De-Escalation of Locoregional Therapy in Low-Risk Disease for DCIS and Early-Stage Invasive Cancer

BACKGROUND

The widespread adoption of screening mammography in the United States has led to an associated increase in the incidence of both ductal carcinoma in situ (DCIS; stage 0 breast cancer) and early-stage invasive breast cancer. One study has estimated that in the United States alone, the incidence of early-stage breast cancer has more than doubled in the last three decades, with this increase almost exclusively observed in women of screening age, ≥ 40 years.¹ Outcomes following current treatment guidelines for stage 0 and I breast cancer are excellent; between 2009 and 2015, 62% of new breast cancer was localized to the breast at the time of diagnosis, and the 5-year mortality for these early-stage patients was 98.8%.²

Such improvements in outcome have been accompanied by important treatment-related sequelae; thus, the Breast Cancer Steering Committee of the National Cancer Institute, which provides guidance for prioritization and conduct for cooperative group trials in the United States, has declared “decreasing toxicity/treatment/costs associated with therapy with negligible clinically meaningful benefits” to be one of its highest priorities.^3(p1) The notion of de-escalation has been a challenging one for both patients and physicians. However, early breast cancer, with its attendant excellent outcomes, provides the ideal setting in which to de-escalate locoregional management and continues to be a proving ground for incorporation of biomarker-driven care. For both DCIS and early-stage invasive cancer, there are important ongoing opportunities to “right-size” both surgery and radiation to reduce treatment toxicities while better understanding the trade-offs of omitting some treatment modalities. While doing so, the highest priority will be to achieve excellent oncologic outcomes and high quality of life for these patients.

PUBLISHED RANDOMIZED TRIALS FOR DCIS

Five prospective, randomized clinical trials have evaluated the role of radiation treatment after lumpectomy for patients with DCIS.^4-11 In these randomized trials, radiation treatment was typically given to the whole breast, usually without a boost to the site of the primary tumor. On balance, these randomized trials have demonstrated that adding radiation treatment to the whole breast after surgical excision reduces the risk of local recurrence in the ipsilateral breast, as well as the subset of invasive local recurrence in the ipsilateral breast. Although these improvements in local recurrence with the addition of radiation treatment are large (typically by about half) and statistically significant, no associated improvement in breast cancer survival or overall survival (OS) has been found with the addition of radiation treatment after lumpectomy. However, the lack of a survival benefit should be interpreted cautiously because any potential survival benefit is an underpowered endpoint based on the number of patients in this setting.

The meta-analysis from the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) analyzed the first four reported randomized trials of radiation treatment after lumpectomy for 3,729 women.¹² The 10-year rate of local recurrence in the ipsilateral breast was reduced from 28.1% to 12.9% (P < .00001) by adding radiation treatment. Similarly, the 10-year rate of the subset of invasive local recurrence in the ipsilateral breast was reduced from 15.4% to 6.8% (P < .001). Despite these large differences, no differences were found for the 10-year rates of breast cancer mortality (3.7% v 4.1%, respectively; P > .1) or all-cause mortality (8.2% v 8.4%, respectively; P > .1).

The EBCTCG meta-analysis defined a potentially favorable subgroup of patients based on the combination of low nuclear grade, negative margins, and tumor size ≤ 2 cm. In this subset of 291 potentially favorable patients, the 10-year rate of local recurrence after lumpectomy was reduced from 30.1% without radiotherapy to 12.1% with radiotherapy (P = .002). After the EBCTCG meta-analysis, McCormick et al^8,9 published the results from a randomized clinical trial of radiation treatment after lumpectomy for DCIS prospectively defined as having good risk characteristics (unicentric disease, tumor size ≤ 2.5 cm, low or intermediate nuclear grade, and negative margins ≥ 3 mm). This study was the only randomized trial specifically restricted to DCIS with favorable, low-risk clinical characteristics. The trial was coordinated by NRG (National Surgical Adjuvant Breast and Bowel Project [NSABP], Radiation Therapy Oncology Group [RTOG], and Gynecologic Oncology Group [GOG]). In the most recent results from this study (NRG/RTOG 9804 [ClinicalTrials.gov identifier: NCT00003857]; formerly RTOG 9804), adding radiation treatment reduced the 12-year rates of local recurrence from 11.4% to 2.8% (hazard ratio [HR], 0.26; P = .0001) and the subset of invasive local recurrence from 5.8% to 1.5% (HR, 0.34; P = .016). However, no differences were found for OS or disease-free survival.

