Founder's key components: GPX4, SLC7A11 (system xc-), FSP1 (AIFM2), glutathione; ACSL4 drives the lipid peroxidation these restrain.
DRY status: none located. Queries as listed in the DRY column.
DISCONFIRMING FIRST, and for this node the disconfirming evidence is decisive
1. The field's own 2024 review does not connect fasting to ferroptosis at all. [ROW C1 / no fasting modality tested / C] Berndt C, Alborzinia H, Amen VS, and others, corresponding author Conrad M. "Ferroptosis in health and disease." Redox Biology 2024;75:103211, DOI 10.1016/j.redox.2024.103211, online 30 May 2024. Author list abbreviated in the retrieved record. Two 100,000-character portions were read, characters 0 to 200,000 of 549,681.
- The words "fasting" and "caloric restriction" do not occur in either portion. "Nutrient deprivation" does not occur. "Cysteine deprivation" does not occur.
- "Starvation" occurs three times, all at the cell level: glucose starvation, cysteine starvation, selenium starvation.
- No human in vivo evidence of ferroptosis with a named validated marker is reported. Patient-level observations that are reported are presented as findings rather than validated biomarkers: iron, MDA and decreased GSH in ulcerative colitis colonic tissue; increased 4-HNE immunoreactivity in Crohn's disease intestinal epithelial cells; increased lipid peroxidation in patients with selenoprotein synthesis defects. A set of 36 ferroptosis-linked lncRNAs is described as a potential prognostic tool.
- Diet is discussed only as dietary vitamin E content in mice, where it masks or unmasks Gpx4 knockout phenotypes, and where a vitamin-E-deprived diet causes widespread endothelial cell death. That is a micronutrient effect in a knockout, not a fasting effect.
- Reading with the denominator stated: in 200,000 characters of a 2024 field review covering ferroptosis in health and disease, fasting is absent. If there were a fasting-to-ferroptosis literature worth grading, this is where it would appear.
2. The one mechanistic link that IS in the review runs in the opposite direction to the popular story. [ROW C1 / no fasting modality tested / C] From Berndt 2024: glucose starvation activates AMPK, which suppresses synthesis of certain polyunsaturated fatty acids and INHIBITS ferroptosis by phosphorylating and inactivating acetyl-CoA carboxylase. System not specified beyond general cell studies.
- Energy stress, on this mechanism, is ferroptosis-PROTECTIVE. A reader who assumes fasting promotes ferroptotic clearance of damaged cells has the sign backwards on the one energy-related mechanism the review states.
- Also from the review: cysteine starvation triggers CHAC1 through the ATF4 amino-acid starvation response, which hydrolyzes glutathione, and likely depletes GSH faster than it inhibits GSH synthesis. Under selenium starvation, selenium is preferentially directed to GPX4 and TrxR via higher SBP2 affinity. Both are amino-acid and micronutrient mechanisms in cells. Neither is a fasting intervention.
- Conflict reported unresolved: cysteine starvation sensitizes to ferroptosis while glucose starvation protects against it. Human fasting is neither of those two conditions in isolation, and no source located resolves which dominates in a fasting human.
3. Human iron data under fasting exist, are internally contradictory, and are not ferroptosis. [ROW H3 / TRE + IF-ADF / C] La Vignera S, Condorelli R. "Effects of the Mediterranean Diet, Ketogenic Diet, and Intermittent Fasting on Iron Metabolism and Iron Deficiency Risk in Adults with Obesity: A Systematic Literature Review." Nutrients 2026;18(16):2681, DOI 10.3390/nu18162681, published 17 August 2026.
- Table 1, intermittent fasting column, 7 studies reporting, ranges only: ferritin baseline 78 to 108 microgram/L with a post-intervention change of -12 to -22 percent; hemoglobin baseline 13.0 to 14.1 g/dL with change not exceeding +/-0.3 g/dL; IL-6 baseline 2.9 to 5.8 pg/mL with change -22 to -38 percent; hepcidin baseline 48 to 88 ng/mL with change -15 to -28 percent.
