Dry Fasting — The Record

Pathway atlas · node 10 of 41

HIF / hypoxia

What fasting does to this node, by modality, by system and by grade. Disconfirming evidence is reported first.

Founder's components: HIF1A, EPAS1 (HIF2α), ARNT, VHL, EGLN/PHD.

DRY status

DRY: PRESENT at the output layer, and the direction is DOWN. Papagiannopoulos-Vatopaidinos et al. 2020, 5-day absolute dry fast, n=10: ERYTHROPOIETIN DECREASED. Erythropoietin is the canonical HIF target gene product. Hematocrit and erythrocyte count rose in the same study, which is consistent with hemoconcentration from fluid loss rather than with erythropoietic drive. No HIF1A, EPAS1, ARNT, VHL or EGLN/PHD was measured in that study or in any dry-fast study located. Papagiannopoulos et al. 2013 (5-day food and water deprivation, n=10) measured hemoglobin oxygen saturation and found it stable, and measured no HIF component. Koppold et al. 2021 and Mähler et al. measured none. This is the only DRY row in the whole slice that lands on a node output with a stated direction, and it points the opposite way from the intuitive reading that water deprivation drives hypoxia signaling. Queries run: the four DRY queries under node 1, plus "fasting HIF-1alpha stabilization hypoxia signaling intermittent fasting mouse", "Ramadan fasting erythropoietin HIF-1alpha VEGF human serum study", "fasting erythropoietin humans serum increase decrease study hypoxia inducible factor human".

Disconfirming, nulls and contradictions first

Graded rows

#ModalitySystemTierGradeFindingSource
10.1PF (3 d, water allowed)Rat left ventricle, male Wistar, n=5 proteomics, 5 to 13 other assaysA1CHIF-1α mRNA -23 percent, protein -28 percent. Egln3 mRNA -46 percent with PHD3 protein unchanged. HIF-1 KEGG proteins altered: SERPINE1 up, CDKN1B up, TF down, HK2 down, TFRC down. Targets by RT-qPCR: Pdk4 +722 percent, Hmox1 +138 percent, Hk2 -63 percent, Gja1 -16 percent; no change in Adora2b, Gapdh, Nos2, Nox4, Vegfa. Western validation: PDK4 +571 percent, HK2 protein -16 percent. IL-6 and TNF-alpha protein unchanged. THE ONLY STUDY IN THIS SLICE THAT MEASURED A HIF COMPONENT UNDER FASTING, AND THE DIRECTION IS DOWN.Benak et al. 2024, Front Physiol 15:1462014, DOI 10.3389/fphys.2024.1462014
10.2DRY (5 d absolute)Human, n=10H3CERYTHROPOIETIN DECREASED. Hematocrit and erythrocytes increased, consistent with hemoconcentration. Output layer; no HIF component measured.Papagiannopoulos-Vatopaidinos et al. 2020, Complement Med Res 27(4):242-251, PMID 31958788, DOI 10.1159/000505201
10.3DRY (5 d food and water deprivation)Human, n=10, 4 men 6 women, mean age 47.5, mean BMI 28H3CHemoglobin oxygen saturation STABLE across 5 days. Blood pressure and heart rate stable. Glucose nadir 3.4 mmol/L on day 3 (p = 0.007). Osmolality rose to 302 mOsm/L by day 5. Urea 8.2 mmol/L, Na+ 148 mEq/L, Cl- 105 mEq/L on day 5. K+ peak mean 5.29 mEq/L on day 2, with one participant approaching 7 mEq/L. Creatinine rose slightly but significantly, peak mean 8 µmol/L day 5. Creatinine clearance 150 to 250 mL/min on day 4 (+67 percent), falling to 125 mL/min on day 8 (-17 percent). Weight -1,390 plus or minus 60 g/day. NO HIF OR OTHER SIGNALING COMPONENT MEASURED; the paper states a hormonal profile was not run. INTERNAL INCONSISTENCY: the abstract says creatinine clearance rose "up to 167%" while the results text gives a maximum of +67 percent. Both figures recorded, not reconciled.Papagiannopoulos et al. 2013, Forsch Komplementmed 20(6):427-433, PMID 24434757, DOI 10.1159/000357718
10.4none (no fasting tested)Mouse intestine, HIF-2α gain and loss of functionMCIntestinal HIF-2α enriched in L-cells with GPR40, potentiates fatty-acid-induced GLP-1 secretion through ERK. Overexpression raised GLP-1 and reduced diet-induced visceral adiposity, glucose intolerance and hepatic steatosis; disruption decreased GLP-1 secretion. NO FASTING INTERVENTION.Mooli et al. 2021, Cell Mol Gastroenterol Hepatol 13(4):1057-1072, PMID 34902628, DOI 10.1016/j.jcmgh.2021.12.004

Open questions


# RECEIPT BLOCK

Nodes done

10 of 10. Nodes 1, 2, 3, 4, 5, 6, 7, 8, 9, 10. None deferred.

