detection measures (cancer detection and recall measures).²⁹ Five studies were identified.^{24,30,31,32,33} The studies included in the Houssami et al. (2018) systematic review, with the exception of Skaane et al. (2014),³³ all included a comparison of DM and DM plus DBT in addition to the synthesized digital mammography (sDM) plus DBT arm and as such were included in Table 6 and Appendix Table 1.

Table 7. Characteristics of Systematic Reviews of DBT Plus s2D Mammography

DBT: digital breast tomosynthesis; DM: digital mammography;NR: not reported; s2D: synthesized 2-dimensional.
Results of the systematic review are shown in Table 8. Meta-analyses were not conducted; instead, qualitative summaries were provided.
Cancer detection rates appear similar between DM plus DBT (range, 5.45 to 8.5 per 1,000 screens) and s2D plus DBT (range, 5.03 – 8.8 per
1,000 screens). The recall rates appear heterogeneous across included studies. The mean glandular dose for s2D plus DBT was 55% to 58%
of DM plus DBT. The systematic review did not include a risk of bias or quality assessment. However, all of the included studies had
limitations similar to the studies in the previous setting, i.e., lack of follow-up for interval cancers or confirmation of negative results.

Table 8. Results of Systematic Reviews of DBT Plus s2D Mammography

DBT: digital breast tomosynthesis; DM: digital mammography; NR: not reported; s2D: synthesized 2-dimensional.
^aTwo studies reported recall and three studies reported false-positive recall.
The Skanne et al. (2014) study from the systematic review and other studies published following the systematic review are briefly summarized
in Table 9.³³ None has sufficient follow-up to evaluate interval cancers.

Table 9. Other Studies of DBT Plus sDM for Breast Cancer Screening

CI: confidence interval; DBT: digital breast tomosynthesis; DM: digital mammography; PPV: positive predictive value; sDM: synthesized digital mammography; s2D:
synthesized 2-dimensional.

Clinically Useful
A test is clinically useful if the use of the results informs management decisions that improve the net health outcome of care. The net health outcome can be improved if patients receive correct therapy, or more effective therapy, or avoid unnecessary therapy, or avoid unnecessary
testing.

Direct Evidence
Direct evidence of clinical utility is provided by studies that have compared health outcomes for patients managed with and without the test.

Because these are intervention studies, the preferred evidence would be from RCTs.

There is no direct evidence from trials comparing health outcomes in patients screened for breast cancer using DBT and mammography.

Chain of Evidence
Indirect evidence on clinical utility rests on clinical validity. If the evidence is insufficient to demonstrate test performance, no inferences can be made about clinical utility.

Given that the utility of breast cancer screening with mammography has been established, a chain of evidence should demonstrate that screening performance of DBT plus synthesized 2D is equivalent to that of standard mammography alone. Available studies have reported that replacing mammography with DBT plus s2D might increase cancer detection and reduce recall rates. However, performance characteristics are uncertain due to the limitations described above in the section on the clinical utility of DBT plus acquired mammography, and thus it is not possible to construct a chain of evidence.

Subsection Summary: Screening With 3D DBT Plus Synthesized 2D Mammography
Preliminary results of one RCT, two prospective and three retrospective studies have assessed 3D DBT plus s2D mammography, which has lower radiation exposure than 3D DBT plus DM. In the RCT, the rate of cancers detected was similar for DBT plus s2D compared to DM but recall rates were lower. Two observational studies found higher detection rates with 3D DBT plus s2D compared with DM, one found similar detection rates with 3D DBT plus s2D compared with DM, and two found similar detection rates with 3D DBT plus s2D compared with 3D DBT plus DM. When comparing the recall rate of 3D DBT plus s2D with DM alone, one prospective observational study and one RCT found a higher recall rate for DM and one prospective study found similar rates, while the retrospective studies had mixed findings. However, the potential for overdiagnosis cannot be ascertained because of the study designs, and interval cancer rates are not yet available. The nonrandomized designs lack long-term follow-up to assess false-negative results. The RCT is designed to continue follow-up for 2 years with completion expected in 2022.

There is a lack of direct evidence on the clinical utility of DBT from screening trials comparing health outcomes in patients screened for breast cancer with DBT vs mammography. Due to limitations in the studies on diagnostic accuracy, it is not possible to construct a chain of evidence.

