Covariate clustering: Women with breast cancer in southwestern Paraná, Brazil

Abstract

IntroductionDue to the high incidence and aggressiveness of breast cancer, understanding specific factors associated with the profile of the disease is necessary. Thus, the study aimed to analyze data from 155 patients with breast cancer, grouping them according to their clinicopathological characteristics, attended at a reference hospital for Oncology, in 2015–2020, in the southwest region of Paraná, Brazil. Material and MethodsUsing multivariate statistical analysis, sample data were divided into three clusters. The heterogeneity between clusters was obtained by Ward's method. The clinicopathological variables obtained from the patients' medical records were: the presence of intratumoral emboli and lymph nodes, menopausal status, molecular subtype, histological grade, TNM staging of the disease, tumor size, age at diagnóstico, weight, height, and body mass index. ResultsIt is observed that 70% of the patients were in menopause at diagnóstico, 31.5% had tumors containing emboli, and 41% had positive lymph nodes. The prevalence of Luminal B subtype, intermediate histological grade, and TNM staging II was verified. The prevalence of the disease was higher in women aged over 50 years, representing 66% of cases. The BMI of the patients ranged from 17.63 kg/m2 to 51.26 kg/m2, with 73.55% above 25 kg/m2. Using the spatial distribution of patients, cluster analysis identified the regions with the worst averages of clinicopathological variables and the highest number of cancer cases. ConclusionThrough the statistical analysis, it was possible to determine the heterogeneity of the data, so the patients were separated into three clusters. When analyzing the obtained clusters, each one of them had specific characteristics.