Fasting, Calorie Cuts, or Keto: What Lowers Inflammation?

Dietary restriction can lower inflammatory biomarkers, but the best current evidence does not support choosing a diet because it supposedly targets one cytokine better than another. Calorie restriction, intermittent fasting, and ketogenic diets each improved at least one marker in trials of adults with overweight or obesity. Yet when researchers matched calorie intake or achieved weight loss, many of the apparent differences between approaches disappeared.

The more useful question is therefore less glamorous: which approach can you sustain long enough to reduce-and maintain-fat mass? Biology rarely objects to a less cinematic answer.

Three diet strategies, three apparent biomarker patterns

A 2024 systematic review and meta-analysis by Aamir et al. offers the cleanest direct comparison of calorie restriction (CR) and intermittent fasting (IF). It included randomized controlled trials lasting at least eight weeks in adults with overweight or obesity: 15 CR trials and six IF trials. The investigators examined C-reactive protein (CRP), interleukin-6 (IL-6), and tumor necrosis factor-alpha (TNF-alpha), all blood markers associated with inflammation. They are biomarkers, not proof that a diet prevents heart disease, diabetes, cancer, or any other clinical outcome.

CR significantly reduced CRP, with a standardized mean difference (SMD) of -0.15 (95% CI, -0.30 to -0.00; p=0.04), and IL-6, with an SMD of -0.31 (95% CI, -0.51 to -0.10; p=0.004). CR did not significantly reduce TNF-alpha. IF showed the opposite pattern in that analysis: TNF-alpha fell significantly, with an SMD of -0.32 (95% CI, -0.63 to -0.02; p=0.04), while CRP and IL-6 did not significantly change.

That pattern makes a neat headline: calorie restriction for CRP and IL-6, fasting for TNF-alpha. The data deserve more restraint. Aamir et al. did not report effect sizes and confidence intervals for the null comparisons, so the published findings cannot produce a complete, comparable three-by-three table. More importantly, the IF finding rests on six trials and its confidence interval only barely excludes no effect.

Consistency differs sharply by marker

The CRP result for CR had no observed between-study heterogeneity (I²=0%). That makes it the most consistent finding in this comparison, although the effect size itself was small. The IL-6 estimate was less stable: I²=73% indicates substantial variation among studies, and the analysis did not identify why results differed. A pooled IL-6 estimate of -0.31 is not a reliable forecast for any individual person or protocol.

The authors also flagged concerns about randomization and allocation concealment in included trials. Meta-analysis can combine studies; it cannot repair weak trial conduct.

Keto has the largest reported CRP drop-on the thinnest evidence

A separate 2024 meta-analysis of ketogenic-diet randomized trials included five trials and 218 adults with overweight or obesity. CRP fell by -0.62 mg/dL. Numerically, that is the largest reported CRP change among the central reviews, but it is an absolute-unit estimate, whereas Aamir et al. reported standardized mean differences. Those figures are not interchangeable, so they cannot establish that keto outperforms CR.

The keto evidence is preliminary for a more basic reason: five studies met only the review authors' minimum threshold for assessing publication bias. Most included protocols likely used very-low-calorie ketogenic diets, which combine sharp carbohydrate restriction with severe calorie deficits. If CRP falls, the trial design cannot tell us how much credit belongs to ketosis and how much belongs to the deficit and resulting weight loss.

A proposed explanation involves beta-hydroxybutyrate, a ketone body that may inhibit the NLRP3 inflammasome and thereby affect IL-1beta-related CRP production. It is biologically plausible, but a plausible pathway is not clinical proof. The keto meta-analysis did not establish the size of any ketone-specific effect independent of calorie restriction.

A 2024 meta-analysis by Kazeminasab et al. helps explain the problem. It found that low-carbohydrate diets and concurrent caloric restriction may compound one another's effects on CRP. That supports the possibility that very-low-calorie keto protocols act through both carbohydrate restriction and energy deficit, while leaving their relative contributions unresolved.

Fasting subtype rankings are useful clues, not diet prescriptions

A 2025 network meta-analysis by Khalafi et al. examined 21 studies involving 839 adults with BMIs from 22 to 43 kg/m². Its rankings placed time-restricted feeding (TRF) highest for TNF-alpha reduction, the 5:2 pattern highest for CRP, and alternate-day fasting in between.

These rankings generate hypotheses, but they do not prove that meal timing steers inflammation through distinct pathways. The network included two non-randomized controlled trials and participants ranging from lean to severely obese. That wider population differs from the overweight-and-obesity samples in the CR and keto reviews. A P-score ranks interventions within a network; it does not tell a reader how large or clinically meaningful the difference will be for them.

Circadian alignment of daily fasting windows and the more severe twice-weekly restriction of 5:2 are possible explanations for subtype differences. The meta-analysis did not test those mechanisms directly. Treating a ranking as a mechanism would be like treating a restaurant review as a recipe.

Equalize weight loss, and the diet differences fade

The most consequential evidence comes from studies designed to reduce the usual confounder: different amounts of weight loss. A 2025 meta-analysis by Hamsho et al. pooled randomized trials comparing isocaloric IF and CR-approaches matched for total calorie intake. It found no significant IF advantage over CR.

A 2025 exploratory sub-study by Turner et al. compared IF plus early time-restricted eating with CR after matching participants by achieved weight loss. Once fat-mass reduction was equalized, systemic and adipose-tissue inflammatory markers did not differ significantly between strategies.

Duration does not rescue a clear timing advantage. In a 12-month randomized trial by Lin et al. in adults with obesity, time-restricted eating and daily CR produced statistically equivalent reductions in TNF-alpha and IL-6 at 12 months.

These studies do not prove dietary format has zero independent effect. They do make a stronger case that fat-mass reduction is a major, and likely dominant, driver of the anti-inflammatory changes seen across these approaches. The striking biomarker pattern in uncontrolled meta-analyses may partly reflect different rates or amounts of weight loss rather than distinct inflammatory targets.

What the evidence can-and cannot-tell you

CR has the deepest and most consistent evidence for CRP reduction: 15 randomized trials and I²=0% in Aamir et al.'s analysis. IF has a statistically significant TNF-alpha result, especially with TRF in the network ranking, but its direct evidence base is smaller and the result becomes less persuasive when weight loss is matched. Keto has an intriguing CRP signal, but only five trials in 218 people, with carbohydrate restriction tightly entangled with caloric deficit.

For inflammation-related biomarkers, sustained fat-mass loss currently matters more than whether you create the deficit by daily calorie restriction, fasting windows, or a ketogenic diet.

None of these reviews answers the durability question: whether biomarker improvements persist after weight stabilizes and an active deficit ends. They also measure surrogate markers, not inflammatory disease events or longevity. A lower CRP value can be informative, but it is not a longevity certificate printed by your eating schedule.

Choose the structure you can follow safely and consistently. Daily calorie restriction may suit someone who prefers regular meals; time-restricted eating may simplify decisions for someone who dislikes tracking; keto may fit someone who genuinely prefers its food pattern and can meet nutritional needs. Choosing one solely to "target" TNF-alpha, IL-6, or CRP runs ahead of the evidence. The studies support adherence and sustained weight management far better than biomarker micromanagement.