Founder's components: NFKB1/2, RELA/B, REL, IKBKB, NFKBIA, TNFAIP3. Second most important node in this slice. The literature is large, split, and confounded by weight loss. It is also almost entirely measured at the wrong layer.
THE LAYER PROBLEM, STATED FIRST
Every human fasting row on this node measures NF-κB OUTPUT (circulating IL-6, TNF-alpha, IL-1beta, CRP), not NF-κB. No human fasting study located measured NFKB1, NFKB2, RELA, RELB, REL, IKBKB, NFKBIA or TNFAIP3 protein, nuclear translocation, or IκB degradation. The only evidence touching an actual component is Chung et al. 2001, a review of calorie restriction in rodents reporting that NF-κB activation was suppressed by preventing IκBα and IκBβ dissociation, with no numbers retrieved and the species not stated in the abstract. A cytokine panel is not a measurement of this node.
DRY status
DRY: PRESENT, and the two dry-fast studies point in OPPOSITE directions.
- Al lami et al. 2024, 30-day dawn-to-dusk dry fast (no food and no water during daylight), adults with increased BMI, assessed within 2 weeks before, at week 4, and within 2 weeks after: multiple cytokines DECREASED. IL-1beta fell at week 4 (p = 0.045) and the decrease persisted after the fasting period (p = 0.024). Also significantly decreased: IL-15 (p = 0.014), IL-1 receptor antagonist (p = 0.041), macrophage-derived chemokine (p = 0.013), CXCL9 (p = 0.027), FGF2 (p = 0.010), IL-12p40 (p = 0.038), IL-22 (p = 0.025), TNF-alpha (p = 0.046). Body weight (p = 0.032), BMI (p = 0.028), hip (p = 0.007) and waist (p = 0.016 to 0.033) decreased, and resting heart rate decreased (p = 0.023). Baseline IL-8 correlated with BMI measures (r = 0.57, p = 0.042). Waist change correlated positively with IL-15, MDC, PDGF, soluble CD40L and VEGF-A, which ties the cytokine changes to the anthropometric changes.
- Papagiannopoulos-Vatopaidinos et al. 2020, 5-day ABSOLUTE dry fast, n=10: CRP INCREASED. Alongside increases in ADH, ACTH, cortisol (about +495 percent), urine osmolality, renin, angiotensin II, aldosterone, total antioxidant capacity, uric acid, albumin, erythrocytes, hematocrit, and noradrenaline and dopamine excretion.
The conflict is reported unresolved. A 30-day intermittent dry fast with nightly rehydration lowered multiple cytokines; a 5-day absolute dry fast with no rehydration raised CRP. The modalities differ in absoluteness, duration and rehydration, and no retrieved source reconciles them. Neither side is selected. Queries run for DRY on this node: the four DRY queries under node 1, plus "Ramadan fasting CRP IL-6 TNF-alpha meta-analysis inflammatory markers", "intermittent fasting NF-kB p65 human PBMC measured protein expression trial" (budget-stopped), "caloric restriction NF-kB signaling aging rodent p65 nuclear translocation reduced".
