Development and Clinical Validation

The Oncotype DX® Breast DCIS Score result was developed based upon analyses of multiple correlative science studies comparing gene expression profiles between invasive and DCIS tumor samples. The selected genes included in the Breast DCIS Score algorithm include five proliferation genes, progesterone receptor (PR), GSTM1 and five reference genes. The algorithm was developed using scaling and category cut-points based on the analysis of the Breast DCIS Score result in a separate cohort of DCIS patients.1

The E5194 trial was a prospective cooperative group trial that enrolled a total of 670 patients.2 The primary objective was to evaluate 5- and 10-year actuarial ipsilateral breast event (IBE) rates after local excision alone in a selected population of patients with DCIS. Key eligibility criteria included patients with a minimum 3 mm negative margin and either tumors of low or intermediate grade that were ≤ 2.5 cm in size or high grade tumors that were ≤ 1 cm. Radiation treatment was not allowed and tamoxifen therapy was permitted at the discretion of clinicians beginning in 2000. Results from this trial have been previously published by Hughes and colleagues.

Initial validation of the Breast DCIS Score result was performed in a prospectively designed study of archived tumor specimens from 327 patients who participated in the previously described E5194 trial.3 All aspects of the assay, including pre-analytic and analytic methods, gene coefficients, and risk group categories, were pre-specified prior to initiation of sample processing and clinical validation. Central pathology review was performed by breast pathologists using the College of American Pathologists DCIS guidelines. The primary objective of the study was to determine if the Breast DCIS Score result was significantly associated with the risk of an IBE, defined as local recurrence of DCIS or invasive carcinoma. The study’s secondary objective was to determine whether the Breast DCIS Score result provided prognostic information beyond standard clinical and pathologic measures. Clinical demographics and tumor characteristics of patients enrolled in the validation study are described below. Overall the patient population in the validation study was similar to that of the E5194 trial, except for median tumor size, which was slightly larger in the validation study cohort (Table 4).

Table 4: Study Patient Characteristics (N=327)
Patient age 61 years (Median)
Post-menopausal 248 (76%)
Tumor size 7 mm (Median)
Tumor size < 10 mm 260 (79%)
Negative margins > 5 mm 214 (65%)
Tamoxifen use 96 (29%)
ER positive (RT-PCR) 318 (97%)
Study cohort: Cohort 1 273 (83%)

Cohort 2 54 (17%)
*Similar to parent trial for all variables except for tumor size
Cohort 1 low or intermediate grade group and Cohort 2 high grade group

The study met its primary objective as the Breast DCIS Score result was predictive of the 10-year risk of any IBE. The Breast DCIS Score result as a continuous variable was significantly associated with developing an IBE (HR = 2.38, 95% CI = 1.19 – 4.60; p = 0.01). In the primary analysis adjustment for tamoxifen as a covariate did not significantly alter the study’s results (HR = 2.31, 95% CI = 1.15 – 4.49; p = 0.02). The Kaplan-Meier curves below (Figure1) shows the estimated risk of any IBE and invasive IBE over a 10-year time period according to the three pre-specified risk groups (low < 39, intermediate 39-54, and high ≥ 55).

Figure 1: 10-Year Risk of IBE (A) and Invasive IBE (B) by Breast DCIS Score Groups
Ten Year Risk DCIS

DCIS Ten Year Risk

The 10-year risk of any IBE (DCIS or invasive carcinoma) was 10.6% in the low risk group compared to 25.9% in the high risk group (p=0.006). The risk for developing ipsilateral invasive carcinoma only was 3.7 % in the low risk group compared to 19.2% in the high risk group (p=0.003). Approximately 70% of all patients enrolled in the study were in the low risk group. In multivariate analyses, the DCIS Score result, tumor size, and menopausal status were identified to be statistically significant predictors of the risk of local recurrence (p ≤ 0.02). The hazard ratio for the score remained unchanged after adjusting for tumor size and menopausal status thereby demonstrating that the Breast DCIS Score result provides independent prognostic information beyond these other risk factors.

