Overcoming Barriers in Ductal Carcinoma In Situ Management: From Overtreatment to Optimal Treatment

Ductal carcinoma in situ (DCIS) is an intraepithelial breast neoplasm accounting for up to 25% of US breast cancers.^1,2 Treatment for this nonlethal diagnosis remains controversial and ranges widely from no surgery to total mastectomy, often with radiation (RT) and/or endocrine therapy (ET). Recent SEER data show that only 2% of US women receive no surgical treatment, whereas 28% undergo mastectomy, often with sentinel node biopsy.³ Although mastectomy rates for DCIS have decreased over the past 40 years, use of RT after wide local excision (WLE) increased by nearly 20%³ along with a modest increase in ET use.⁴ Consequently, DCIS has become a major driver of breast cancer overdiagnosis and overtreatment.⁵ A 2009 National Institutes of Health (NIH) State of Science Conference recommended concerted efforts to decrease indolent DCIS diagnoses, unnecessary surgeries, and excessive adjuvant treatment.⁶ Yet, today, despite promising imaging and molecular assays to help optimize DCIS management, these patterns remain largely unchanged or perhaps more prevalent. In this commentary, we describe ongoing controversies in DCIS diagnosis and management that could influence future clinical trial design.

The natural history of DCIS is poorly understood, and the clinical benefit from increased diagnosis and resulting interventions is difficult to quantify. DCIS does not metastasize, but some will become invasive breast cancer (IBC).⁷ However, IBC frequency has not decreased proportionately with greater DCIS incidence.⁸ Therefore, clinicians and patients will benefit from improved clinical decision tools that account for the impact of observation versus intervention on outcomes other than survival, including local control and quality of life (QOL).

In an effort to decrease unnecessary DCIS detection in lower-risk women, strategies that shift screening from population-based⁹ or age-based¹⁰ to risk-based approaches may be beneficial. For example, the Wisdom Trial assigns women to screening regimens on the basis of individual risk (risk prediction model plus polygenic risk score).¹¹ However, available population-based risk models lack precision for individual clinical decision making, and could put some patients at risk of developing later-stage invasive cancers.¹²

Advanced imaging may help further refine both individual risk estimates and discrimination between higher- and lower-risk DCIS. Mammographic density is associated with breast cancer risk,¹³ and greater normal tissue background parenchymal enhancement levels on magnetic resonance imaging (MRI) may improve individual risk assessments over density alone.^14-16 Risk assessment tools on the basis of deep learning mammography¹⁷ and MRI¹⁸ models may outperform established models like Tyrer-Cuzick.

MRI has greater sensitivity for high- versus low-grade DCIS, is nearly 100% sensitive for invasive carcinoma, and is not affected by density. Thus, MRI could decrease overdiagnosis of low-grade DCIS if used to determine which types of calcifications warrant biopsy¹⁹ or for screening women with denser breasts if used alone.²⁰ But, MRI is not without barriers like cost, false-positives, claustrophobia, and exposure to IV contrast. Still, initial studies evaluating economic feasibility are promising,^21,22 and abbreviated breast MRI sequences²³ could reduce imaging and interpretation times and costs without jeopardizing performance.²⁴

There are also questions about the need to immediately surgically excise DCIS. The Comparing an Operation to Monitoring, with or without Endocrine Therapy for low-risk DCIS (COMET),²⁵ LOw RISk DCIS (LORIS),²⁶ and Management of LOw-Risk DCIS (LORD)²⁷ trials compare surgery to active surveillance (±ET) to identify women who could forego or delay upfront surgery. Unfortunately, both LORD and LORIS trials closed prematurely without meeting accrual goals because of recruitment challenges. Also, a recent single-institution study suggested that application of the COMET Trial's eligibility criteria to a retrospective cohort resulted in an IBC underdiagnosis of 12% over 10 years.²⁸ These challenges highlight a critical need to develop more accurate imaging and tissue-based measures to identify low-risk DCIS if active surveillance is to become a viable approach.

