Every claim, every source.

This 2024 Nature Cell Biology paper from the Madeo lab identified spermidine — a polyamine found in many foods (wheat germ, soybeans, mushrooms, aged cheeses) and produced endogenously — as the essential mediator of fasting-induced autophagy. The authors ran experiments across multiple model systems: yeast, nematodes, mouse cells, and human cell lines (U2OS osteosarcoma cells and H4 neuroglioma cells). Across all systems, blocking spermidine synthesis with the inhibitor DFMO suppressed fasting-induced autophagy — and supplementing exogenous spermidine (100 µM) rescued the autophagy response. The paper also reports human-cohort metabolomics: across multiple cohorts of fasting participants (61 to 109 volunteers per cohort, fasting durations 3 to 16 days), serum spermidine levels rose during fasting. Human PBMCs showed increased hypusination of eIF5A — a downstream effect linking spermidine to translation control and autophagy machinery. The paper's mechanistic claim is significant: spermidine is not just correlated with fasting-induced autophagy; it is required for the response to occur.

This study used data from the Canadian Study of Adolescent Health Behaviors — a national survey of 2,762 adolescents and young adults — to ask how common intermittent fasting is and whether it correlates with eating-disorder behaviors. The engagement numbers were striking: roughly 48 percent of women, 38 percent of men, and 52 percent of transgender or gender-non-conforming respondents reported practicing some form of intermittent fasting in the past 12 months. The researchers used the Eating Disorder Examination Questionnaire alongside modified Poisson regression to measure associations. Across all three gender groups, intermittent fasting in the past 12 months and the past 30 days was significantly associated with elevated eating-disorder psychopathology — disordered cognitions, restrictive behaviors, and binge-purge cycles. The pattern was strongest and most consistent in women. The authors do not claim fasting causes eating disorders; the data are cross-sectional and cannot prove direction. They argue clinicians screening young patients should treat self-reported intermittent fasting as a meaningful flag.

This randomized controlled trial enrolled 71 adults with metabolic syndrome and randomized them to a five-day modified Buchinger-style fast followed by a modified DASH diet versus DASH diet alone. Investigators measured 16S rRNA gut microbiome composition, ambulatory blood pressure, antihypertensive medication requirements, and standard cardiometabolic biomarkers at baseline, immediately post-fast, and at three months follow-up. The fasting plus DASH arm showed greater reductions in systolic blood pressure, in the requirement for antihypertensive medications, and in body-mass index at three months than the DASH-only arm. Gut microbiome analysis identified specific bacterial taxa — including changes in genera linked to short-chain fatty acid production and to microbial pathways relevant to host metabolic regulation — that responded to the fast, with changes that partly persisted into the post-fast period. The paper is one of the few human RCTs to combine a multi-day fasting intervention with comprehensive microbiome characterization and clinically meaningful blood pressure endpoints.

This Cleveland Clinic-affiliated study followed 1,403 patients who were eligible for a protein-sparing modified fast program over 5 years to answer the question the original 1970s PSMF literature could not: does the dramatic short-term weight loss persist? Of those eligible, 879 (63 percent) actually initiated PSMF; the remaining 524 (37 percent) pursued other dietary approaches and served as a comparison cohort. The 1-year outcomes were dramatic and favored PSMF: -7.6 percent body weight in the PSMF arm versus -1.8 percent in the comparison arm, a 5.8-percentage-point difference (p less than 0.01). At 3 years, PSMF still showed an advantage but smaller: -2.3 percent vs -0.9 percent, a 1.4-point difference. By 5 years, the difference had effectively disappeared: -1.4 percent vs -1.0 percent (p=0.64, not statistically significant). The proportion achieving clinically meaningful (≥5 percent) weight loss told the same story: PSMF was strongly favored at 1 and 3 years, equivalent at 5 years. The honest conclusion: PSMF produces substantial short-term weight loss with good durability through year 3, but by year 5 the advantage over conventional dietary care is gone.

The REDUCE-IT trial randomized 8,179 statin-treated adults with elevated triglycerides and either established cardiovascular disease or diabetes plus risk factors to receive 2 grams of icosapent ethyl twice daily (a purified prescription-grade EPA preparation, total daily dose 4 grams) or matching placebo. After a median follow-up of 4.9 years, the icosapent ethyl arm experienced a 25% relative reduction in the primary composite endpoint of major adverse cardiovascular events (cardiovascular death, nonfatal myocardial infarction, nonfatal stroke, coronary revascularization, or unstable angina) compared with placebo. Reductions were observed across multiple individual endpoint components, including cardiovascular death. The trial reignited debate over high-dose omega-3 cardiovascular prevention after several earlier mixed-result trials and stands as the largest, longest, and methodologically strongest RCT supporting a cardiovascular benefit from a specific EPA preparation.