Randomized clinical trials have studied the value of adjuvant endocrine treatment after lumpectomy, as well as the role of aromatase inhibitors in the postmenopausal patient.^4,7,10,13-17 After lumpectomy plus radiation treatment in the NSABP B-24 study, patients were randomly assigned to receive adjuvant tamoxifen versus not receiving it.^7,10,13 Adding tamoxifen reduced the 10-year risk of any breast cancer event (defined as ipsilateral plus contralateral invasive carcinoma or DCIS) only for estrogen receptor (ER)-positive DCIS (HR, 0.49; P < .001), but not for ER-negative DCIS (HR, 0.84; P = .59). In the UK/ANZ (United Kingdom, Australia, and New Zealand) study, adding tamoxifen reduced all breast cancer events for the overall group of patients (HR, 0.71; P = .002) and for the subset of patients not receiving radiotherapy (P = .002).⁴ However, no similar benefit for tamoxifen was seen for the subset of patients receiving radiotherapy (P = .8).

OMISSION OF RADIATION TREATMENT AFTER BREAST-CONSERVING SURGERY FOR DCIS

Using Clinical and Pathologic Characteristics to Define Low-Risk DCIS and to De-Escalate Treatment After Lumpectomy

A number of arguments support de-escalation of treatment and omitting radiotherapy after lumpectomy for DCIS. First, no randomized trial has shown a benefit for survival or distant metastatic disease by adding either radiation treatment or adjuvant endocrine therapy after lumpectomy, although recognizing that randomized studies are underpowered for these endpoints. Second, the long-term breast cancer–specific mortality rates are similar and low whether patients are treated with mastectomy, lumpectomy, or lumpectomy plus radiation treatment.^18,19 Third, after an initial lumpectomy, only about half of the local recurrences in the ipsilateral breast are invasive carcinoma. Fourth, salvage treatment of local recurrence after an initial lumpectomy is often successful.

Many retrospective studies have attempted to individualize patient management and to omit radiation treatment after lumpectomy in low-risk DCIS as defined by clinical and pathologic characteristics.^20-30 However, the specific clinical and pathologic characteristics used to select for low-risk DCIS vary widely across studies, and no consensus has emerged. The most common characteristics used to define low-risk DCIS include (1) tumor size, (2) negative margins of resection, (3) grade, and (4) patient age. The lack of consensus is highlighted by the National Comprehensive Cancer Network, which does not include any specific definition of “low risk” for DCIS.²⁰

More recently, three clinical studies, including RTOG 9804, have attempted to define prospectively low-risk DCIS using clinical and pathologic characteristics (Table 1).^8,9,31-33 When restricted to the patients with grade 1-2 DCIS treated without radiation treatment after lumpectomy, the 10-year rates of local recurrence were 9.8%-15.6%, which appear to be less than historical controls. In addition, an increasing use of endocrine therapy across these three studies appears to be associated with a decreasing risk of local recurrence for patients treated with lumpectomy only, without radiation treatment.

TABLE 1.

Summary of Open and Enrolling Studies Prospectively Identifying Patients With Low-Risk DCIS for Treatment Using Breast-Conservation Surgery With or Without RT

Using Molecular Profiling to Define Low-Risk DCIS and to De-Escalate Treatment After Lumpectomy

Given the successful integration of multigene molecular profiling in the management of invasive breast carcinoma, similar efforts have been made to develop and implement multigene molecular profiling in the management of DCIS, especially with the goal of de-escalation by omitting radiation treatment after lumpectomy^34-40 De-escalation of therapy is further supported by the lack of improvement in survival by adding radiation treatment as well as by the small, but real, potential adverse effects of adding radiation treatment. Thus, molecular profiling has the potential to help individualize management, for example, by allowing for the individual patient with low-risk disease to weigh the relative risk-benefit of decreasing upfront treatment in exchange for a small increase in subsequent local recurrence.