- Direct internal contradiction in the same paper: the body text says hepcidin is ELEVATED during acute fasting and reduced only after significant weight loss, while Table 1 reports hepcidin DOWN 15 to 28 percent for intermittent fasting. The paper attributes the discrepancy to timing and does not reconcile it. Reported unresolved.
- The body text also reports serum iron reduced during acute short-term fasting (a hypoferremic response) and transferrin increased acutely, both within physiological range, and warns that acute hypoferremia measured during a fast may be misread as iron deficiency.
- Provenance weakness, measured: the human IF iron claims are cited to reference 16, which the retrieved record identifies as de Cabo and Mattson 2019, a review, not an IF trial with iron outcomes. Teng et al. 2021 is cited for IF weight-loss effects but its title describes diet-induced weight loss and bariatric surgery. The specific trials behind the iron numbers cannot be identified from the retrieved text.
- Ferroptosis is not mentioned in the retrieved portion of this paper.
- Reading: iron handling changes under fasting in humans. Iron handling is not ferroptosis. Ferritin, hepcidin and transferrin saturation are not markers of iron-dependent lipid-peroxide cell death, and using them that way would be exactly the substitution this atlas exists to prevent.
Located, not opened, and therefore ungraded
- Liu X, Peng S, Tang G, Xu G, Xie Y, Shen D, Zhu M, Huang Y, Wang X, Yu H, Huang M, Luo Y. "Fasting-mimicking diet synergizes with ferroptosis against quiescent, chemotherapy-resistant cells." eBioMedicine 2023, article 104496, DOI 10.1016/j.ebiom.2023.104496. Citation confirmed from a Crossref-backed index. The abstract was not retrieved and the experimental system is unknown: cell culture, mouse, or both. linkinghub.elsevier.com returned a redirect shell and sciencedirect.com refused. Therefore: no tier, no grade, no direction, no claim. This is the single paper most likely to be cited as "fasting plus ferroptosis", and this atlas currently cannot say what it measured. Highest-priority retrieval debt on this node.
- "Intermittent fasting ameliorates neuronal ferroptosis and cognitive impairment in mice after traumatic brain injury", Nutrition (article S0899900723000229 on the publisher index). Mouse by its title. sciencedirect.com refused. Retrieval debt. Note the direction implied by the title is ferroptosis-REDUCING, which is a third direction alongside the two in conflict above.
Node 18 verdict
Grade: C. Best tier C1. There is no A1 row and no human row.
- No human fasting study measuring any ferroptosis endpoint was located. Not GPX4, not SLC7A11, not FSP1/AIFM2, not glutathione in a cell-death context, not ACSL4, not a lipid-peroxide measurement interpreted as ferroptosis.
- The cell-level mechanisms that touch nutrient supply point in two opposite directions (cysteine starvation sensitizing, glucose starvation protecting) and a located-not-opened mouse title points in a third (IF reducing neuronal ferroptosis after injury).
- This node must not read as a human fasting effect, and it must not be written as a direction at all until the conflict is resolved. The defensible public sentence is that ferroptosis is a recent and rapidly moving field, that its validated in vivo biomarkers are not settled even in its own 2024 field review, and that fasting does not appear in that review.
Open questions
1. Open the eBioMedicine 2023 paper and establish its system. Until then it is uncitable here. 2. Does any human fasting cohort have stored samples in which 4-HNE, GPX4 activity or an oxidized-lipid panel could be measured? Not located. 3. Which dominates in a fasting human: the cysteine-starvation sensitization or the AMPK-mediated protection? Unresolved and currently unresolvable from the located sources. 4. Does plasma cysteine actually fall in a fasting human, which is the precondition for the sensitizing mechanism to apply at all? Not measured in any source opened in this run.
# RECEIPT BLOCK
Nodes
Nodes done: 8 of 8 (11, 12, 13, 14, 15, 16, 17, 18). Zero deferred.