Disconfirming, nulls and contradictions first
- The trial-by-trial count of human randomized trials is overwhelmingly null on the pro-inflammatory cytokines. Mulas et al. 2023, reviewing 11 randomized trials (5 TRE, 6 IF-ADF): IL-6 measured in 5 trials, 0 decreased, 5 no change. TNF-alpha measured in 4 trials, 0 decreased, 4 no change. CRP measured in 8 trials, 3 decreased (all three IF-ADF: Bowen, Razavi, Varady 2013) and 5 no change (4 TRE, 1 IF-ADF: Cho). Verbatim conclusion: "intermittent fasting has little or no effect on key circulating pro-inflammatory cytokines". [H1, grade B+]
- A 12-month randomized comparison found nothing on any of the three. Lin et al. 2025, TRE versus daily calorie restriction in adults with obesity: both arms lost roughly 4 to 5 percent of body weight, and neither produced a statistically significant change in TNF-alpha, IL-6 or CRP, within groups or between groups, at month 12. CRP was positively related to body weight, visceral fat mass and insulin resistance; IL-6 and TNF-alpha were related to no metabolic marker. [H1, grade B+]
- The human 36 hour transcriptome looks like a clean NF-κB suppression and does not survive correction. Elliott et al. 2014, n=10, PBMC: TNF -1.74, IL1B -2.04, CXCL2 -2.06, CCL3 -2.11, IER3 -1.95, NFKBIZ -1.57 at nominal p < 0.05, all NF-κB-associated. The authors state that under FDR < 5 percent, "no genes would be considered to be significantly affected by the extended fast". The most NF-κB-shaped human gene list in this slice is formally a null result. [H3, grade C]
- One prospective controlled Ramadan study found an INCREASE. Khoshandam Ghashang et al. 2021, n=50 healthy young males in summer in Germany, fasting group measured at four timepoints and non-fasting controls at two: IL-1beta increased significantly from T2 to T4 and from T3 to T4, and there were no significant differences between the fasting and non-fasting groups at either shared timepoint for any cytokine measured (IL-1beta, IL-6, IL-8, IL-10, IL-12, TNF-alpha). The between-group null is as important as the within-group rise. [RAMADAN, H2, grade C]
- A rodent 3-day fast found no cytokine change at all in the target tissue. Benak et al. 2024: IL-6 and TNF-alpha protein unchanged in rat left ventricle after 3 days of food deprivation with water. The authors used that null to argue that reduced basal inflammation does not explain the HIF-1α change they did see. [A1, grade C]
- UNRESOLVED CONFLICT AMONG THE META-ANALYSES, ALL SIDES NAMED.
- Khalafi et al. 2025 (21 studies, 839 participants): TNF-alpha SMD -0.31, p = 0.009; CRP SMD -0.19, p = 0.04; leptin SMD -0.57, p = 0.005. IL-6 and adiponectin NOT significant. Time-restricted feeding showed the largest TNF-alpha reduction (-0.39, p = 0.001); the 5:2 diet ranked highest for CRP. TRF held the highest probability ranking for IL-6, TNF-alpha, leptin and adiponectin, but the IL-6 and adiponectin effects were not statistically significant.
- Lü et al. 2025 (8 RCTs, 573 participants, metabolic syndrome): IL-6 reduced, WMD -0.58, 95 percent CI -1.08 to -0.08, p = 0.022. This directly contradicts Khalafi on IL-6.
- Mulas et al. 2023 (11 randomized trials, counted trial by trial): little or no effect on pro-inflammatory cytokines.
- Lin et al. 2025 (12-month randomized): null on all three. Three of these are 2023 to 2025 syntheses of overlapping literature and they do not agree with each other. The split is reported as the state of the evidence. No winner is picked.
- The weight-loss confound is named by the reviewers themselves, not by this atlas. Mulas et al. note that for CRP "it stands to reason that >6% weight loss may be necessary to see significant modulations in CRP levels with intermittent fasting", and that for TRE the weight loss "was not high enough to produce favorable modulations". Lin et al. found CRP tracked body weight, visceral fat and insulin resistance. Al lami et al. found cytokine changes correlating with waist change. Any claim that fasting lowers inflammation independently of weight loss is unsupported by everything retrieved here.