Rows by tier

Counted directly off the numbered rows in the node tables above. Each row is counted once, at the strongest system that supports it.

TierCountRows
H1 human RCT, fasting as the intervention102.2, 2.3, 2.4, 3.1, 3.2, 3.3, 9.1, 9.2, 9.3, 9.4
H2 human controlled, crossover or prospective cohort33.6, 9.7, 9.8
H3 human single-arm, pre/post or ex vivo human tissue191.1, 1.2, 2.1, 2.5, 2.6, 2.7, 2.9, 2.10, 3.4, 3.5, 8.1, 8.2, 8.3, 9.5, 9.6, 9.9, 9.10, 10.2, 10.3
A1 mammalian in vivo, fasting as the intervention131.3, 2.8, 4.1, 4.2, 4.3, 5.3, 6.1, 6.2, 6.3, 7.1, 7.2, 9.11, 10.1
A2 non-mammalian in vivo14.4
C1 cell culture under nutrient or serum withdrawal19.13 (its human-monocyte arm; the same row also carries an A1 mouse arm, counted once at C1)
M mechanism only, no fasting intervention tested71.4, 3.7, 5.1, 5.2, 8.4, 9.12, 10.4

TOTAL ROWS = 54. 10 + 3 + 19 + 13 + 1 + 1 + 7 = 54. The sum is stated so it can be checked.

Rows by node: node 1 = 4, node 2 = 10, node 3 = 7, node 4 = 4, node 5 = 3, node 6 = 3, node 7 = 2, node 8 = 4, node 9 = 13, node 10 = 4. Sum = 54.

Nodes whose BEST available row is C1 or M, and which therefore must never be written as a human fasting effect: node 5 (Notch), best row M. Nodes whose best row is animal only: node 4 (Hedgehog) A1, node 6 (Wnt / beta-catenin) A1, node 7 (Hippo) A1. Nodes whose own components are unmeasured in human fasting but whose ligands or outputs carry human rows: nodes 1, 3, 8, 9 and 10. The only node with a human measurement of its own components under a fasting intervention is node 2, from a single study of 8 men at 72 hours.

Grades by node

NodeBest gradeBest tierNode components measured under fasting in humans?
1 RTK / RAS / MAP-kinaseCH3 (ligand) / A1No
2 PI3K / AKT / mTORB+ (ligand layer), B (node layer)H1 / H3YES, one study, n=8, 72 h
3 JAK / STATB+ for the IL-6 null; C for the nodeH1 (ligand) / M (node)No
4 HedgehogCA1No
5 NotchCMNo, and none in animals either
6 Wnt / β-cateninCA1No
7 HippoCA1No
8 TGF-β / SMADCH3 (null)Ligand only (myostatin), null
9 NF-κBB+ at the output layer, C at the node layerH1 / MNo
10 HIF / hypoxiaCA1 / H3 (output)No

No node in this slice reaches grade A. No node has multiple independent human trials converging on a measurement of the node's own components.

DRY column summary

NodeDRY status
1Ligand only: FGF2 decreased, p=0.010, 30-day dawn-to-dusk dry fast. No pathway measurement.
2NONE LOCATED, for any component and for IGF-1, IGFBP-1, IGFBP-3 or insulin.
3Ligand only: IL-15, IL-22, IL-12p40 decreased. No JAK, TYK2, STAT or SOCS measured.
4NONE LOCATED.
5NONE LOCATED.
6NONE LOCATED.
7NONE LOCATED.
8NONE LOCATED.
9PRESENT, output layer, TWO DRY STUDIES IN OPPOSITE DIRECTIONS (30-day intermittent dry fast lowered multiple cytokines; 5-day absolute dry fast raised CRP). Unresolved.
10PRESENT, output layer: erythropoietin DECREASED in a 5-day absolute dry fast. Direction is DOWN.

