Founder's key components: NFE2L2 (NRF2), KEAP1, CUL3, antioxidant-response-element genes.
DRY status: none located. Queries as listed in the DRY column. No NRF2, KEAP1 or ARE-gene measurement under any no-water protocol located.
DISCONFIRMING FIRST
1. The widely cited "fasting induces NRF2 in mouse AND human" paper did not fast a single human. [ROW A1 / PF / B] Kulkarni SR, Donepudi AC, Xu J, Wei W, Cheng QC, Driscoll MV, Johnson DA, Johnson JA, Li X, Slitt AL. "Fasting induces nuclear factor E2-related factor 2 and ATP-binding cassette transporters via protein kinase A and sirtuin-1 in mouse and human." Antioxidants and Redox Signaling 2014;20(1):15-30, DOI 10.1089/ars.2012.5082, PMID 23725046.
- Mouse in vivo: fasting increased NRF2 target gene expression in intact mouse liver, roughly 1.5 to 5-fold.
- Fasting-induced NRF2 target induction was reduced in liver and hepatocytes from SIRT1 liver-specific null mice and from NRF2-null mice, which is a genuine causal test and the strongest row on this node.
- The human component was hepatocytes treated with the cAMP analogue 8-Br-cAMP, with the PKA inhibitor H-89 reducing the effect, plus promoter-binding analysis of the human ABCC2 promoter. There is no human in vivo work in this paper. The retrieved record does not specify primary human cells versus a cell line, which is a retrieval debt, and does not state the fasting duration, which is a second one.
- Reading: the title says "in mouse and human". The human half is pharmacology in cells. Any surface that cites this paper as human fasting evidence for NRF2 is citing it wrong, and that is the specific exposure this node creates.
2. The only human NRF2 row is an uncontrolled single-arm cohort whose own internal correlations point at intake, not abstinence, and whose sirtuin results run the other way. [ROW H3 / RAMADAN / B] Madkour MI, El-Serafi AT, Jahrami HA, Sherif NM, Hassan RE, Awadallah S, Faris MAIE. "Ramadan diurnal intermittent fasting modulates SOD2, TFAM, Nrf2, and sirtuins (SIRT1, SIRT3) gene expressions in subjects with overweight and obesity." Diabetes Research and Clinical Practice 2019;155:107801, DOI 10.1016/j.diabres.2019.107801. PMID not in the retrieved record.
- n = 56 overweight or obese, non-diabetic (34 male, 22 female), mean age 35.72 +/- 12.35, BMI 30.74 +/- 3.60, plus 6 healthy-weight controls (age 29.8 +/- 14.0, BMI 21.4 +/- 2.20). June to July 2016.
- Modality: Ramadan diurnal fasting, no oral intake including water from dawn to sunset, about 15 h per day, baseline one week before Ramadan, follow-up after 28 to 30 consecutive fasting days.
- NRF2 reported up 411.5 percent. SOD2 up 54.1 percent. TFAM up 90.5 percent. Hop if read as a ratio: 1 + 4.115 = 5.1-fold; this conversion is arithmetic performed here, not a reported value, and the paper reports no fold change.
- No P value is given for NRF2, SOD2, TFAM or SIRT1 in the retrieved record.
- The retrieved record states the NRF2, SOD2 and TFAM percentages are framed as obese subjects versus controls, so they may not be within-subject change. With 56 in one group and 6 in the other, a between-group framing would be a very different and much weaker claim than a pre-post one. This ambiguity is unresolved and the number must not be published as a within-subject fasting effect until the full text settles it.
- In the same subjects, SIRT3 fell 61.8 percent, P < 0.001, and SIRT1 fell 10.4 percent as a non-significant trend. The "fasting switches on the redox and longevity program" reading is contradicted inside the same cohort by the largest and only clearly significant effect in the paper, which is a decrease.
- Energy intake above 2000 kcal/day correlated positively with SOD2 (r = 0.84) and TFAM (r = 0.90), P < 0.05. The antioxidant gene signal tracked INTAKE. For a Ramadan cohort, where the total daily intake is consumed at night, this points at the feeding, not at the abstinence.
- Tissue sampled is not stated in the retrieved record, which is a retrieval debt for a gene-expression study.
