Fig. 4

Results from partial least squares supervised analysis (PLS-DA). Models built using significantly different serum fatty acids in (A) preoperative patients and 12 M follow-up (n = 27), B preoperative patients and 24 M follow-up (n = 19), C patients at 12 M and 24 M follow-up (n = 18), D preoperative patients (n = 28) and control (n = 25), E patients at 12 M follow-up (n = 27) and controls (n = 25), F patients at 24 M follow-up (n = 19) and controls (n = 25). Fatty acids included as variables were selected based on significance in paired t-tests for comparisons between preoperative patients and follow-ups (see Tables S4, S5 and S6 respectively) or two-way t-Student’s test for comparisons between breast cancer patients and controls (Table S8)