Abstract PD2-3: High body mass index (BMI) and worse response to neoadjuvant chemotherapy (NACT) by breast cancer phenotype: Own data and external validation on German breast group (GBG) patients

Abstract

Abstract Introduction It remains controversial to what extent BMI predicts pathologic complete response (pCR) following NACT. We evaluated pCR by ‘BMI category’ in different breast cancer subtypes and validated our findings in a larger dataset from the GBG. Patients and methods A retrospective study from UZL with consecutive breast cancer patients treated between 01-01-2000 and 31-12-2011 with different types of NACT +/- trastuzumab (dose capping BSA≥2m2) followed by surgery. pCR1 (ypT0N0) and pCR2 (ypT0/isN0) were reported in each of the intrinsic breast cancer subtypes based on ER and HER-2 expression. The effect of BMI [BMI 1(≤25kg/m2), BMI 2 (>25-29,9 kg/m2) and BMI 3 (≥ 30 kg/m2)] on both pCR definitions was analyzed in univariate (Chi square test) and multivariate (logistic regression) model. Negative ER and HER-2 status were defined according to ASCO/CAP guidelines (e.g. <1% ER positive tumor cells). Data were validated by GBG using different NACT +/- trastuzumab regimen (no dose capping BSA≥2m2) and similar biomarkers for defining breast cancer subtypes although ER-negative status was defined as <10% ER positive tumor cells. Results We included 267 breast cancer cases while the GBG validation was done in 8874 cases. Multivariate model for UZL patients didn't retain ‘BMI category’ as a predictor for any of the pCR definitions. However, univariate analysis found a significant association between BMI and pCR2 (p = 0,048). The tables show the numerical differences for pCR2 in each BMI category by breast cancer subtype for UZL and GBG patients. UZL data suggest a numerical trend with lower pCR2 rate in obese women, which seems more pronounced in ER-negative breast cancer. Results from the much larger GBG dataset with more power to assess ‘BMI category’ for pCR2 in different breast cancer subtypes using a multivariate model are presented in another abstract. Conclusion Despite the small cohort of patients, our data suggest a lower pCR rate in obese women. External validation by GBG confirms a significant association between BMI and pCR2 in uni- and multivariate model, which is also significant for luminal A/B breast cancer. The most obvious reason for our observation was dose capping but based on the GBG dataset where dose capping was avoided where possible, we speculate that immunologic factors, the microenvironment or alternative signaling pathways may affect sensitivity to NACT +/- trastuzumab. Numerical pCR2 differences by BMI category and breast cancer subtype for UZL patients Luminal A/BLuminal HER2HER 2 likeTNBCTotalUZLn = 111n = 40n = 39n = 75n = 267BMI 16.6% (4/61)35.0% (7/20)63.2% (12/19)31.4% (11/35)25.2% (34/135)BMI 23.1% (1/32)50.0% (7/14)56.3% (9/16)27.3% (6/22)27.3% (23/84)BMI 30% (0/18)33.3% (2/6)50.0% (2/4)15.8% (3/19)16.6% (8/48) Numerical pCR2 differences by BMI category and breast cancer subtype for GBG patients Luminal A/BLuminal HER2HER 2 likeTNBCTotalGBGn = 3250n = 1077n = 806n = 1570n = 8847BMI 111.6%(187/1618)26.7%(147/550)47.3%(191/404)38.9%(300/772)22.5%(982/4358)BMI 210.5%(128/1026)24.9% (81/325)44.2%(111/251)37.4% (187/500)21.2%(596/2813)BMI 37.9% (48/606)22.8%(46/202)41.1%(62/151)31.9%(95/298)18.3%(312/1703) Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr PD2-3.