- Design: single-arm pre-post observational, no randomization, no control arm matched for the Ramadan confounders of fluid restriction, shifted sleep, shifted meal timing and changed diet composition.
3. A human 6-day water-only fast study that was in a position to measure redox did not measure it. [ROW M / PF / C] Fang et al. 2023, Immunity & Ageing 20:44 (full citation under node 15). 31 humans, 6 days water-only, red blood cell proteomics with 78 differentially expressed proteins at day 6. Oxidative stress, NRF2 and antioxidant enzymes (SOD, catalase, GPx) are not mentioned in the retrieved content. Recorded as a stated absence, not as a null.
Supporting evidence
[ROW A1 / IF-ADF / B] Hemead HM, Hamed MF, Saleh AA, El-Sayed YS. Reprod Fertil Dev 2025;37:RD24187, DOI 10.1071/RD24187. Male rats on a high-fat diet: IF upregulated SIRT-1/NRF2 in testicular tissue, downregulated NLRP3 and p38 MAPK, lowered malondialdehyde and myeloperoxidase, raised glutathione and catalase. Findings and reference string READ from La Vignera and Condorelli 2026, article not opened, so magnitudes and P values are a retrieval debt.
[ROW A1 / KD + IF-ADF / C] Üstündağ ÜV, Özer A, Koca HB, Koca O, Karaman Mİ, Güneş M. "Exploring the Impact of Ketogenic Diet and Intermittent Fasting on Male Rats' Testicular Health." J Food Biochem 2023;2023:5562120, DOI 10.1155/2023/5562120. Male rats, cited for antioxidant effects. Reference string only. KD is labeled clearly and is not fasting.
Cross-reference, not a separate row. From Kulkarni 2014, already counted above: the NRF2-null and SIRT1-liver-null mouse results are the only causal necessity test on this node in any species.
Conflict left open
Mouse liver shows a 1.5 to 5-fold NRF2 target induction with fasting that requires SIRT1 and NRF2 (Kulkarni 2014). The single human cohort reports NRF2 up with SIRT1 and SIRT3 down (Madkour 2019). The mouse mechanism runs through SIRT1; the human cohort's sirtuins went the wrong way. Reported unresolved.
Node 17 verdict, stated plainly because the founder's catalogue leans on this node
Grade: B. Best tier H3, and that H3 row is a single-arm observational Ramadan cohort with no P value for NRF2 in the retrieved record, an unresolved between-versus-within-subject ambiguity, an unstated tissue, and an internal correlation pointing at intake. The animal evidence is better than the human evidence and includes a knockout necessity test.
- Supportable: fasting raises NRF2 target gene expression in mouse liver, and the effect requires SIRT1 and NRF2. Intermittent fasting raises SIRT-1/NRF2 signaling in rat testis.
- Supportable with the caveats attached: one human Ramadan cohort reported higher NFE2L2 expression after a month of dawn-to-dusk fasting.
- Not supportable: that fasting activates NRF2 in humans. Not supportable: that Kulkarni 2014 is human fasting evidence.
- Separate point, stated because it protects the catalogue: botanical NRF2 activation is a different literature with different evidence, and nothing in this node transfers to it in either direction. A blend's NRF2 claim must stand on the blend's own evidence, not on fasting's.
Open questions
1. Open the Madkour 2019 full text and settle whether the +411.5 percent NRF2 figure is within-subject or obese-versus-control, and name the tissue. This single question decides whether a human NRF2 row exists at all. 2. Open "Modulation of Anti-Oxidant, Anti-Inflammatory and Metabolism-Controlling Genes Expressions by Ramadan Intermittent Fasting: A Prospective Observational Study", locator PMC7258531. It appears to be the same research group and possibly the same cohort, which would mean one stream, not two. Retrieval debt. 3. Has any human fasting study measured KEAP1 or CUL3? Not located. 4. Has any human prolonged fast measured oxidative-stress end products (8-isoprostane, MDA, protein carbonyls) alongside NRF2? Not located in this sweep. If such data show oxidative stress rising while NRF2 transcripts rise, both can be true and the node's direction becomes a compensation story rather than a benefit story.
← 16. Autophagy All 41 nodes 18. Ferroptosis →
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.