21-Gene Assay and Breast Cancer Mortality in Ductal Carcinoma in Situ

Eileen Rakovitch, MD, MSc, FRCPC; Rinku Sutradhar, PhD; Sharon Nofech-Mozes, MD, FRCPC; Sumei Gu, MSc; Cindy Fong, BSc, CCRP; Wedad Hanna, MD, FRCPC; Lawrence Paszat, MD, MSc, FRCPC

Disclosures

J Natl Cancer Inst. 2021;113(5):572-579. 

In This Article

Discussion

Past studies have shown that the DCIS score expression assay improves the prediction of LR risk after BCS for DCIS, but it does not predict breast cancer mortality.[13–15] This is not surprising, because the 12-gene DS assay is predicated on a gene expression signature related to cellular proliferation. In contrast, the 21-gene RS assay includes additional genes implicated in the mechanisms of progression to invasion, cellular adhesion, and metastases. The presence of a high RS in hormone-receptor positive, HER2/neu negative invasive breast cancers is associated with an increased risk of distant metastases, worse survival, and chemotherapy benefit.[18,19] To our knowledge, this is the first study demonstrating that a high RS (defined a priori as >25) is also associated with a higher risk of breast cancer mortality in patients with pure DCIS treated with BCS. The 21-gene RS, in combination with clinical risk factors (age at diagnosis), can improve the risk stratification of DCIS by discriminating the subset of patients with biologically aggressive, potentially fatal disease, for whom breast RT may confer a survival benefit by reducing the risk of invasive LR.

One-third of DCIS lesions had high RS expression, including 25% of those with low-risk clinical features, demonstrating that putative molecular alterations are expressed in a clinically significant proportion of DCIS lesions irrespective of the presence of adverse clinicopathological features.

Our findings corroborate past studies that, overall, the risk of breast cancer mortality after breast-conserving therapy for all subsets of patients diagnosed with DCIS is low.[2] However, these studies were based on clinical and pathological factors without genomic profiling. In this analysis, we found that the RS can identify a subset of patients with DCIS who are at higher risk of invasive LR and a higher risk of breast cancer mortality; however, the impact of the RS on the risk of breast cancer mortality is not uniform across all patients. For women treated by BCS alone, the presence of a high RS is associated with an increased risk of invasive LR and an increased risk of breast cancer mortality, especially in women aged 50 years or younger at diagnosis who have less competing comorbidities. Younger women with a high RS who were treated by BCS alone had the highest risk of invasive LR and an 11-fold increased risk of breast cancer mortality, culminating in a 20-year risk of breast cancer mortality of 9.4%. This is the only subgroup for which the risk of death from breast cancer exceeded the risk of death from other causes (in fact, there were no deaths from other causes). Furthermore, for individuals with a high RS, treatment with RT was associated with a 71% reduction in the 20-year risk of breast cancer mortality; the absolute risk reduction was similar in women aged 50 years or younger (5.4%) or older than 50 years (4.4%) at diagnosis, however, the competing risk of death from other causes was much higher in older women. For women with a low RS, the risk of breast cancer mortality was low with no difference between women treated with or without RT.

A high RS was also associated with a small increased risk of DCIS LR, which can be explained by the presence of the proliferative signature within the RS and was not associated with a higher risk of contralateral breast cancer (Supplementary Figures 2 and 3, available online). This suggests that the increased risk of breast cancer mortality with a high RS is mediated by a higher risk of an ipsilateral invasive LR and that breast RT mitigates this risk by reducing the risk of invasive LR.

Our findings do not imply a lack of benefit from breast RT for individuals with a low RS [because RT reduces the risk of LR, invasive LR, and mastectomy[2,25,26]], nor does it imply a lack of utility of existing clinical assays that predict LR risk.[13,14,16] Population-based analyses report that one-third of women (including those aged 50 years or younger) do not receive breast RT after BCS,[11,26] and even fewer women receive endocrine therapy.[15,26] Although additional research is needed to determine which assay most accurately predicts the risks of LR and invasive LR, the 21-gene RS is the only assay shown to predict breast cancer mortality (Supplementary Figure 4, available online). This additional information can help more effectively target treatments to those at greatest risk and more effectively identify appropriate candidates for de-escalation strategies.

Inherent in population-based studies, treatment with RT was not randomized; therefore, we implemented a propensity score-adjusted multivariable analysis using a competing risk model to calculate cumulative incidence functions and provide accurate estimates accounting for the competing risk of death from other causes (Inverse Probability of Treatment Weighting model yielded similar findings). Individuals in the study cohort were diagnosed with DCIS from 1993 to 2003, but the pathological covariates used in this analysis were ascertained from a systematic pathology review performed in 2012–2015; a representative tissue block from the index DCIS lesion was used to obtain the DS and the RS in 2014–2015. As such, the findings are relevant to individuals diagnosed with DCIS in the current era. Tamoxifen use was limited (<15%);[14] therefore, the data reported reflect outcomes in the absence of endocrine therapy.

Baseline assessment of hormone receptor (estrogen or progesterone receptor) and HER2/neu status was not available, but we were able to ascertain receptor status data from the RS (reverse transcription polymerase chain reaction) assessment. Overall, 1255 (92.1%) of DCIS lesions were estrogen-receptor positive, 19.8% were HER2 positive, and 2% were triple (hormone receptor and HER2) negative [corroborating prior reports[27,28]]. There was no difference in the 10-year risk of breast cancer mortality by biomarker subgroup (Supplementary Table 1, available online). This suggests that the RS provides additional prognostic information beyond estrogen receptor, progesterone receptor, and HER2 expression.

In conclusion, the 21-gene RS provides additional prognostic information beyond LR risk estimates provided by current assays, which can improve risk stratification and management of DCIS. The presence of a high RS (>25) identifies the subset of patients with DCIS who have a higher risk of invasive LR and subsequent breast cancer mortality. For women with a high RS (especially those aged 50 years or younger or those with no comorbidity), treatment with breast RT (and possibly tamoxifen) may reduce the risk of breast cancer mortality by reducing the risk of invasive LR. Conversely, for women with a low RS, the risk of death from breast cancer is low, and the predominant risk of death is because of other (nonbreast cancer related) causes; for these women, de-escalation efforts should optimally balance patients' preference to lower local recurrence risk and optimize breast preservation with the costs and adverse effects of treatment.

processing....