Editorial NexoScience
Few pieces of nutrition advice feel as unquestionable as reaching for the olive oil bottle. A randomized crossover trial published in the Journal of the American Heart Association put that reflex to the test inside a diet that was already healthy, and the result went the other way. The version with barely any extra virgin olive oil lowered LDL cholesterol more than the version soaked in it, and it also did better on total cholesterol, glucose and body weight.
The trial was designed to isolate the oil, not the diet. Forty adults at borderline to high cardiovascular risk, defined as an estimated risk of 5% or more, completed the protocol at the University of Florida. Each of them followed, in random order, four weeks of a whole-food plant-based vegan diet high in extra virgin olive oil, at four tablespoons a day or roughly 52 grams, and four weeks of exactly the same diet with less than a teaspoon a day. Between the two phases there was a one-week washout. The oil was supplied by the researchers and had to be consumed raw, the participants attended weekly virtual cooking classes with a dietitian and chef, and nobody was told to count calories or portions. The trial is registered on ClinicalTrials.gov under the identifier NCT04828447.
Both diets improved the cardiometabolic profile, but one did it harder. During the first four-week period, LDL cholesterol fell by 25.5 mg/dL after the low oil diet and by 16.7 mg/dL after the high oil one, a gap that did not reach statistical significance on its own, with P=0.162. The other markers pointed the same way and some did clear that bar. Total cholesterol dropped 33.8 versus 19.0 mg/dL, with P=0.035. Apolipoprotein B fell 14.8 versus 5.5 mg/dL, with P=0.053. Glucose fell 17.8 versus 8.8 mg/dL, with P=0.082. Average body weight went from 89.0 kilograms to 83.1 after the low oil phase and to 83.8 after the high oil one, a difference of 0.7 kilograms in favour of the low oil version, with P=0.011.
The order in which people ate turned out to matter more than the diets themselves. This is the finding the authors consider most informative. Among those who went from high to low oil, LDL cholesterol dropped a further 12.7 mg/dL when the oil was removed, with P=0.036. Among those who went the other way, from low to high, adding the oil raised LDL by 15.8 mg/dL, with P=0.022, and dragged along glucose, up 15.3 mg/dL with P=0.004, and total cholesterol, up 29.4 mg/dL with P=0.0002. Averaged across the whole trial without accounting for order, the difference between the two diets on LDL was a trivial 1.6 mg/dL, with P=0.722. The signal only appears when you look at the sequence.
The amount of fat was not the variable that decided the outcome. Fat made up 48% of energy during the high oil phase and 32% during the low one, and yet both phases lowered LDL cholesterol compared with the participants’ usual omnivorous diet. The difference was where the fat came from. During the low oil weeks it arrived mostly through avocado, nuts, seeds and whole olives, foods that keep their fibre and their phytochemicals intact instead of being pressed out of them. Dietary fibre rose by around 8 to 9 grams a day in both phases, and the high oil phase carried slightly more saturated fat, which may have contributed to the gap.
None of this makes extra virgin olive oil a bad food. The case for it comes largely from observational studies of the Mediterranean diet, where the oil replaces butter, lard, mayonnaise and animal fats, and in that swap the benefit is real. What those studies do not show is any advantage when the oil replaces other vegetable oils, and an analysis of the PREDIMED cohort found a 25% lower risk of major cardiovascular events among the highest consumers of whole olives, which points at the food rather than the extract. That is the crack this trial opens. The authors write that extra virgin olive oil “may not be the beneficial additive” of a Mediterranean diet in this particular context, and that adding it after a period of low intake can block further reductions in cholesterol.
Some of the numbers cut against the tidy version of the story. HDL cholesterol, the kind usually filed under good, also fell more on the low oil diet, by 10.5 against 5.0 mg/dL with P=0.025, although current evidence suggests HDL function matters more than the raw figure. Lipoprotein(a), a largely genetic risk factor, dropped 17.1 mg/dL in the high to low sequence with P=0.005, something the authors did not expect. And two inflammation markers, interleukin-6 and high-sensitivity C-reactive protein, moved in opposite directions, which the researchers attribute to the small sample or to assay error.
The limitations here weigh as much as the result. This was a randomized trial, so within its own design it can talk about cause and not just correlation, but the design is narrow. Forty people, four weeks per phase, a mean age of 64.4 years and a sample made up mostly of well-educated white women, which the authors themselves flag as a limit on generalisability. The one-week washout proved too short and left carryover effects, which is precisely why the order mattered so much. Calories and macronutrients were not matched between phases, food intake was self-reported, and the whole thing ran during the pandemic. Above all, this was tested inside a whole-food plant-based vegan diet, not inside an ordinary omnivorous one, where an olive oil that displaces butter is a completely different question.
The practical reading is not to throw out the bottle. It is that within a diet already built on unprocessed plant foods, piling on extra virgin olive oil does not add the extra benefit that is usually assumed, and in this trial it took some away. Whether that holds over more than four weeks, and in people who do not eat this way, is still open.
Study: “Recipe for Heart Health: A Randomized Crossover Trial on Cardiometabolic Effects of Extra Virgin Olive Oil Within a Whole-Food Plant-Based Vegan Diet”, by Krenek AM, Mathews A, Guo J, Courville AB, Pepine CJ, Chung ST and Aggarwal M. Published in the Journal of the American Heart Association (2024). DOI: 10.1161/JAHA.124.035034.