This NEJM review summarizes evidence that intermittent fasting regimens — alternate-day fasting, time-restricted eating, and periodic multi-day fasts — engage a "metabolic switch" from glucose-derived energy to fat- and ketone-derived energy after hepatic glycogen is depleted, typically within 12–36 hours of fasting depending on the individual and the protocol. The authors argue that repeated exposure to this switch produces adaptive responses across organ systems, including improved insulin sensitivity, reduced inflammation, increased mitochondrial biogenesis, enhanced autophagy, and improved stress resistance in cells. The review compiles findings from animal models alongside the available human trials at the time of publication. The review notes that, despite preclinical signals being strong and consistent, the human evidence base is more heterogeneous: the largest gains in metabolic markers (fasting insulin, HOMA-IR, lipid profile, inflammatory markers) appear in adults with obesity or metabolic syndrome, while effects in lean, metabolically healthy individuals are smaller. The authors flag practical issues — adherence over months, the early-fast hunger and irritability phase, and the lack of long-term outcome data — as the main barriers to clinical adoption rather than safety in healthy adults.

This is the first supervised controlled-feeding trial designed specifically to isolate intermittent fasting's metabolic effects from weight loss. Men with prediabetes were enrolled in a randomized crossover trial: 5 weeks of early time-restricted feeding (eTRF — a 6-hour eating window, with the last meal before 3 p.m.), followed by 5 weeks of a control schedule (12-hour eating window), then crossover. Critically, participants were fed enough food to maintain their weight in both conditions — the eating window changed, but total energy intake did not. Even without weight loss, eTRF improved insulin sensitivity, beta-cell responsiveness, systolic and diastolic blood pressure, oxidative stress (8-isoprostane), and evening appetite. The improvements demonstrate that intermittent fasting's cardiometabolic benefits are not solely mediated by weight loss — circadian alignment of eating and the duration of the daily fasting window have independent effects. The paper has been highly influential because it isolated the eating-window mechanism from the calorie-deficit mechanism.

This is the largest published study of sustained nutritional ketosis as a T2D management strategy. The Virta Health study enrolled 349 adults with type-2 diabetes — 262 in the continuous care intervention (CCI, an app-mediated remote-care program with macronutrient guidance toward sustained nutritional ketosis) and 87 in usual care. The design was open-label and non-randomized (participants self-selected into the intervention), so it sits below DiRECT's RCT evidence in the hierarchy — but the sample is larger and the duration is longer. At one year, the intervention group's HbA1c fell from 7.6 to 6.3 percent (the threshold for diabetes remission), mean weight loss was 13.8 kg, and 94 percent of insulin users reduced or eliminated insulin therapy. Sulfonylureas were discontinued completely in the CCI group. Secondary markers improved across the board: HOMA-IR dropped 55 percent, hsCRP dropped 39 percent, triglycerides dropped 24 percent, HDL-C rose 18 percent. The usual-care arm showed no meaningful change on any of these endpoints.

DiRECT is the trial that proved type-2 diabetes is reversible through structured weight loss in routine primary care. 306 adults aged 20–65 with T2D diagnosed within the past six years and BMI 27–45 were enrolled across 49 GP practices in Scotland and Tyneside; the practices, not the patients, were randomised. The intervention had three phases: total diet replacement (an 825–853 kcal/day formula diet for 3–5 months) with diabetes and blood-pressure medications stopped, structured food reintroduction over 2–8 weeks, then long-term weight-maintenance support. At 12 months, 46% of intervention participants achieved diabetes remission (HbA1c < 6.5% off all glucose-lowering medications) compared to 4% of usual-care controls. Mean weight loss was 10 kg in the intervention arm versus 1 kg in the control arm. Remission tracked weight loss tightly: 86% of those losing ≥15 kg achieved remission, while none who gained weight did.

This Australian Institute of Sport study is the most prominent counter-evidence to keto-adapted athletic performance claims. Burke and colleagues randomized 29 elite race walkers to one of three 3-week dietary conditions during intensified training: continuously high carbohydrate availability (HCHO), periodized carbohydrate availability (PCHO — same total intake but timed around training), or low-carbohydrate high-fat (LCHF — under 50 g/day carbs, 78 percent of energy from fat). All three diets were isocaloric. The findings cut against simple "keto is good for endurance" narratives. Peak aerobic capacity (VO2max) improved across all three diets. But race-walking economy — the oxygen cost per unit speed at race-relevant velocities — got worse on LCHF. The keto-adapted walkers needed more oxygen to walk at the same pace, even with elevated fat oxidation. Net result: 10 km race time did not improve on LCHF (about -1.6 percent change, not statistically meaningful) while both carbohydrate-available groups improved 5–7 percent. The conclusion was unambiguous: for elite endurance athletes performing at race-relevant intensities, LCHF impaired performance despite increasing fat oxidation. The paper has been replicated by the same group with different cohorts.