Node grades
| Node | Display name | Grade | Best tier | Reads as a human fasting effect? |
|---|---|---|---|---|
| 11 | Estrogen receptor (ER) | C receptor / B ligand axis | M receptor / H1 ligand | No at receptor. Ligand only, conflicting. |
| 12 | Androgen receptor (AR) | C receptor / B+ ligand axis | M receptor / H1 ligand | No at receptor. Ligand yes, with two large nulls. |
| 13 | Cell cycle (G1→S control) | C | A1 | No. Mouse. |
| 14 | p53 network | C, zero opened rows | M | No. Nothing opened in any species. |
| 15 | Apoptosis (intrinsic and extrinsic) | C | H3, and both H3 rows are null | No. Two human nulls. |
| 16 | Autophagy | B markers / C flux | H1 (exploratory, post hoc) | Markers yes. Flux no. |
| 17 | NRF2 / KEAP1 | B | H3, single-arm observational | No. Mouse and rat carry it. |
| 18 | Ferroptosis | C | C1 | No. No human row exists. |
Rows by tier
Computed mechanically from the ROW tags in this file, not estimated. Total in-node ROW tags: 42. Reproducible check: a raw grep for [ROW returns 43 matches. One of those is the format legend in the "How to read this" section at the top of the file and is not a row. 43 minus 1 equals 42. Any future count that does not subtract the legend will be one high.
- H1: 6 (node 11 x1, node 12 x2, node 16 x3)
- H2: 2 (node 11 x1, node 12 x1)
- H3: 17 (node 11 x4, node 12 x3, node 13 x1, node 15 x3, node 16 x4, node 17 x1, node 18 x1)
- A1: 11 (node 11 x1, node 12 x4, node 13 x1, node 15 x2, node 17 x3)
- A2: 1 (node 16 x1)
- C1: 2 (node 18 x2)
- M: 3 (node 13 x1, node 14 x1, node 17 x1)
Rows per node: 11 has 7, 12 has 10, 13 has 3, 14 has 1, 15 has 5, 16 has 8, 17 has 5, 18 has 3.
Two of the 42 tagged rows carry no evidence weight and are excluded from any claim:
- node 16,
A2 / not a fasting study / not graded: the 2016 Nobel record, present only to fix the species. - node 16,
H1 / PF / no result: the PROFASTA protocol, which has a registered design and zero results.
Evidence-bearing rows: 40.
Three rows were re-tiered during the census because their first tag put a human measurement against a node component that study did not measure. All three are now M, per the brief's definition that a link inferred from a shared upstream node is mechanism: node 13 (Moro IGF-1, measured H1 in humans but not a G1/S component), node 14 (Vendelbo mTOR/ULK1, fasting tested but p53 not measured), node 17 (Fang 6-day water-only fast, fasting tested but redox not measured). One line was also stripped of its ROW tag because it was a cross-reference to the Kulkarni 2014 row rather than a second row, which had double-counted one A1 under node 17.
Tier distribution read plainly: the human rows in this slice are overwhelmingly H3, 17 of 40, and the H1 count of 6 includes 3 node-16 rows of which one is a protocol with no results and one is an exploratory post hoc analysis. No node in this slice has an A grade, and no node has two independent human trials converging.