Once intervention is planned, surgery remains the upfront standard. However, re-excisions are needed in 30%-40% of cases, and it is unclear whether MRI can reduce re-excisions without increasing mastectomy rates.^3,29 The high re-excision rates observed with DCIS are because of poor delineation of noncalcified components of DCIS on mammography, lack of palpability, and guidelines from 2016 recommending ≥ 2-mm margins.³⁰ MRI may better delineate DCIS extent, with numerous studies showing that MRI size more closely matches pathology span than mammography.^31,32 Overall, MRI sensitivity to accurately determine disease extent approaches 90% versus 55% with mammography.^30,33

Still, the potential for surgical outcome improvements with MRI remains unclear. A meta-analysis³⁴ and systematic review³⁵ found limited evidence that MRI reduced reoperation rate, although many of the included studies predated published guidelines to improve quality of imaging acquisition³⁶ and interpretation.³⁷ The single-arm ECOG-ACRIN E4112 DCIS trial showed that when MRI is performed in accordance with American College of Radiology accreditation requirements, rate of successful WLE is high (96%) while re-excision (22%) is low.³⁸ Recent retrospective observational studies showed that MRI decreased number of surgeries³⁹ and lowered positive margin rate without increasing mastectomies.⁴⁰ Finally, a small trial of MRI (IRCIS) showed a lower reoperation rate in the MRI arm in the per-protocol analysis (hazard ratio = 0.59; 95% CI, 0.35 to 1.0), although not in the intention-to-treat analysis (hazard ratio = 0.68; CI, 0.4 to 1.1).⁴¹ Given conflicting data, uncertainty persists about whether MRI can optimize DCIS surgical management. This is likely not going to be resolved without larger definitive trials.⁴² Additional technologies are in development to provide real-time assessment of margins using radiofrequency electrical fields^43,44 and advanced localization techniques,^45,46 but these have yet to be rigorously studied to determine impact on surgical outcomes.

Understanding which patients with DCIS require additional therapy after surgery is key to reduce overtreatment. RT is often offered after WLE to reduce risk of in-breast recurrence (IBR),^47,48 which varies substantially with clinicopathologic factors (CPF), including extent, grade, comedonecrosis, margin, age, family history, estrogen receptor status, and use of ET.^49,50 Nonetheless, nearly 75% of women receive RT^3,51,52 with no consensus about who can safely avoid it. The Radiation Therapy Oncology Group (RTOG) trial 9804 randomly assigned women with low-risk DCIS to WLE ± RT, and found IBR rates to be low overall, albeit higher in those who did not receive RT (15.1% v 7.1% at 15 years), and modestly continuing to increase with longer follow-up.^53-55 ECOG E5194 omitted RT in clinically low-risk DCIS, and reported a 14.4% 12-year IBR rate in non–high-grade lesions < 2.5 cm and 24.6% in high-grade lesions < 1 cm.⁵⁶ Clinical models incorporating CPF to guide DCIS treatment report a wide range of IBR risk depending on CPF, treatment, and length of follow-up.^49,50 In total, it remains challenging to individualize treatment decisions, and consequently, most patients receive RT.

Two tissue-based assays to assess DCIS IBR risk are commercially available in the United States. Oncotype DCIS Score (EXACT Sciences, Madison, WI) is a 12-gene molecular assay that estimates a 10-year IBR risk. The results were initially validated using banked tissue from E5194 (no RT) and found significantly lower 10-year IBR rates in the low-risk group (IBR 10.6%) versus the intermediate-risk (IBR 26.7%) and high-risk (IBR 25.9%) groups.⁵⁷ Further prognostic validation of this assay was done in the Ontario Registry⁵⁸ including impact of combined CPF and DCIS score,^59,60 and it was also used in the E4112 trial to guide RT use.³⁸ E4112 showed that use of DCIS score results in lower RT utilization compared with historical data. Half of the study participants had low DCIS scores, and more than 90% of them accepted a recommendation to forego RT.³⁸

A second assay, DCISionRT (PreludeDx, Laguna Hills, CA), used a unique combination of CPF to assess risk of recurrence on a 0-10 scale. Scores ≥ 3 denote high-risk DCIS that corresponds to a 10-year IBR-invasive risk ≥ 10%. DCISionRT was initially studied in archived DCIS and found to distinguish women into a low-risk group (IBR 7%) and a high-risk group (IBR 23%). It also suggested a benefit from RT no > 1% in the low-risk group compared with 12% in the high-risk group^61,62 with similar findings in a separate registry data set, in which nearly half of the RTOG-eligible low-risk DCIS were reclassified as higher-risk.⁶³ This and other series suggest that the use of a risk score may affect RT recommendations beyond routinely available CPF, yet neither assay has been prospectively tested in randomized trials, which may explain their low penetration in routine care.⁶⁴

The wide range of dose-fractionation and shorter treatment schedules available in DCIS further complicate studies assessing RT. In the setting of whole breast irradiation, moderately hypofractionated schedules of 15-20 treatments are now preferred over more protracted courses.⁶⁵ Partial breast irradiation is available to many patients with lower-risk DCIS,⁶⁶ and data are emerging on very short-course extreme hypofractionation to the whole breast.⁶⁷ As more convenient RT options become available, their perceived burden may be lower. Thus, RT approaches, dose and fractionation, and utilization will evolve over time, and this also requires further study.

