Analyzing blood fatty acid (FA) patterns could better capture FA interactions and aid in characterizing the relation to prostate cancer (PCa). We aimed to assessdata-derived FA patterns and PCa risk. We conducted a nested case-control among men in the Physicians' Health Study. FA levels were measured in pre-diagnostic whole blood samples of 476 cases and their matched controls to identify 28 individual FA. FA patterns were identified using principal component analysis. Conditional logistic regression was used to estimate RRs and 95% CIs of PCa across quintiles of FA patterns. Two patterns explaining 40.9% of total variation were identified. Pattern 1 was characterized by high levels of trans-FA and α-linolenic acid along with low levels of n-3 PUFA. Pattern 2 was characterized by high levels of 14 and 16 carbon saturated FA and MUFA along with low levels of α-linolenic acid and saturated FA of 18 or more carbons; this pattern largely reflects de novo lipogenesis. There was a positive but nonsignificant association between pattern 1 and PCa risk (RRQ5 vs Q1=1.37, 95%CI: 0.91-2.05). Pattern 2 was statistically significantly associated with higher PCa risk (RRQ5 vs Q1=1.63, 95%CI: 1.04-2.55). This association was similar across tumor stage, grade, and clinical aggressiveness categories. The two FA patterns were consistent with known interactions between FA intake and metabolism. A pattern suggestive of higher activity in the de novo lipogenesis pathway was related to higher PCa risk.