The 12-gene Oncotype DX DCIS Score (also referred to as the DCIS Score) is a prognostic test that was developed to determine the 10-year risks of local recurrence and the subset of invasive local recurrence after initial lumpectomy without radiation treatment.^37,39 The DCIS Score consists of seven tumor genes and five reference genes. This molecular test was developed and validated for clinical use in accordance with the rigorous criteria for tumor biomarkers as specified by Simon et al.⁴¹ Importantly, two independent validation studies for the DCIS Score were performed, highlighting the reproducibility of this molecular test.^37,39

The first validation of the DCIS Score was performed using DCIS tumor specimens and clinical information from 327 patients enrolled in ECOG-ACRIN (Eastern Cooperative Oncology Group American College of Radiology Imaging Network) E5194.³⁹ In this validation study, the 10-year risks of local recurrence were 10.6% for the low-risk group, 26.7% for the intermediate-risk group, and 25.9% for the high-risk group (P = .006). The 10-year risks for the subset of invasive local recurrence were 3.7%, 12.3%, and 19.2%, respectively (P = .003). On multivariable analysis, the DCIS Score, tumor size, and menopausal status were statistically significant for local recurrence (all P ≤ .02). Similar findings were reported by Rakovitch et al³⁷ in a second, independent validation study using a population-based cohort of patients from Ontario, Canada.

The DCIS Score identifies a significant fraction of patients with underlying higher risk biology that is not seen using standard clinical and pathologic characteristics to select for lower risk biology. In the E5194 study, 30% (97/327) of the patients were found to have intermediate- or high-risk molecular profiling, and in the Ontario cohort, 38% (216/571).

The prognostic information from the DCIS Score has been further refined by combining the DCIS Score with the clinical and pathologic features of patient age and tumor size.³⁶ Using this combination of factors increases the number of lower-risk patients, thus increasing the number of patients eligible for omitting radiation treatment. For example, if one judges the threshold for eligibility for omitting radiation treatment to be ≤ 10% risk of local recurrence at 10 years, then 47% of patients meet this threshold, and if ≤ 8%, then 26% meet this threshold. Given that the risk of developing a contralateral breast cancer is approximately 7%-8% at 10 years, many patients have molecular profiling that indicates similar, or even lower, risk in the ipsilateral breast.

In an analysis of the interaction of the DCIS Score with radiation treatment, Rakovitch et al³⁸ reported that adding radiation treatment after lumpectomy reduces the 10-year rate of local recurrence by about half, regardless of the DCIS Score. For a lower risk DCIS Score, the 10-year rate of local recurrence was reduced from 10.6% without radiotherapy to 5.0% with radiotherapy, and for a higher risk DCIS Score, the respective 10-year rates were 25.4% without radiotherapy and 12.6% with radiotherapy.

The Decision Score is a second test developed for patients with DCIS.^34,40 The Decision Score is derived from seven cancer-related genes combined with four clinical and pathologic factors (patient age, tumor size, margin status, and palpability). The Decision Score is then dichotomized into two groups (low risk and elevated risk).

On multivariate analysis for the overall group of patients, the Decision Score, the use of radiation treatment, and year of diagnosis were statistically significant for local recurrence. The Decision Score and the use of radiation treatment were statistically significant for the subset of invasive local recurrence. When the Decision Score was separated into low-risk and high-risk groups, the use of radiation treatment remained statistically significant for the elevated-risk group for local recurrence and for the subset of invasive local recurrence, but not for the low-risk group for local recurrence or the subset of invasive local recurrence. Based on the published findings from these various molecularly based studies, a number of prospective studies have been designed and are currently open to enrollment (Table 2). The cost-effectiveness of molecular testing for DCIS in the context of omission of radiation has been evaluated, and has not been shown to offer a value advantage from a population perspective.⁴² Thus, the decision of when to use molecular diagnostics remains an individual choice, specifically benefiting those patients whose treatment could be altered by this information.

