External Validation of the Breast Cancer Surgery Risk Calculator (BCSRc): A Predictive Model for Postoperative Complications.

BACKGROUND: The breast cancer surgical risk calculator (BCSRc) is a prognostic tool that determines a breast cancer patient's unique risk of acute complications following each possible surgical intervention. When used in the preoperative setting, it can help to stratify patients with an increased complication risk and enhance the patient-physician informed decision-making process. The objective of this study was to externally validate the four models used in the BCSRc on a large cohort of patients who underwent breast cancer surgery. METHODS: The BCSRc was developed by using a retrospective cohort from the National Surgical Quality Improvement Program database from 2005 to 2018. Four models were built by using logistic regression methods to predict the following composite outcomes: overall, infectious, hematologic, and internal organ complications. This study obtained a new cohort of patients from the National Surgical Quality Improvement Program by utilizing participant user files from 2019 to 2020. The area under the curve, brier score, and Hosmer-Lemeshow goodness of fit test measured model performance, accuracy, and calibration, respectively. RESULTS: A total of 192,095 patients met inclusion criteria in the development of the BCSRc, and the validation cohort included 60,144 women. The area under the curve during external validation for each model was approximately 0.70. Accuracy, or Brier scores, were all between 0.04 and 0.003. Model calibration using the Hosmer-Lemeshow statistic found all p-values > 0.05. All of these model coefficients will be updated on the web-based BCSRc platform: www.breastcalc.org . CONCLUSIONS: The BCSRc continues to show excellent external-validation measures. Collectively, this prognostic tool can enhance the decision-making process, help stratify patients with an increased complication risk, and improve expectant management.

Surgical Predictive Model for Breast Cancer Patients Assessing Acute Postoperative Complications: The Breast Cancer Surgery Risk Calculator.

BACKGROUND: Prognostic tools, such as risk calculators, improve the patient-physician informed decision-making process. These tools are limited for breast cancer patients when assessing surgical complication risk preoperatively. OBJECTIVE: In this study, we aimed to assess predictors associated with acute postoperative complications for breast cancer patients and then develop a predictive model that calculates a complication probability using patient risk factors. METHODS: We performed a retrospective cohort study using the National Surgical Quality Improvement Program (NSQIP) database from 2005 to 2017. Women diagnosed with ductal carcinoma in situ or invasive breast cancer who underwent either breast conservation or mastectomy procedures were included in this predictive modeling scheme. Four models were built using logistic regression methods to predict the following composite outcomes: overall, infectious, hematologic, and internal organ complications. Model performance, accuracy and calibration measures during internal/external validation included area under the curve, Brier score, and Hosmer-Lemeshow statistic, respectively. RESULTS: A total of 163,613 women met the inclusion criteria. The area under the curve for each model was as follows: overall, 0.70; infectious, 0.67; hematologic, 0.84; and internal organ, 0.74. Brier scores were all between 0.04 and 0.003. Model calibration using the Hosmer-Lemeshow statistic found all p-values to be > 0.05. Using model coefficients, individualized risk can be calculated on the web-based Breast Cancer Surgery Risk Calculator (BCSRc) platform ( www.breastcalc.org ). CONCLUSION: We developed an internally and externally validated risk calculator that estimates a breast cancer patient's unique risk of acute complications following each surgical intervention. Preoperative use of the BCSRc can potentially help stratify patients with an increased complication risk and improve expectations during the decision-making process.

Molecular detection of micrometastatic breast cancer in histopathology-negative axillary lymph nodes fails to predict breast cancer recurrence: a final analysis of a prospective multi-institutional cohort study.

BACKGROUND: To address the clinical relevance of molecular detection of occult breast cancer in sentinel lymph nodes and nonsentinel axillary lymph nodes (ALN), we initiated the Minimally Invasive Molecular Staging of Breast Cancer (MIMS) trial, a multi-institutional prospective cohort study. This trial represents the first prospective cohort study in which a multimarker, real-time reverse transcription polymerase chain reaction (RT-PCR) analysis was applied to the detection of breast cancer micrometastases in ALN. MATERIALS AND METHODS: Sentinel and/or nonsentinel ALN from 501 breast cancer subjects with T1-T3 primary tumors were analyzed by standard histopathology and multimarker, real-time RT-PCR analysis. Seven breast cancer-associated genes (mam, mamB, PIP, CK19, muc1, PSE, and CEA) known to be overexpressed in metastatic breast cancer compared with control lymph nodes were used. Follow-up data were collected for 5 years. RESULTS: Of the 501 breast cancer subjects enrolled, 348 were node negative and completed the 5-year follow-up. Of these patients (n = 94), 27% demonstrated evidence of molecular overexpression. The 5-year relapse-free survival rate was 95.4% (95% confidence interval [95% CI], 92.4-97.2%). No single gene or combination of study genes was predictive of recurrence. CONCLUSIONS: The genes in this study panel failed to be predictive of clinical relapse. This may be a function of several factors: the low event rate at 5 years, the particular gene set, the methodology used for detection/analysis or that our original hypothesis was wrong and that the presence of positive marker signal by real-time RT-PCR is not associated with a worsened clinical outcome.