- The β-hydroxybutyrate and NLRP3 result is frequently recruited to this node and belongs to neither this node nor to fasting. Youm et al. 2015: BHB suppresses NLRP3 inflammasome activation by urate crystals, ATP and lipotoxic fatty acids; acetoacetate, butyrate and acetate do not; BHB does not affect the NLRC4, AIM2 or non-canonical caspase-11 inflammasomes; the mechanism is preventing K+ efflux and reducing ASC oligomerization and speck formation. In human monocytes BHB reduced NLRP3-mediated IL-1beta and IL-18 (C1). In mouse models of Muckle-Wells syndrome, familial cold autoinflammatory syndrome and urate crystal peritonitis, BHB or a ketogenic diet reduced caspase-1 activation and IL-1beta secretion (A1, modality KD). NO HUMAN FASTING INTERVENTION WAS RUN IN THAT STUDY. NLRP3 is also not an NF-κB component; it is a parallel inflammasome. Two separate category errors sit in the usual retelling. [C1 and A1, grade C]
Graded rows
| # | Modality | System | Tier | Grade | Finding | Source |
|---|---|---|---|---|---|---|
| 9.1 | TRE and IF-ADF | Human, 11 randomized trials | H1 | B+ | IL-6: 0 of 5 decreased. TNF-alpha: 0 of 4 decreased. CRP: 3 of 8 decreased, all three IF-ADF. OUTPUT LAYER. | Mulas et al. 2023, Front Nutr 10:1146924, DOI 10.3389/fnut.2023.1146924 |
| 9.2 | TRE vs CR, 12 months | Human, obesity, randomized, secondary analysis | H1 | B+ | No significant change in TNF-alpha, IL-6 or CRP within or between groups at 12 months, despite 4 to 5 percent weight loss in both arms. CRP related to body weight, visceral fat and insulin resistance; IL-6 and TNF-alpha related to no metabolic marker. | Lin et al. 2025, Nutrients 17(7):1130, PMID 40218888, DOI 10.3390/nu17071130 |
| 9.3 | Mixed IF (pooled) | Human, 21 studies, 839 participants | H1 | B+ | TNF-alpha SMD -0.31 p=0.009; CRP SMD -0.19 p=0.04; leptin SMD -0.57 p=0.005. IL-6 and adiponectin NOT significant. TRF largest TNF-alpha reduction -0.39 p=0.001; 5:2 ranked highest for CRP. CIs not reported in the retrieved abstract. | Khalafi et al. 2025, Nutrients 17(15):2388, PMID 40805975, DOI 10.3390/nu17152388 |
| 9.4 | Mixed IF (pooled) | Human, 8 RCTs, 573 participants | H1 | B+ | IL-6 reduced, WMD -0.58, 95% CI -1.08 to -0.08, p = 0.022. CONTRADICTS ROW 9.3 ON IL-6. Also FBS -3.34 (-6.24 to -0.45) p=0.024; HbA1c -0.08 (-0.13 to -0.02) p=0.005; HOMA-IR -0.60 (-0.91 to -0.28) p<0.001; LDL-c -6.42 (-12.26 to -0.58) p=0.031. | Lü et al. 2025, J Health Popul Nutr 44(1):293, PMID 40826125, DOI 10.1186/s41043-025-01039-2 |
| 9.5 | DRY (30 d dawn-to-dusk) | Human, increased BMI, single-arm observational | H3 | C | IL-1beta down at week 4 (p=0.045), persisting after (p=0.024); TNF-alpha down (p=0.046); IL-15 (p=0.014), IL-1RA (p=0.041), MDC (p=0.013), CXCL9 (p=0.027), FGF2 (p=0.010), IL-12p40 (p=0.038), IL-22 (p=0.025) down. Weight, BMI, waist, hip and resting heart rate down. Cytokine changes correlated with waist change. n NOT STATED in the retrieved record. OUTPUT LAYER. | Al lami et al. 2024, Metabolism Open 21:100274, PMID 38455231, DOI 10.1016/j.metop.2024.100274 |
| 9.6 | DRY (5 d absolute) | Human, n=10 | H3 | C | CRP INCREASED. With increases in ADH, ACTH, cortisol (about +495 percent), urine osmolality, renin, angiotensin II, aldosterone, total antioxidant capacity, uric acid, albumin, erythrocytes, hematocrit, noradrenaline and dopamine excretion; decreases in body weight, waist, erythropoietin, adrenaline excretion and Na+ excretion. Vitamin C half-life about 4.8 days. CONTRADICTS ROW 9.5. | Papagiannopoulos-Vatopaidinos et al. 2020, Complement Med Res 27(4):242-251, PMID 31958788, DOI 10.1159/000505201 |
| 9.7 | RAMADAN (1 month) | Human, n=40 fasting vs n=28 matched controls | H2 | B | IL-6 and CRP significantly LOWER during Ramadan than at baseline, p < 0.001 each, with no change in controls. Homocysteine also lower. Total cholesterol, triglycerides and LDL unchanged. Absolute values not retrieved. | Aksungar et al. 2007, Ann Nutr Metab 51(1):88-95, PMID 17374948, DOI 10.1159/000100954 |