TABLE 2.

Summary of Open and Enrolling Prospective Studies Using Molecularly Based Approaches for DCIS of the Breast

OMISSION OF SURGERY FOR LOW-RISK DCIS

The treatment of DCIS has been predicated on the assumption that without treatment, all DCIS would progress to invasive cancer. The excellent prognosis of patients treated for DCIS with the current treatment approach has been well documented, with 10-year breast cancer–specific survival approaching 98%.¹⁹ However, there has been increasing recognition of the variable rate and likelihood of invasive progression among different DCIS lesions. A few case reports from single institutions as well as from the Nurses’ Health study indicated that up to 50% of DCIS may progress to invasive cancer over a 30-year period.^43,44 However, these reports were small, with limited data. A recent evaluation of data from the SEER registry of women diagnosed with DCIS from 1992-2014 who did not undergo surgery at the time of diagnosis indicates that the rate of invasive progression is highly variable between individuals and that the 10-year net risk of ipsilateral invasive breast cancer ranged from 15%-28%, depending on age at diagnosis and histologic features.⁴⁵

Moreover, there are some data to support that some DCIS may resolve without surgery, with endocrine therapy alone. CALGB 40903 (ClinicalTrials.gov identifier: NCT01439711) was a prospective study in which postmenopausal women diagnosed with ER-positive DCIS received treatment with 6 months of preoperative letrozole. Overall, over 80% of patients exhibited an imaging response to treatment (Fig 1). More notably, however, there was a 15% rate of pathologic complete response (pCR), including a 75% rate of pCR among patients with low-grade disease.⁴⁶ Such data support a more individualized approach for DCIS, which could be based on projected risk of invasive progression rather than merely the diagnosis of DCIS.

FIG 1.

Magnetic resonance imaging (MRI)–based response pattern of hormone receptor–positive ductal carcinoma in situ to preoperative letrozole therapy in CALGB 40903. Patients exhibited variable patterns of response, with almost one third of study patients demonstrating complete or near-complete imaging response to therapy by 6 months. Example of patient with near-complete response to treatment at (A) baseline, (B) 3-month, and (C) 6-month MRI timepoints.

Based on these observations, there has been growing interest in testing whether there exists a group of patients with DCIS with sufficiently low risk of invasive progression such that close surveillance, rather than surgery and/or radiation, could be reasonably considered. Since 2015, there has been activation of 3 prospective randomized trials in the United States, United Kingdom, and Europe, and one registry study in Japan for low-risk DCIS, all of which seek to determine whether an active surveillance strategy, with or without endocrine therapy, may be noninferior to usual care (Table 3).^47-50 For patients on active surveillance, surgery is only undertaken if there is progression of disease. The definition of “low-risk” disease differs slightly among studies, but all studies exclude high-grade DCIS, and both the COMET (Clinical Trials.gov identifier: NCT02926911) and LORETTA trials (UMIN000028298), which include the use of endocrine therapy, require that DCIS be ER and/or progesterone receptor (PR) positive.

TABLE 3.

Summary of Open and Enrolling International Prospective Trials for Omission of Surgery in Low-Risk DCIS

The primary endpoint of these studies is either time to progression to invasive cancer or 2-, 5-, and 10-year rate of invasive progression, and the overall study designs are sufficiently similar to allow for informative meta-analyses. However, additional meaningful endpoints, such as quality of life, mastectomy-free survival, and OS, will also be evaluated to assess the quality-of-life trade-offs between these two approaches. Biospecimen collection is integrated into these trials, and it is hoped that these efforts will permit discovery of novel markers of invasive progression for DCIS.