Four dry-fast human studies were opened in total: Papagiannopoulos et al. 2013 (5-day absolute food and water deprivation, n=10), Papagiannopoulos-Vatopaidinos et al. 2020 (5-day absolute dry fast, n=10), Koppold et al. 2021 (19-day daytime dry fast, n=34), Mähler et al. (daytime dry fast, n=11 men), plus Al lami et al. 2024 (30-day dawn-to-dusk dry fast, n not stated). Across all five, ZERO components of any of these ten nodes were measured. Every DRY row in this file is a ligand or an output. Ezpeleta et al. 2024, the Nutrition Reviews narrative review of prolonged water fasting that screened PubMed, Embase and Cochrane, contains NO human dry-fasting evidence and explicitly excludes religious fasting. That is the protective finding: the dry-fasting pathway literature does not exist at the node level, in any language channel swept.

Language channels

LanguageQueryVerdict
Englishall primary queries in this runPRODUCTIVE
Russianсухое голодание исследование Филонов метаболизм результатыEMPTY at witness level. Returned consumer and popular sources only (lifehacker.ru, books.yandex.ru, flibusta.su, vkusvill.ru, habr.com). No peer-reviewed primary research. This is the Filonov dry-fasting tradition language and it yielded no publisher article record.
Chinese禁食 间歇性断食 mTOR 信号通路 炎症 NF-κB 研究PRODUCTIVE as a channel, records NOT opened. Surfaced Chinese-language journal records: kyxuebao.kmmu.edu.cn article 7ffcd0cd-e48f-4292-a830-89d49a128d6a; xuebao.shsmu.edu.cn 2023;43(9):1131 (intermittent fasting plus thermogenic fat activation in obese mice); manu37.magtech.com.cn/yyxb 2025;47(1):99. Logged as retrieval debt D27.
Japanese断食 mTOR シグナル 研究 IGF-1 ヒト 臨床試験EMPTY for Japanese-language primary research. Returned English-language PMC and MDPI records plus one news article (newsweekjapan.jp).
GermanRamadan Fasten Studie Entzündungsmarker Zytokine Insulin IGF-1 ErgebnissePRODUCTIVE for German-institution work, EMPTY for German-language text. Surfaced Khoshandam Ghashang et al. 2021 (Hannover), Koppold et al. 2021 (Berlin) and Mähler et al. (Berlin), all published in English and all opened and used.
Frenchjeûne prolongé humain IGF-1 mTOR inflammation étude clinique résultatsEMPTY for French-language primary research. Returned French-language secondary coverage (lanutrition.fr, bionity.com/fr) of Pietzner et al. 2024, whose English primary was located and opened separately, plus academic.oup.com/edrv/article/46/6/856/8211151 which was not opened.
Arabicصيام رمضان دراسة الالتهاب السيتوكينات عامل النمو IGF-1 نتائجPRODUCTIVE. Surfaced Rahbar et al. 2019, which was opened and used as rows 2.6 and 3.5, plus jnfh.mums.ac.ir. All retrieved records are English-language.
Persianروزه داری رمضان مطالعه التهاب سیتوکین IGF-1 نتایجPRODUCTIVE as a channel, records NOT opened. Surfaced Persian-language journal PDFs: ijem.sbmu.ac.ir article-1-405-fa.pdf and article-1-754-fa.pdf; ijdld.tums.ac.ir article-1-138-fa.pdf; rjms.iums.ac.ir article-1-2715-en.pdf; sid.ir PDFs. Logged as retrieval debt D26.
TurkishRamazan orucu çalışma inflamasyon sitokin IGF-1 insülin sonuçlarPRODUCTIVE as a channel, records NOT opened and all below witness standard. Surfaced a Gazi University lecture PDF on "Ramazan açlığının moleküler mekanizması", theses at acikbilim.yok.gov.tr (20.500.12812/31408 and /108467) and nek.istanbul.edu.tr thesis 59858, plus dergipark.org.tr. Theses and lecture slides are not publisher article records. Logged as retrieval debt D28.
Spanish"ayuno intermitente" estudio humanos IGF-1 inflamación marcadores resultados ensayoEMPTY at witness level. Returned undergraduate theses (zaguan.unizar.es, uvadoc.uva.es, dspace.umh.es), consumer sites and secondary coverage (intramed.net, psiquiatria.com).
Korean간헐적 단식 연구 염증 지표 IGF-1 인슐린 임상시험 결과EMPTY at witness level. Returned English-language MDPI and PMC records plus Korean-language blog posts (devstarsj.github.io, wikidocs.net). Note that Choi et al. 2024 (rows 7.2) is Korean-institution work published in English and was located through the English channel.