The second prospectively designed clinical validation study of the Oncotype DX Breast DCIS Score assay was conducted in a contemporary population-based cohort of 571 patients diagnosed with DCIS in Ontario, Canada from 1994-2003.4 DCIS diagnosis was reconfirmed by central breast pathology review and study endpoints, methods and analysis plan were all predefined. The primary objective was to evaluate if the Breast DCIS Score is an independent predictor of the risk of local recurrence (DCIS or invasive) and secondary endpoint was to determine if the Breast DCIS Score result is independently associated with local recurrence after adjusting for significant clinical and pathologic variables. Inclusion criteria included patients with pure DCIS who underwent breast-conserving surgery only and had negative surgical resection margins. Median follow-up was 9.6 years and patient demographics and tumor characteristics are described below in Table 5.

Table 5: Study Patient Characteristics (N=571)
Characteristic Number
Age = 50 years 459 (81%)
Nuclear Grade
Low 55 (10%)
Intermediate 332 (58%)
High 184 (32%)
Comedo Necrosis 350 (61%)
Solid Subtype 358 (63%)
Tumor size
= 10 mm 150 (26%)
> 10 mm 140 (25%)
Missing 281 (49%)
Multifocality 114 (20%)
ER+ by RT-PCR 541 (95%)
HER2+ by RT-PCR 100 (17.5%)
Presence of at least 2 foci of DCIS in the same quadrant at least 5mm apart

The study found the Breast DCIS Score result to independently predict and quantify local recurrence risk. In the primary analysis, the Breast DCIS Score result was significantly associated with any local recurrence in estrogen receptor positive patients (HR = 2.26, 95% CI = 1.41-3.59; p < 0.001) as well as all patients regardless of estrogen receptor status (HR = 2.15; 95% CI = 1.43-3.22; p < 0.001). Figure 2 below shows the 10-year estimated risk of local recurrence of invasive carcinoma and DCIS for the pre-specified risk groups (low < 39, intermediate 39-54, and high ≥ 55).

Figure 2: 10-Year Risk of an Invasive (A) and DCIS (B) Recurrence by Breast DCIS Score Groups

10 Year DCIS Risk

The 10-year risk of a local invasive carcinoma recurrence was 8.0% in the low risk group compared to 20.9% and 15.5% in the intermediate and high risk groups respectively (p = 0.03). The risk of developing a local recurrence of DCIS was 5.4% in the low risk group compared with 14.1% and 13.7% in the intermediate and high risk groups, respectively (p = 0.002). Approximately 62% of all patients studied were in the low risk group. In multivariable analysis the DCIS Score result was a significant predictor of local recurrence (HR = 1.68, 95% CI = 1.08-2.62; p = 0.02) and provided independent recurrence risk information beyond clinical and pathologic measures including age at diagnosis, tumor size, tumor subtype, and multifocality (p = 0.02). The investigators concluded that this study further validates the Breast DCIS Score result as an independent predictor of 10-year risk of local recurrence among women with DCIS treated with breast-conserving surgery. This study’s results are consistent with the initial E5194 validation study thereby reconfirming those findings in a contemporary real-world cohort of DCIS patients.




1. Baehner FL, Yoshizawa CN, Butler SM, et al. The development of the DCIS Score: scaling and normalization in the Marin Medical Laboratories cohort. Poster presented: ASCO Breast Cancer Symposium; 2012; San Francisco, CA.
2. Hughes LL, Wang M, Page DL, et al. Local excision alone without irradiation for ductal carcinoma in situ of the breast: a trial of the Eastern Cooperative Oncology Group. J Clin Oncol. 2009;27(32):5319-5324.
3. Solin L, Gray R, Baehner FL, et al. A multigene expression assay to predict local recurrence risk for ductal carcinoma in situ of the breast. J Natl Cancer Inst. 2013;105(10):
4. Rakovitch E, Nofech-Mozes S, Hanna W, et al. A population-based validation of the DCIS score predicting recurrence risk in individuals treated by breast-conserving surgery. Oral presentation: San Antonio Breast Cancer Symposium; 2014; San Antonio, TX.


Next: Clinical Impact and Health Economic Studies