BREAST-CONSERVING SURGERY WITHOUT RADIATION TREATMENT OF EARLY-STAGE INVASIVE BREAST CANCER

In the setting of breast-conservation treatment of early-stage invasive breast carcinoma, omitting radiotherapy for the lowest-risk patient subgroups has been an active area of clinical investigation. The search for these patient subgroups is supported by the decreasing rates of local recurrence in more contemporary groups of patients who received breast-conservation surgery. Early studies focused on using older patient age to select for low-risk subgroups for omitting radiation treatment. More recent and ongoing studies have focused on using biologic approaches to select for low-risk patient subgroups (Table 4).

TABLE 4.

Summary of Published and Enrolling Prospective De-Escalation Studies in Low-Risk Patients With Early-Stage Invasive Breast Carcinoma Treated With Breast-Conservation Surgery and Endocrine Therapy, With or Without RT

The approach to omitting radiation treatment in low-risk subgroups is based on the assumption that even though adding radiation treatment results in a small improvement in local control, there is no demonstrated improvement in other endpoints, such as OS, breast cancer–specific survival, and freedom from distant metastatic disease. However, because prospective studies, including randomized trials, used local failure as the primary study endpoint to design the respective study sample sizes, these studies were substantially underpowered for other endpoints. In the EBCTCG meta-analyses of randomized trials of radiation treatment after breast-conservation surgery for early-stage invasive breast carcinoma, four local recurrences in the breast were directly associated with one avoidable breast cancer death.^51,52 Thus, even in low-risk subgroups, omitting radiation treatment is potentially associated with a small increase (eg, 1%-2%) in breast cancer deaths, which is too small to be evaluated with the numbers of patients reported to date.

The rate of local recurrence has been shown to be lower in older patients. Based on this observation, a number of randomized clinical trials as well as SEER data have evaluated the role of omitting radiation treatment after lumpectomy using patient age as a key eligibility factor, typically for patients ≥ 65 or ≥ 70 years of age.^53-58 These randomized trials demonstrated a small, but statistically significant, improvement in the 5-year and 10-year rates of local recurrence associated with adding radiation treatment, but without any demonstrable impact on breast cancer–specific mortality or OS. The study with the longest reported follow-up, CALGB 9343, found no significant differences in time to mastectomy, time to distant metastasis, breast cancer–specific survival, or OS between patients randomly assigned to tamoxifen alone or tamoxifen and radiation therapy at a median of follow-up of 12.6 years.⁵⁴ Importantly, when individualizing treatment options for the older patient, comorbidity and life expectancy are also key considerations.

Retrospective studies have demonstrated low rates of 10-year local recurrence (eg, ≤ 5%) after breast-conservation treatment with low-risk molecular profiling.^59-65 Based on these observations, as well as the generally decreasing risk of local recurrence in contemporary studies, a number of prospective studies based on molecular profiling have been designed with the goal of potentially eliminating radiation treatment of the lowest-risk patient subgroups (Table 4). One limitation of these studies is that only one was a randomized study to evaluate the potential improvement in local recurrence associated with adding radiation treatment, even in these low-risk subgroups. Another limitation is that these studies are underpowered for events other than local control (eg, breast cancer mortality or distant metastatic disease). Regardless, these studies will begin to identify potential patient subgroups for de-escalation of radiation in the setting of breast-conservation treatment.

SELECTIVE DE-ESCALATION OF AXILLARY SURGERY FOR EARLY-STAGE INVASIVE BREAST CANCER

De-escalation of locoregional treatment of early-stage invasive cancer has also included consideration of omission of sentinel node biopsy in select individuals. Despite the low morbidity of sentinel lymph node biopsy (SLNB) compared with axillary node dissection (ALND), prospective trials have nevertheless reported up to a 10% risk of neuropathy, lymphedema, and limited shoulder mobility after SLNB, all of which are more pronounced among older women.^66,67 Thus, efforts have focused on reduction of axillary surgery in those patients who are either at low risk for axillary metastases or in those for whom this information would not likely alter treatment recommendations. Although some have argued that SLNB lacks benefit in women who present with a clinically negative axilla and for whom chemotherapy would clearly be recommended (eg, T1 N0 human epidermal growth factor receptor 2–positive breast cancer), omission of SLNB has primarily been studied in women ≥ 70 years, including those with small ER-negative tumors for which chemotherapy might not be advised, even in the setting of low-volume nodal disease.