Languages
| Language | Status | Query |
|---|---|---|
| English | Productive | 14 distinct queries across the 8 nodes (autophagy markers and flux, Klionsky guidelines, EM in human liver and gut, Ramadan sex hormones, NRF2 human and mouse, ferroptosis and FMD, iron and hepcidin, p53 and fasting, post-fast refeeding, TRE and testosterone, human lymphocyte apoptosis) |
| Russian | Productive channel, empty literature | сухое голодание аутофагия Филонов исследование человек. Returns were books and popular pages only (Filonov, Сухое лечебное голодание: мифы и реальность), zero marker measurements, zero peer-reviewed human studies |
| Chinese | NOT RUN, channel blocked | planned: 禁食 自噬 人体 LC3 研究 |
| Japanese | NOT RUN, channel blocked | planned: 断食 オートファジー ヒト 研究 大隅 |
| German | NOT RUN, channel blocked | planned: Fasten Autophagie Mensch Studie Buchinger |
| French | NOT RUN, channel blocked | planned: jeûne autophagie humain étude marqueurs |
| Arabic | NOT RUN, channel blocked | planned: الصيام الالتهام الذاتي الإنسان دراسة |
| Persian | NOT RUN, channel blocked | planned: روزه اتوفاژی انسان مطالعه |
| Turkish | NOT RUN, channel blocked | planned: oruç otofaji insan çalışma |
| Spanish | NOT RUN, channel blocked | planned: ayuno autofagia humanos estudio marcadores |
| Korean | NOT RUN, channel blocked | planned: 단식 자가포식 인간 연구 |
Languages: 2 of 11 queried. 9 of 11 not run. Cause recorded below. These 9 are reported as BLOCKED, not as EMPTY. An unrun query is not a result and is not written as one anywhere in this file.
Channels blocked, each with its reason, none retried after its first refusal
| Channel | Reason | When |
|---|---|---|
| eutils / NCBI E-utilities | blocked from this environment per the brief | before start |
| Europe PMC REST (www.ebi.ac.uk) | agent proxy connect_rejected, HTTP 403 | probe |
| pubmed.ncbi.nlm.nih.gov | proxy connect_rejected 403 via curl; reCAPTCHA interstitial via WebFetch | probe and 1 fetch |
| pmc.ncbi.nlm.nih.gov | reCAPTCHA "Checking your browser" interstitial | 1 fetch |
| api.openalex.org | proxy, no response (000) | probe |
| api.crossref.org | proxy, no response (000) | probe |
| api.semanticscholar.org | proxy, no response (000) | probe |
| europepmc.org (website) | site client_error, nothing retrieved | 1 fetch |
| doaj.org | proxy, no response (000) | probe |
| journals.physiology.org | site client_error on PDF | 1 fetch |
| onlinelibrary.wiley.com | site client_error | 1 fetch |
| www.cell.com | site client_error | 1 fetch |
| linkinghub.elsevier.com | returned a redirect shell with no article content | 1 fetch |
| www.sciencedirect.com | productive once, then client_error; treated as blocked thereafter | 2 fetches |
| clinicaltrials.gov | returned the site template only, no study record (JS-rendered) | 2 fetches |
| aminer.org | returned site metadata only | 1 fetch |
| WebSearch (all languages) | shared per-turn budget of 200 calls exhausted by the agents in this turn. This is the cause of the 9 unrun language channels and of 3 unrun node queries. | mid-run |
Channels productive: www.mdpi.com (11 fetches including offset reads), journals.plos.org (2), link.springer.com (3), www.nature.com (1), www.nobelprize.org (2), pure.johnshopkins.edu (1), academic.oup.com (1), dbpia.nl.go.kr (1), digital.library.adelaide.edu.au (1), ouci.dntb.gov.ua (1), stockwelllab.fas.columbia.edu (2), doi.org as a redirect resolver (2), www.sciencedirect.com (1 of 2).
Sources opened: 20 distinct sources. Used as the denominator for the DRY absence statement.