Productive: 6 of 11 (English, Chinese, German, Arabic, Persian, Turkish). Empty at witness level: 5 of 11 (Russian, Japanese, French, Spanish, Korean). Not one non-English channel produced a node-level fasting measurement that the English channel had not already produced.

Channels blocked

ChannelReasonAction taken
pmc.ncbi.nlm.nih.govreCAPTCHA browser check on first fetchDead, never retried. Switched to api.openalex.org pmcid: path lookups, which supplied the same records.
pubmed.ncbi.nlm.nih.govMixed. PMID 24434757 returned a full record; PMID 28202779 returned an empty shell with no article content.Not retried after the first empty return. Switched to api.openalex.org.
www.nature.comclient_error on /articles/s41467-024-45260-9, nothing retrievedDead, never retried. link.springer.com served Nature-group content successfully instead (used for Yilmaz et al. 2012).
elifesciences.orgJavaScript gate on both the article path and the .pdf pathDead after the PDF attempt. No row written from it. Retrieval debt D25.
Bash curl to api.openalex.orgHTTP 000 on connect, and reading /root/.ccr/README.md for the proxy fix was denied by the auto-mode classifierDead. All literature retrieval run through WebFetch. TLS was not disabled and the proxy was not unset.
WebSearchHard budget stop: "limit: 200 WebSearch calls per turn, shared by every agent in it"Five queued queries were refused. All subsequent retrieval ran on WebFetch against URLs already surfaced in this run.
api.openalex.orgRATE_LIMIT_EXCEEDED twice, both on 10-wide parallel batchesNOT dead. Throttled to 1 to 3 concurrent and recovered. The one instructed single retry succeeded.

INSTRUMENT FAULT FOUND AND RECORDED

api.openalex.org ?search=... returns the WRONG RESULT SET through WebFetch: the query parameters are stripped and the endpoint returns the globally most-cited works instead. Four separate searches for fasting papers all returned Lowry 1951 (protein assay), Bradford 1976 (dye-binding assay) and Laemmli 1970 (bacteriophage T4). A reader who trusted the instrument would have recorded three irrelevant method papers as the fasting literature. The path forms /works/pmid:X, /works/pmcid:PMCX, /works/doi:X and /works/WX are unaffected and were used for every openalex row in this file. Anyone reusing this channel must use the path form only.

Retrieval debts

Sources located but not opened, or opened without the needed quantity. None of these supports any row.