This prospective cohort study followed 1,272 adolescent females over multiple years, identifying which baseline psychosocial and behavioral risk factors predicted the future onset of distinct DSM-5 eating disorders (anorexia nervosa, bulimia nervosa, binge eating disorder, and other eating disorder presentations). The study design enabled differentiation among predictors — body dissatisfaction and dietary restriction predicted multiple eating-disorder onsets, but specific risk-factor profiles also distinguished anorexia onset from bulimia onset from binge-eating onset. Stice and colleagues are among the leading research groups in eating-disorder prevention; this paper supplies one of the cleaner empirical bases for who is at elevated risk for which eating-disorder presentation, with implications for how broader nutrition interventions should screen and refer at-risk participants.

This randomized pilot study asked the cleanest possible head-to-head question for intermittent fasting: when matched for the goal of weight loss, does alternate-day fasting beat ordinary daily caloric restriction? Adults with obesity (BMI ≥30, age 18–55) were randomized to either zero-calorie alternate-day fasting (ADF, n=14) or moderate daily caloric restriction (CR at -400 kcal/day, n=12) for 8 weeks, followed by 24 weeks of unsupervised follow-up. The ADF arm achieved a substantially larger calculated energy deficit (about 376 kcal/day greater than CR), yet the actual weight loss was statistically indistinguishable: ADF -8.2 kg vs CR -7.1 kg over 8 weeks. Body composition, lipids, and insulin sensitivity index showed no significant between-group differences. Safety was strong — no adverse effects, 93 percent completion in the ADF arm. Twenty-four-week unsupervised follow-up showed similar weight regain in both groups, but the ADF arm trended toward more favorable lean-mass preservation. The honest conclusion: ADF is a safe and tolerable alternative to daily restriction with equivalent short-term outcomes, not a superior intervention.

This eight-week randomized trial enrolled 34 resistance-trained males and assigned them either to a 16:8 time-restricted-feeding pattern (eating window 1pm to 8pm) or to a normal-eating-pattern control while continuing standardized resistance training across both arms. The TRF group showed reductions in fat mass, fasting glucose and insulin, IGF-1, leptin, and inflammatory markers (IL-6, TNFα), and an improved testosterone-to-cortisol ratio, while maintaining muscle area and maximal strength on standard one-rep-max testing. Total energy and protein intake were matched approximately between groups. The study is one of the cleaner demonstrations that an intermittent-fasting-style eating pattern can be combined with resistance training without performance decrement and with favorable body-composition and biomarker changes in already-trained adults.

The FASTER (Fat-Adapted Substrate utilization in Trained Elite Runners) study compared 20 elite ultra-endurance athletes — 10 habitually consuming a high-carbohydrate diet (59 percent carbs) and 10 long-term keto-adapted (10 percent carbs, 70 percent fat, average 20 months on the diet) — across maximal and submaximal exercise testing. The headline finding was record-setting: peak fat oxidation in the keto-adapted athletes was 2.3-fold higher than in the carb-adapted group (1.54 vs 0.67 grams per minute), the highest fat-oxidation rates ever recorded in humans during exercise. During submaximal exercise (3-hour run at 64 percent VO2max), fat contributed 88 percent of the energy in keto-adapted athletes versus 56 percent in carb-adapted athletes. Notably, muscle glycogen utilization and post-exercise glycogen repletion were similar between groups despite the dramatic substrate-source shift — meaning keto-adapted athletes used proportionally less carbohydrate from glycogen stores during the run, so their glycogen actually lasted longer. The paper transformed how the field thinks about athletic substrate use: humans can adapt to fat as their dominant fuel without losing the ability to use carbohydrate when it matters.

This Cell Metabolism paper from Valter Longo's USC group introduced the fasting-mimicking diet (FMD) — a 5-day periodic dietary protocol designed to deliver fasting's molecular benefits while keeping participants able to consume modest amounts of plant-based food. The paper has two parts. In aged mice, monthly FMD cycles for several months produced multi-system regeneration: hippocampal neurogenesis rose, IGF-1 dropped, PKA activity decreased, NeuroD1 expression increased, and cognitive performance improved on standard mouse cognition tests. In a 38-participant pilot human RCT, three monthly FMD cycles (each 5 days) produced reductions in body weight, body fat, blood pressure, fasting glucose, and IGF-1 without significant adverse events. The paper is foundational because it bridged rodent CR research and practical human protocol design — providing a structured, safe framework for delivering fasting benefits without continuous calorie restriction. Longo subsequently commercialized the protocol as ProLon, a packaged 5-day FMD product. The paper's data quality is solid but the commercial development complicates how it should be cited.

