Does Intraoperative Radiation Therapy Result in Improved Overall Survival in Early-Stage Breast Cancer? An Analysis of the National Cancer Database

Abstract

The TARGIT-A trial of kilovoltage intraoperative radiation therapy (IORT) vs. external beam radiation therapy (EBRT) demonstrated a significant reduction in non-breast cancer mortality (nBCM) with a trend towards improved overall survival (OS) in women that received IORT. The trend towards improved OS was attributed to an increase in cardiac mortality in EBRT patients. We set to determine whether EBRT results in inferior OS compared to IORT in women with low-risk, early-stage breast cancer (BC) in a United States database with a hypothesis that the two approaches should result in equivalent OS. We used the National Cancer Database (NCDB) to identify women with low-risk BC treated with lumpectomy and either IORT or EBRT from 2004-2016. Low-risk BC included: age≥50 years, stage 0-I (ductal histology), lymph node-negative, ER+. Patients were categorized into 1 of 2 groups: IORT (20 Gy/1 fraction and no further EBRT) or EBRT (4256-6000 cGy to the breast [no regional lymph node irradiation]). The primary endpoint was OS. We compared OS between the IORT and EBRT patients using the Kaplan-Meier method and Cox-regression multivariate analysis (p<.05 considered significant). We used propensity-score matching (PSM) as an independent test of the Cox regression analysis. We identified 81,889 women that met the inclusion criteria (1,321 IORT; 80,568 EBRT). Compared to EBRT patients, IORT patients were older (mean age = 68.3 years vs. 64.8 years, p<.0001), had a similar proportion of patients with no comorbidities (83.1% vs. 84.7%, p = .13), and were less likely to have DCIS (11.3% vs. 27.5%, p<.0001), grade 3 tumors (9.9% vs. 16.8%, p<.0001), or PR- disease (9.3% vs. 14.7%, p<.0001). IORT patients were more often treated at academic medical centers (53.8% vs. 31.4%, p<.0001). Median follow-up was 44.9 months (IQR 23.7 months-75.9 months). The 4-year OS for IORT vs. EBRT was 94.9% vs. 95.7% (HR = 1.24, 95% CI 0.93-1.64, p = .15). Only tumor laterality and receipt of lymph node surgery were not associated with OS on univariate analysis. After adjustment for all other potential confounders (age, comorbidity, race, facility type, insurance status, income level, education level, stage, tumor size, grade, PR status, hormone therapy receipt, and chemotherapy receipt) the HR for OS from IORT vs. EBRT was 0.90 (95% CI 0.67-1.21, p = .48). The PSM cohort included 1320 of a possible 1321 matched pairs. The 4-year OS for IORT vs. EBRT was 94.9% vs. 95.9% with a HR = 1.13 (0.79-1.63, p = 0.51). In summary, we found that there was no significant trend towards improved OS between patients with low-risk BC treated with IORT vs. EBRT. We included only the lowest risk BC patients, including those with DCIS, to allow for the detection a potentially small harmful effect of EBRT. Further study is warranted to determine if IORT or other partial breast irradiation techniques result in reduced nBCM, especially in an era of more precise, conformal EBRT delivery techniques.