Evidence for safe omission of surgical staging of the axilla was provided by the CALGB 9343 study, a prospective trial of women ≥ 70 years of age who underwent breast-conserving surgery for stage I ER-positive disease and who were then randomly assigned to either tamoxifen and radiation or tamoxifen alone. Of interest, over 60% of women in the study did not undergo any axillary surgery. Of the 244 women who had ALND, there were no axillary recurrences, but importantly, there were also no axillary recurrences among women who had tamoxifen and radiation and only 6/200 axillary recurrences among those who received tamoxifen alone without radiation. At a median follow-up of 12.6 years, there was no difference in breast cancer survival or OS between groups.⁵⁴

Two additional prospective randomized trials have also addressed outcomes after omission of axillary surgery among older women with early-stage breast cancer.^68,69 Similar to CALGB 9343, both studies found a small increase in axillary recurrence in patients who did not have axillary surgery compared with those who underwent axillary surgery. In an analysis that combined the data from the two studies, there were no differences in in-breast recurrence, breast cancer–specific survival, or OS.

Among women ≥ 70 years of age in the National Cancer Database diagnosed with stage I ER-positive breast cancer between 2004 and 2010, 89% of patients had some type of axillary surgery, including 95% of patients who had mastectomy. However, even among node-positive patients, only 23% of patients 76-79 years of age and 11% of women 80-85 years of age received adjuvant chemotherapy.⁷⁰

In 2016, the Society of Surgical Oncology published its “Choosing Wisely” guidelines, which included a recommendation to consider omission of SLNB in clinically node-negative women > 70 years of age with hormone receptor–positive invasive breast cancer.⁷¹ The European Society of Breast Cancer Specialists had issued similar guidance in 2012, stating that although SLNB should be considered routine in healthy older women, omission of SLNB and completion ALND might be reasonable in some older patients whose treatment recommendations and outcomes were not likely to be affected by axillary staging.⁷²

Currently, there is only 1 active single-institution trial at Cedars Sinai Hospital (ClinicalTrials.gov identifier: NCT02564848), a single-arm registry trial studying the safety of SLNB omission in women ≥ 65 years of age with clinically node-negative T1-2 ER-positive breast cancer. There are several other trials in development using biomarkers to identify women at lowest risk. These efforts to de-escalate axillary surgery in selected patients will require multidisciplinary teamwork, including geriatric oncologists who can help provide a careful definition of physiologic age and life expectancy, to identify those patients who benefit least from axillary surgery.

In conclusion, the ultimate goal of personalized medicine is to right size therapies by using clinical criteria, emerging biomarkers, and medical evidence. Achievement of this objective will require both more and less: greater use of biologically targeted therapies as well as identifying opportunities to scale back those treatments that do not offer meaningful clinical benefit. Conceptually, the latter is more challenging than the former, because omission of certain treatments may be mistakenly confounded with either negligence or increased risk. Thus, future efforts focused on selective de-escalation of treatments will pose unique challenges and will require new tools, including innovative approaches to noninferiority trial design, thoughtful definition of more meaningful clinical endpoints (including addressing the fundamental question of how to define “clinically meaningful”), rational biomarker development, and more effective approaches to both eliciting and determining patient preferences for care. These efforts will require more creative engagement and communication with patients to realize the most patient-centered and equitable outcomes for patients with early-stage breast cancer.

AUTHOR CONTRIBUTIONS

Conception and design: E. Shelley Hwang

Collection and assembly of data: E. Shelley Hwang

Data analysis and interpretation: E. Shelley Hwang

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

De-Escalation of Locoregional Therapy in Low-Risk Disease for DCIS and Early-Stage Invasive Cancer

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).

Lawrence Solin

Consulting or Advisory Role: Genomic Health

Speakers' Bureau: Genomic Health

No other potential conflicts of interest were reported.