How Come EPA+DHA Improved but Insulin Action Didn't? stephen phinney, Beyond Obesity (Independent Scientist) 9 January 2015 The data in this report are very interesting, but the authors¿ interpretation is puzzling. It appears that the authors have decided a priori that the only important fatty acids in muscle PL vis-à-vis insulin action are EPA and DHA. Thus they proudly report that DHA increases from 2.08 to 2.81 wt% and EPA from 0.86 to 1.13 across the 24-week ¿maintenance¿ period (which actually resulted in a significant loss of fat mass due to Orlistat Rx). However in this same time period, there were no commensurate improvements in serum insulin, glucose, or HOMA-IR. In fact, all three of these parameters of insulin action got worse (albeit non-significantly) as EPA and DHA contents in the membrane PL rose significantly. Twenty years ago we published data on serum PL fatty acids before, during, and after a VLCD resulting in 18 kg of weight loss (ref below). The most prominent change among the membrane highly unsaturated fatty acids was a prompt and dramatic rise in arachidonate (20:4n-6) during the VLCD, followed by an equally prompt and dramatic decline back to pre-diet content despite maintaining a major weight loss. The data in this current report corroborate and extend our findings into muscle membranes as well (ie, 12.04 before, 13.64 after, and back down to 12.67 wt% after 24 weeks). Calculated as wt%, the total content of fatty acids in the PL fraction is a ¿zero-sum game¿. That is, if one fraction increases, the sum of the others has to go down in equal proportion. Further, since 20:4n-6 competes directly with EPA + DHA for the sn-2 position in phospholipids, the 0.97 wt% decline in arachidonate (13.64 ¿ 12.67) that the authors report can effectively explain the total rise of EPA+DHA (1.00 wt%) during the maintenance period. Thus the net total of these three fatty acids did not change during maintenance, and thus it is no surprise that none of the measures of insulin action got better (despite a further loss of 3 kg of fat mass), and in fact most trended worse (eg, HOMA-IR from 2.2 to 2.8, and serum insulin from 57 to 69). In Borkman¿s seminal NEJM study (ref #6 in the current report), among the individual skeletal muscle PL fatty acids, the strongest correlations with serum insulin in both cohorts were for arachidonate (-0.63 and -0.73). Neither EPA nor DHA as individual fatty acids were significantly negatively correlated with serum insulin (or positively correlated with measured insulin sensitivity in the subjects who has glucose clamps done). There is much work yet to be done in the area of muscle membrane fatty acids and insulin sensitivity, but to make progress we need to look objectively at the data. Much of the existing data indicates that arachidonate undergoes dramatic redistribution during VLCD (and also carbohydrate-restricted) weight loss, and that its changes may play a dominant role in the improved insulin action that occurs during nutritional ketosis. Stephen Phinney, MD, PhD Professor of Medicine emeritus, UC Davis Phinney SD, Davis PG, Johnson SB, Holman RT. Obesity and weight loss alter serum polyunsaturated lipids in humans. Am J Clin Nutr. 1991 Apr;53(4):831-8. Competing interests Author of: The New Atkins for a New You (2010), and The Art and Science of Low Carbohydraqte Living (2011).