Retrieval debts, explicit
Sources located but not opened, or opened but incomplete. None of these carries a grade anywhere above. 1. eBioMedicine 2023, DOI 10.1016/j.ebiom.2023.104496, Liu et al, FMD and ferroptosis. Citation confirmed, abstract and experimental system unknown. Highest priority; node 18 cannot be graded properly without it. 2. Bensalem et al. 2021, "Measurement of autophagic flux in humans: an optimized method for blood samples", Autophagy. The method underneath both human flux datasets. Not opened. 3. J Physiol 2025;603(10):3019-3032, Bensalem et al. DOI and PMID not retrieved; only the repository record was opened, so effect sizes and confidence intervals beyond P = 0.04 are missing. 4. "Training state and skeletal muscle autophagy in response to 36 h of fasting", J Appl Physiol 2018. Human, directly on node 16. Host refused. 5. "Muscle fiber type-specific autophagy responses following an overnight fast and mixed meal ingestion in human skeletal muscle", Am J Physiol Endocrinol Metab 2022. Human, node 16. Host refused. 6. "Increasing autophagy with fasting and diet: separating fact from fiction", locator PMC13673093. Closest existing review to node 16's question. Host captcha. 7. Jamshed 2019 Table S1, the full gene list, including whether SQSTM1/p62 was measured. 8. Madkour 2019 full text: tissue sampled, P values for NRF2/SOD2/TFAM, and whether the +411.5 percent is within-subject or obese-versus-control. This one question decides node 17's human row. 9. "Modulation of Anti-Oxidant, Anti-Inflammatory and Metabolism-Controlling Genes Expressions by Ramadan Intermittent Fasting", locator PMC7258531. Possibly the same cohort as Madkour 2019, which would make it one stream and not two. 10. Cell Metab 2022, S1550-4131(22)00182-6, fasting-induced deep quiescence in muscle stem cells via ketone bodies. Node 13. 11. "p21Cip1 plays a critical role in the physiological adaptation to fasting through activation of PPARalpha", locator PMC5056372. Nodes 13 and 14. 12. "Hepatic p53 is regulated by transcription factor FOXO1 and acutely controls glycogen homeostasis", locator PMC9399478. Node 14, the only located p53-and-fasting paper. 13. Imada 2024 Nature: actual tumor incidence numbers in refed versus fasted versus fed mice. 14. Ito et al. 2010 AJP-GI 298:G916-G926: direction and magnitude of fasting-induced rat intestinal apoptosis beyond the title claim. 15. Hemead 2025 Reprod Fertil Dev 37:RD24187: magnitudes and P values for the rat NRF2 result. 16. Cameron 1991: whether water was permitted during the 48 h fast, and the P values. 17. Rautou et al. 2008 Gastroenterology S0016-5085(08)00903-7; Kheloufi 2014 DOI 10.1155/2014/701064; Restellini 2013 DOI 10.1155/2013/749169: human hepatocyte autophagy in anorexia nervosa. 18. PROFASTA full author list (truncated on the page) and Kröpfl 2022 full author list (truncated). 19. La Vignera 2026 iron review: reference strings for refs 16 and 24, and the per-study iron values behind Table 1's ranges. 20. Nine language channels listed in the language table. 21. Three unrun node queries: English DRY query (all nodes), p53-and-fasting second construction (node 14), human cell-cycle marker query (node 13).
Citation conflict found and recorded
Poursalehian et al. 2024, PLoS ONE 19(5):e0299695, DOI 10.1371/journal.pone.0299695. Opening the article directly gives the authors as Poursalehian M, Mohseni S, Shadman Z, Mohajeri-Tehrani M, Atlasi R, Khoshniat Nikoo M, Qorbani M, Larijani B. The reference string for the same DOI inside La Vignera and Condorelli 2026 gives Poursalehian M, Soltanieh S, Barzegar A, Hassanzadeh Keshteli A, Esmaillzadeh A, Adibi P. Same DOI, two different author lists. The publisher record is treated as authoritative here and the citing review's reference string is recorded as wrong.
Style compliance
Punctuation: not one long dash in the file. American spelling throughout, except inside retrieved titles and journal names, which are reproduced as published and are never respelled. Nothing is written as an excuse. No oncology treatment claim and no claim that anything heals or eliminates any disease: the one tumor finding in this file (Imada 2024) is a mouse risk observation in an Apc-deletion model, reported as a risk and as a caution, with its genotype and species named, and no therapeutic claim of any kind is made from it. No reader is advised to fast or not to fast anywhere in this file. This is a record.
← 17. NRF2 / KEAP1 (oxidative-stress response) All 41 nodes 19. Mismatch repair (MMR) →
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.