| 9.8 | RAMADAN (1 month, summer, Germany) | Human, n=50 healthy young males, prospective controlled | H2 | C | IL-1beta INCREASED significantly from T2 to T4 and T3 to T4 in the fasting group. NO significant difference between fasting and non-fasting groups at T1 or T3 for any cytokine measured. TNF-alpha correlated with body weight, skeletal muscle mass and fat-free mass at T3. Magnitudes and p-values paywalled. | Khoshandam Ghashang et al. 2021, J Complement Integr Med 18(2):425-431, DOI 10.1515/jcim-2019-0209 |
| 9.9 | RAMADAN | Human, n=68 Saudi adults with diabetes, pre vs during | H3 | C | Lower log IL-6 and lower CRP; higher log HDL-C; lower body weight, waist and hip, after adjustment for confounders. No effect sizes or p-values retrieved. | Aljahdali et al. 2024, Curr Diabetes Rev 20(9):e080124225329, PMID 38192134, DOI 10.2174/0115733998274064231114075657 |
| 9.10 | PF (36 h) | Human PBMC, n=10 | H3 | C | TNF -1.74, IL1B -2.04, CXCL2 -2.06, CCL3 -2.11, IER3 -1.95, NFKBIZ -1.57 at nominal p<0.05. NFKBIZ is the only NF-κB-family gene in the list. NO GENE SURVIVED FDR < 5 PERCENT by the authors' own statement. | Elliott et al. 2014, Genes Nutr 9:432, DOI 10.1007/s12263-014-0432-4 |
| 9.11 | PF (3 d, water allowed) | Rat left ventricle, male Wistar, n=5 to 13 | A1 | C | IL-6 and TNF-alpha protein UNCHANGED. Authors used this null to argue against reduced basal inflammation as an explanation for the HIF-1α fall they measured. | Benak et al. 2024, Front Physiol 15:1462014, DOI 10.3389/fphys.2024.1462014 |
| 9.12 | CR | Species not stated in abstract; MeSH lists mice | M | C | Age-related increases in NF-κB, IL-1beta, IL-6, TNF-alpha, COX-2 and inducible nitric oxide synthase attenuated by calorie restriction; NF-κB activation suppressed by preventing IκBα and IκBβ dissociation. THE ONLY EVIDENCE IN THIS NODE TOUCHING AN ACTUAL NF-κB COMPONENT. Review. No numbers, no IKK or p65 data retrieved. CR IS NOT FASTING AND IS LABELED. | Chung et al. 2001, Ann N Y Acad Sci 928(1):327-335, PMID 11795524, DOI 10.1111/j.1749-6632.2001.tb05662.x |
| 9.13 | KD (mouse) and none (human monocytes) | Human monocytes; mouse in vivo disease models | C1 and A1 | C | BHB suppressed NLRP3 activation by urate crystals, ATP and lipotoxic fatty acids; acetoacetate, butyrate and acetate did not; no effect on NLRC4, AIM2 or caspase-11; mechanism via preventing K+ efflux and reducing ASC oligomerization and speck formation. Reduced IL-1beta and IL-18 in human monocytes; BHB or ketogenic diet reduced caspase-1 activation and IL-1beta in mouse models. NO HUMAN FASTING INTERVENTION. NLRP3 IS NOT AN NF-κB COMPONENT. BHB concentrations not retrieved. | Youm et al. 2015, Nat Med 21(3):263-269, PMID 25686106, DOI 10.1038/nm.3804 |
Open questions
- The node itself is unmeasured in human fasting. Closing it requires a p65 nuclear translocation or IκBα degradation measurement in a human fasting trial. None was located.
- The IL-6 disagreement between Khalafi et al. 2025 and Lü et al. 2025 is unexplained. Population differs (general adults versus metabolic syndrome) and trial counts differ (21 versus 8), which are candidate explanations and not retrieved findings.
- The dry-fast direction conflict (rows 9.5 and 9.6) is the most consequential open item on this node for dryfasting.com, because the two studies disagree and one of them is the only absolute dry fast measured.
← 8. TGF-β / SMAD All 41 nodes 10. HIF / hypoxia →
dryfasting.com has been published since 2014. This evidence record was compiled on 22 November 2026. Every row names the fasting modality the study actually ran, the system it was measured in and the grade. Where no source carries a DOI or PMID the row says so rather than supplying one. Nothing is sold on this site.