#SourceWhat is missing
D1Wei et al. 2017, PMID 28202779Numeric IGF-1 change. Abstract came from an OpenAlex inverted index, not the published text. PubMed returned an empty shell.
D2Akasheh et al. 2024, PMID 39458471All numbers and p-values for IGF1, IGFBP1, IGFBP3. PMC full text captcha-blocked.
D3Khalafi et al. 2025, PMID 40805975Confidence intervals; the full list of modalities compared.
D4Mulas et al. 2023, Front Nutr 10:1146924PMID.
D5Elliott et al. 2014, Genes Nutr 9:432PMID; issue and page range.
D6Aksungar et al. 2007, PMID 17374948Absolute IL-6 and CRP concentrations and percent changes.
D7Rahbar et al. 2019, PMID 30774841IGF-1 and IL-2 values and p-values.
D8Aljahdali et al. 2024, PMID 38192134All effect sizes, CIs and p-values; whether other cardiometabolic markers were null or unreported.
D9Khoshandam Ghashang et al. 2021, DOI 10.1515/jcim-2019-0209Paywalled beyond the publisher abstract. Magnitudes, p-values, and explicit per-cytokine null statements. PMID.
D10Faris et al. 2012, PMID 23244540Abstract null in OpenAlex, not open access. The title asserts cytokine attenuation; WHICH cytokines and by how much is unconfirmed. No row written from it.
D11Faris et al. 2019, J Nutr Intermed Metab 15:18-26Located ONLY as a reference entry inside another paper. No results, no DOI, no PMID. This is a Ramadan inflammation meta-analysis and would bear directly on node 9.
D12Brandhorst et al. 2015, PMID 26094889Abstract null. Human-arm design, n, and IGF-1, IGFBP-1, glucose and ketone numbers.
D13Horne et al. 2013, PMID 23220077Abstract null, closed access. Nothing beyond the title retrieved. A randomized crossover water-only fast.
D14Leiper and Molla 2003, PMID 14681711Abstract null, closed access. This is the Ramadan FLUID-RESTRICTION review and bears directly on the DRY column.
D15Han et al. 2021, PMID 33723462Abstract null; nature.com blocked. Fasting-induced FOXO4 in human CD4+ T cells, with mTORC1 among the record's keywords. Would bear on nodes 2 and 9.
D16Al lami et al. 2024, PMID 38455231SAMPLE SIZE not stated in the retrieved record. Exact cytokine-to-timepoint mapping and all effect sizes.
D17Mähler et al., Bahá'í daytime dry fasting preliminary studyJournal, year, volume, pages and DOI. The DOAJ record served did not carry them. No numerical values for any outcome.
D18Antonellis et al. 2021, PMID 34535504Fold changes and exact p-values; the served text reports only "*p < 0.05" against figures.
D19Yilmaz et al. 2012, PMID 22722868Calorie-restriction percent and duration.
D20Li et al. 2023 (PMID 37139422) and Choi et al. 2024 (PMID 38811804)THE DIRECTION of the endogenous YAP or TAZ change during fasting. Neither retrieved abstract states it. This is the blocking gap on node 7. Li et al. also omit the fasting duration.
D21Rodenfels et al. 2014, DOI 10.1101/gad.249763.114Volume, pages and PMID.
D22Mihaylova et al. 2018, DOI 10.1016/j.stem.2018.04.001PMID; full author list truncated in the served record.
D23Chung et al. 2001, PMID 11795524Species not stated in the abstract; no numbers; no IKK or p65 data. This is the only source touching an actual NF-κB component.
D24Dai et al. 2022, PMID 36145236, Nutrients 14(18):3860WHETHER WATER WAS PERMITTED during the 10-day "complete fasting" is not stated. The study therefore cannot be classified PF or DRY and NO ROW WAS WRITTEN FROM IT, despite n=13 and a full metabolic panel. If water was withheld it is the largest absolute dry fast located.
D25The FGF21-to-growth-hormone-signaling mechanismelifesciences.org JavaScript-gated on both the article and PDF paths. No source opened, so NO node 3 mechanism row was written.
D26Persian-language recordsijem.sbmu.ac.ir article-1-405-fa.pdf, article-1-754-fa.pdf; ijdld.tums.ac.ir article-1-138-fa.pdf; rjms.iums.ac.ir article-1-2715-en.pdf; sid.ir PDFs. Located, not opened.
D27Chinese-language recordskyxuebao.kmmu.edu.cn article 7ffcd0cd-e48f-4292-a830-89d49a128d6a; xuebao.shsmu.edu.cn 2023;43(9):1131; manu37.magtech.com.cn/yyxb 2025;47(1):99. Located, not opened.
D28Turkish-language recordsGazi University lecture PDF "Ramazan açlığının moleküler mekanizması"; acikbilim.yok.gov.tr theses 20.500.12812/31408 and /108467; nek.istanbul.edu.tr thesis 59858. Located, not opened, and below witness standard.
D29ISRCTN17045897Surfaced as a possible dry-fasting trial registration. The WebSearch budget ended before it could be read. A registration is below witness standard regardless.
D30Papagiannopoulos et al. 2013, PMID 24434757INTERNAL INCONSISTENCY, not resolved: the abstract reports creatinine clearance rising "up to 167%" while the results text gives a maximum of +67 percent. Both figures are recorded in row 10.3 and neither is adopted over the other.
D31academic.oup.com/edrv/article/46/6/856/8211151Surfaced through the French channel as a likely Endocrine Reviews fasting review. Not opened.

Total retrieval debts: 31.

Sources opened and used

32 sources were opened and are cited in the rows above. Every identifier in this file (PMID, DOI, journal, volume, pages, year, author) was retrieved from an opened record. No identifier was derived, completed or inferred. Where a field could not be retrieved it is named in the debt table rather than filled in.

END OF SLICE A

← 9. NF-κB All 41 nodes 11. Estrogen receptor (ER) →


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.