Obesity and breast cancer prognosis: pre-diagnostic anthropometric measures in relation to patient, tumor, and treatment characteristics.

PURPOSE: Examine the association between obesity and clinical outcomes in early breast cancer and assess if patient, tumor, and treatment characteristics modify such associations in Malmö Diet and Cancer Study patients (MDCS). METHODS: The MDCS enrolled 17,035 Swedish women from 1991 to 1996. At enrollment, participants' body mass index (BMI), waist circumference and body fat percentage measures were collected. We identified all female MDCS participants with invasive breast cancer from 1991 to 2014. Follow-up began at breast cancer diagnosis and ended at breast cancer recurrence (BCR), death, emigration, or June 8, 2020. The World Health Organization guidelines were used to classify BMI, waist circumference, and body fat percentage into three categories of healthy weight, overweight, and obesity. We fit Cox regression models to compute adjusted hazard ratios (HRs) with 95% confidence intervals (CI) of BCR according to body composition. To evaluate effect measure modification, we stratified Cox models by patient, tumor, and treatment characteristics. RESULTS: In total, 263 BCRs were diagnosed over 12,816 person-years among 1099 breast cancer patients with a median follow-up of 11.1 years. Obesity according to BMI (HR = 1.44 [95%CI 1.00-2.07]), waist circumference (HR = 1.31 [95%CI 0.98-1.77]), and body fat percentage (HR = 1.41 [95%CI 1.02-1.98]) was associated with increased risk of BCR compared with healthy weight. Obesity was stronger associated with BCR in patients with low socioeconomic position (HR = 2.55 [95%CI 1.08-6.02]), larger tumors > 20 mm (HR = 2.68 [95%CI 1.42-5.06]), estrogen-receptor-negative breast cancer (HR = 3.13 [95%CI 1.09-8.97]), and with adjuvant chemotherapy treatment (HR = 2.06 [95%CI 1.08-4.31]). CONCLUSION: Higher pre-diagnostic BMI, waist circumference, and body fat percentage was associated with increased risk of BCR. The association between obesity and BCR appears dependent on patient, tumor, and treatment characteristics.