This Canadian government laboratory study tested 52 canned fish products from the 2014 Canadian retail market for bisphenol A (BPA) and three related compounds (BPB, BPE, BPF) using gas chromatography mass spectrometry. The headline finding: BPA was detectable in every one of the 52 products, but at substantially lower levels than a comparable study from five years earlier. The concentration range was 0.96 to 265 nanograms per gram, with an average of 28 ng/g. Three of the four BPA analogues were essentially absent — BPB and BPE were not detected in any product, and BPF appeared in only four samples at low concentrations (1.8 to 5.7 ng/g) — suggesting BPA is still the dominant epoxy resin used in current can liners. The few outliers above 100 ng/g came from a single newly-marketed brand, indicating that brand-level differences in liner formulation drive most of the variation. Industry-wide, the data show measurable downward progress in canned-fish BPA exposure.

This study established a foundational point in microbiome research: dietary changes alter gut microbial composition rapidly and reproducibly. Ten participants alternated between an entirely animal-based diet (meat, eggs, cheese) and an entirely plant-based diet (grains, legumes, fruits, vegetables) for five days each. Microbiome composition shifted within 24 hours of dietary change and reverted within 48 hours of returning to baseline diet. The animal-based diet specifically increased the abundance of bile-tolerant microorganisms (Bilophila wadsworthia, Alistipes putredinis, Bacteroides) and decreased the abundance of Firmicutes that metabolize plant polysaccharides. Functional metagenomic analysis confirmed corresponding shifts in microbial gene expression. The paper is the canonical reference for the rapid-response biology of the human gut microbiome to dietary substrate change and for the bidirectional, plastic nature of these shifts.

This Cell Metabolism paper combined a large NHANES-based human cohort (2,253 adults followed over 18 years) with mouse experiments to ask whether high protein intake — especially animal protein — drives cancer and mortality risk via IGF-1 and growth-hormone signalling. The headline finding is age-dependent. In adults aged 50–65, those reporting high protein intake (≥20 percent of calories from protein) had a 75 percent higher overall mortality and a fourfold higher cancer death risk over the next 18 years compared to low-protein eaters (under 10 percent of calories). The effect was largely abolished when the protein came from plant sources rather than animal sources. After age 65, the relationship reversed: high protein became protective for cancer and overall mortality — though high protein at any age was associated with a fivefold increase in diabetes mortality. Mouse experiments supported the mechanism: high-protein diets accelerated tumour growth and elevated IGF-1, while protein restriction did the opposite. The interpretation is that protein's relationship with longevity is not monotonic; it depends on age, on the protein source, and on what's being optimized for.

The 2012 European Union health-claim regulation that defines the regulatory threshold for what counts as a polyphenol-active olive oil. The claim — "olive oil polyphenols contribute to the protection of blood lipids from oxidative stress" — is authorised only for olive oils that deliver at least 5 milligrams of hydroxytyrosol and its derivatives (oleuropein complex and tyrosol) per 20 grams of olive oil consumed. That works out to a tissue-relevant 250 mg per kilogram of olive oil at the daily-intake reference of 20 g. The claim is supported by a body of EFSA-reviewed mechanistic and intervention evidence on phenolic compounds and oxidative stress markers, and is the most widely cited regulatory basis for the "extra virgin olive oil polyphenols are different from refined olive oil" distinction. The regulation is also the EU's reference point for "extra virgin" olive oil quality grading more broadly (acidity ≤0.8%, panel test, polyphenol content). Refined olive oil typically loses 84–87% of polyphenol content during refining.

This study tested 17 different brands of canned sardines, sourced from six countries and bought at retail in eastern Kentucky, for the four most concerning heavy metals in fish: arsenic, cadmium, lead, and mercury. Each brand was analyzed as a composite of 3 to 4 fish using standard atomic-absorption laboratory methods. The headline finding for sardines was clear: mercury was below the 0.09 microgram-per-gram detection limit in every sample tested. That is well under the 1.0 microgram-per-gram FDA action level for predatory fish like tuna and swordfish, and roughly a tenth of the typical canned-tuna averages reported in FDA surveillance data. Arsenic was the highest of the four metals on average (1.06 µg/g), with the highest values appearing in samples from Norway and Thailand; cadmium was highest in Moroccan brands; lead was highest in Canadian brands. The study supports the narrow but important claim that commercial canned sardines, across multiple sourcing countries, are a low-mercury fish.