38 Comparing Risk Stratification and Radiotherapy Benefit for Patients With DCIS Using a 7-Gene Biosignature as Compared to a Clinicopathologic Nomogram

Background

Multiple prospective trials show over 70% of women with ductal carcinoma in situ (DCIS) do not recur after breast-conserving surgery (BCS) alone and radiotherapy (RT) had a clinically meaningful reduction in 10-year risk after BCS. Clinicopathologic (CP) factor–based criteria or nomograms have been used to select for whom to deescalate treatment, assessing the 10-year ipsilateral breast recurrence (IBR) rate after BCS. In this study, we compared the classification of women by a CP model similar to the MSKCC DCIS nomogram with the 7-gene biosignature and their associated total 10-year IBR rates with and without RT.

Materials and Methods

DCIS patients (n = 926) from 4 international cohorts treated with BCS (negative margins) with (n = 641) and without RT (n = 335) were evaluated for clinicopathological models and the 7-gene biosignature with residual risk subtype (RRt). Formalin-fixed, paraffin-embedded tissue samples for each patient were analyzed at a CLIA lab for the 7-gene biosignature, DCISionRT with RRt. The biosignature reported a decision score (DS) of 0-10 and the presence/absence of the RRt subtype. The CP model used MSKCC DCIS nomogram factors and weighting but omitted the number of excisions and close margin and was categorized using a score <220 as CP low risk. The MSKCC-like score was calculated using the equation score = 100*(RT = 0) + 76*(ET = 0) + FAMHX(pos)*30 + Detection(clinical)*34 + (Grade ≥ 2)*26 + Necrosis (Present)*12 + Year of Diagnosis (≤1998)*58 + (134-1.49*AGE); margin (no-ink on tumor) = 0, number of excisions ≥3 (incomplete) = 0, evaluate with RT = 0). 10-year total IBR rates were evaluated using Cox proportional hazards and Kaplan-Meier analysis by treatment and risk groups in patients treated with and without RT.

Results

The biosignature classified 37% of women treated with BCS as low risk (n = 338) and 63% (n = 588) were classified into the combined Elevated/Residual Risk group. The 10-year IBR rate for patients who did not receive RT in the low-risk group was 5.6% (95% CI, 2.5%-12.1%). RT did not reduce the IBR rate in the low-risk group (P = .71), where the 10-year IBR rate was 4.8% (95% CI, 2.5%-9.1%) after RT. The CP model classified 320 (34%) as CP low risk (score <220) with an 8.3% 10-year IBR rate for BCS without RT and a non-significant slightly lower rate with RT (4.7%; HR, 0.5; 95% CI, 0.2-1.2, P = .12). Eighty-seven percent (n = 277) of the MSKCC low-risk group patients were treated with endocrine therapy (ET), whereas 32% (n = 124) of the biosignature low-risk patients were treated with ET. Thirty-seven percent (n = 124) of the biosignature low-risk and 30% (n = 98) of the MSKCC low-risk patients were not treated with RT. The biosignature reclassified 63% of these CP low-risk patients as DS elevated risk, with 10-year IBR rates of 12.3% for BCS without RT and 4.8% with RT (HR, 0.2; 95% CI, 0.1-0.8, P <.01).

Conclusions

Using 10-year outcome data (n = 926) from multiple contemporary published validation studies, the biosignature reclassified nearly two-thirds of patients in the CP low-risk (MSKCC-like nomogram) group as elevated risk. The re-classified CP low-risk patients had elevated 10-year IBR rates after BCS without RT and had an 80% relative benefit from RT, demonstrating they potentially could have been undertreated. This demonstrates the ability of the 7-gene biosignature to better risk-stratify patients with DCIS following BCS and identify patients with low-risk CP who may benefit from RT.

AFFILIATIONS:

Julie Margenthaler,¹ Karuna Mittal,² Frank Vicini,³ Chirag Shah,⁴ Rachel Rabinovitch,⁵ Pat Whitworth,⁶ Brian Czerniecki,⁷ David Dabbs,² G. Bruce Mann,⁸ Fredrik Wärnberg,⁹ Sheila Weinmann,¹⁰ Michael Leo,¹⁰ Steven C. Shivers,² Troy Bremer²

¹Washington University School of Medicine, St. Louis, MO.

²PreludeDx, Laguna Hills, CA.

³GenesisCare, Farmington Hills, MI.

⁴Cleveland Clinic, Cleveland, OH.

⁵University of Colorado, Colorado Springs, CO.

⁶Nashville Breast Center, Nashville, TN.

⁷Moffitt Cancer Center, Tampa, FL.

⁸University of Melbourne, Melbourne, Australia.

⁹University of Gothenburg, Gothenburg, Sweden.

¹⁰Kaiser Permanente Center for Health Research, Portland, OR.