3D DBT for Diagnosis
Clinical Context and Test Purpose
The purpose of 3D DBT in patients who have screen-detected abnormalities suspicious for breast cancer is to inform a decision whether to biopsy.

The question addressed in this portion of the evidence review is whether there is sufficient evidence that DBT used to detect breast cancer in patients with abnormal findings on breast imaging or clinical exam improves the net health outcome compared with standard techniques. Specifically, is 3D DBT at least as accurate as standard methods for diagnosing breast cancer and is this degree of increased accuracy likely to improve health outcomes via the earlier diagnosis, better patient management decisions, and more appropriate treatment?

The following PICOTS were used to select literature to inform this review.

Patients
The relevant population of interest are individuals with abnormal findings on breast imaging or a clinical examination.

Interventions
The intervention of interest is 3D DBT as an adjunct to 3D mammography for diagnosis.

Comparators
The comparators of interest are standard diagnostic methods. Diagnosis includes both physical examination and imaging. Diagnostic imaging may include diagnostic mammography and ultrasonography. Magnetic resonance imaging for the diagnosis of breast cancer is discussed in evidence review 6.01.29.

Outcomes
The beneficial outcomes of a true-negative test result, are an avoidance of invasive procedures (e.g., biopsy or mastectomy). The beneficial outcomes of a true-positive test result are reductions in overall mortality and breast cancer-specific mortality.

The harmful outcomes of a false-negative test result are a delay in treatment and a potential increase in mortality. The harmful outcomes of false-positive test results are unnecessary invasive procedures.

Timing
DBT for diagnosis would be performed after a positive breast cancer screening examination.

Setting
The test would be performed in an outpatient imaging setting.

Study Selection Criteria
For the evaluation of the clinical validity of DBT, studies that met the following eligibility criteria were selected:

• Prospective studies (preferably in a U.S. setting)
• Comparing DBT plus mammography with diagnostic evaluation alone
• Appropriate reference standard (histopathology)
• Including performance characteristics (e.g., sensitivity, specificity)

Technically Reliable
Assessment of technical reliability focuses on specific tests and operators and requires a review of unpublished and often proprietary information. Review of specific tests, operators, and unpublished data are outside the scope of this evidence review and alternative sources exist. This evidence review focuses on the clinical validity and clinical utility.

Clinically Valid
A test must detect the presence or absence of a condition, the risk of developing a condition in the future, or treatment response (beneficial or adverse).

Prospective Studies
As per the selection criteria, the characteristics of prospective studies are described in Table 10. The reference standard used for all included studies was histopathology. These prospective studies were conducted in Europe and Asia. Heywang-Kobrunner et al. (2017)⁴⁰ and Thibault et al. (2013)⁴¹ used single-view DBT while Seo et al. (2016)⁴² used double-view.

Table 10. Characteristics of Prospective Studies of DBT Diagnostic Performance

BIRADS: Breast Imaging Reporting and Data System; DBT: digital breast tomosynthesis; NR: not reported.

Results of the studies meeting selection criteria are shown in Table 11. Precision estimates for performance characteristics such as sensitivity and specificity were only provided in Teertstra et al. (2010),⁴³ in which the diagnostic performance of DBT was very similar to DM and Seo et al (2016),⁴² who reported that the sensitivity of DM plus DBT was significantly higher than DM alone (pless than 0.001). Only Thibault et al. (2013) compared DBT with DM plus ultrasonography; adding DBT to DM plus ultrasound did not improve the estimated area under the curve.⁴¹

Table 11. Results of Prospective Studies of DBT Diagnostic Performance

DBT: digital breast tomosynthesis; DM: digital mammography; NPV: negative predictive value; NR: not reported; PPV: positive predictive value; US: ultrasonography.

The studies included in the tables above were prospective, consecutively enrolled participants, and used an appropriate reference standard. Notable limitations identified in each study are shown in Tables 12 and 13. Only one study compared DBT with DM plus ultrasonography and one study provided precision estimates for performance characteristics such as sensitivity and specificity.

Table 12. Relevance Limitationsof Prospective Studies of DBT Diagnostic Performance