Abstract P3-08-05: Validation of a nomogram for predicting recurrence among women with ductal carcinoma in situ and breast conserving surgery in an integrated health care system

Abstract

Abstract Current management of ductal carcinoma in situ (DCIS) most often entails removing the lesion by breast-conserving surgery. Quantifying the risk of a patient's ipsilateral breast event (IBE) recurrence, either invasive cancer or DCIS, after breast conserving surgery remains a clinical concern. The aim of this study was to validate the Memorial Sloan-Kettering Cancer Center (MSKCC) nomogram to predict IBE recurrence in patients from our institution. Patients and Methods We retrospectively identified 608 patients in the Kaiser Permanente Northwest integrated healthcare system with a diagnosis of DCIS who had undergone local excision from 1990 through 2007. We assessed the performance of the MSKCC nomogram for predicting IBE recurrence using measures of discrimination (how well risk scores separate those with and without the event) and calibration (agreement between predicted and observed risk). We calculated Harrell's C and R2D to provide estimates of discrimination. We calculated the calibration slope by performing a Cox regression using the prognostic index (PI; predicted log relative hazard based on the original Cox coefficients), with a slope not significantly different from 1 indicating no difference in discrimination from the development sample. We also examined discrimination by comparing the KM curves of 4 risk groups and in a Cox regression as a predictor, which were created using Cox's method (4 groups using the 16th, 50th, & 84th percentiles), as no risk groups were defined in the MSKCC development study. We examined whether there was model misfit by testing each predictor in a Cox regression with an offset of the PI. Finally, we assessed calibration by plotting the observed rates and associated 95% CIs against the predicted probabilities for groups based on 4 risk groups and octiles. Results The median follow-up time for the KPNW cohort was 125 months. The 10-year IBE recurrence rate was 9.5%, 95% CI [7.0%, 13.0%). Harrell's C was .70, which is comparable to what has been found in other validation studies. The PI accounted for 22% of variation in time (R2D =.22, 95% CI [.08, .38]). The test of the calibration slope provided no support that discrimination in this sample differs from the development study (LR χ2(1)=0.21, p=.65). An examination of the Kaplan-Meier curves among the risk groups showed good separation of the high risk group compared to the others, but little separation between the lowest two risk groups. None of the predictors demonstrated evidence for differential weighting from the MSKCC coefficients (p values ranged from .08 to .97). Calibration was good for the lowest, low, and moderate risk groups, but there was underprediction in the high risk group. When examining calibration using octiles, we found a similar pattern to that of other validation studies in which the highest octile had the furthest departure from perfect agreement. Conclusion The MSKCC nomogram for predicting IBE recurrence in patients with DCIS who were treated with local excision have some utility, and our results are consistent with other validation efforts. However, there is much potential to further increase the prediction of recurrence beyond what is possible with the MSKCC nomogram. Citation Format: Leo MC, Francisco M, Jenkins C, Weinmann S. Validation of a nomogram for predicting recurrence among women with ductal carcinoma in situ and breast conserving surgery in an integrated health care system [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P3-08-05.