ET is also commonly used either instead of or in addition to RT. About half of patients with estrogen receptor–positive DCIS receive ET, which is expected to halve the risk of a new breast cancer event in either breast.⁴ Although an aromatase inhibitor may be more effective than tamoxifen,⁶⁸ low estrogen–related toxicities may affect adherence and reduce potential benefit. Thus, tamoxifen remains an option and there is interest in exploring lower doses.⁶⁹

In this context of a wide array of adjuvant therapy options and variations, it is important not to lose sight that reductions in IBR are unlikely to impact survival in DCIS.⁵¹ Therefore, patients at a higher risk of IBR could still reasonably forego RT and/or ET, and undergo salvage treatment at the time of recurrence. There is no uniformly acceptable threshold for what constitutes a low risk of IBR, and individual patient preferences will matter.

These observations point to other important gaps in knowledge, including the impact of patient experience and QOL metrics on individual decision making and their relationship to DCIS outcomes. Data on decision making and impact of management choices on QOL among patients with DCIS remain scant and are largely cross-sectional using surveys and retrospective events.^70,71 E4112 showed that patients' level of knowledge about DCIS and treatment options was associated with post-treatment physical and mental well-being,⁷² and several studies highlight patients' desire for DCIS knowledge and decision support.^73-76 Therefore, validation and implementation studies evaluating tools to facilitate tailored decision making are another pressing need in DCIS management.

External factors are also important. Racial disparities in breast cancer diagnosis and outcomes are well established⁷⁷ and have been exacerbated by the COVID-19 pandemic.⁷⁸ Although incidence of DCIS has stabilized in the United States since 2000, rates have increased in Black women,⁷⁹ and breast cancer mortality after DCIS is higher among non-Hispanic Black women compared with non-Hispanic Whites.⁵¹ Explanations for racial differences in clinical outcomes are most likely a complex interplay between treatment-related disparities and socioenvironmental exposures.^80,81 Future research should leverage concepts such as allostatic load––a composite measure of physiologic dysregulation secondary to adverse socially patterned exposures (eg, poverty⁸²)—to understand the implications of race and socioenvironmental factors on epidemiology,⁸³ treatment receipt, tolerability, and mortality, and identify targets for precision therapy, including biobehavioral intervention.⁸⁴

In future studies, several barriers must also be addressed to allow translational studies to improve DCIS management. First is the need to bank fresh tissue and images across sites in prospective trials to identify novel molecular, pathology, and imaging markers. Follow-up beyond 10 years is needed to capture meaningful outcomes like long-term IBR. Studies must also account for heterogeneity in multimodality care that is driven by factors like region, ethnicity, age, and socioeconomic or insurance status,⁸⁵ which also affect clinical trial participation. Fortunately, success in recruitment to recent DCIS trials such as E4112 and RTOG 9804 indicates that funding agencies, investigators, and study participants will be interested in the next generation of trials.

In summary, the management of DCIS will evolve from overtreatment for the average patient to optimization for an individual once several issues have been addressed: Will fewer cases of DCIS diagnosis compromise benefits from earlier detection of IBC? Could some patients with DCIS avoid surgery? Will advances in imaging help optimize surgery and allow patients to undergo one single procedure, be it breast conservation or mastectomy? Will tumor profiling optimally identify patients who could avoid radiation? Will improved shared decision-making strategies result in greater care satisfaction? Finally, can we identify biologic changes induced by adverse social environments as targets to reduce health disparities in care on long-term DCIS treatment outcomes? Extensive work performed since the 2009 NIH State of Science Conference directive has provided the groundwork of knowledge and technology necessary to guide a new generation of studies to address these key questions and generate evidence that will meaningfully affect clinical outcomes for an individual patient.^86,87 The time to resolve these issues for a more informed future is now.

AUTHOR CONTRIBUTIONS

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Overcoming Barriers in Ductal Carcinoma In Situ Management: From Overtreatment to Optimal Treatment

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).