🧬 Aging Wiki

gdf-15-biomarker

Properties1
aliasesGDF15 serum, GDF-15 plasma, MIC-1 serum, growth differentiation factor 15 biomarker, GDF15 aging biomarker

18 min read

GDF-15 as an Aging Biomarker (Serum/Plasma)

Serum GDF-15 (growth differentiation factor 15) is a single-analyte proteomic biomarker that rises monotonically with age and is among the most robustly validated predictors of all-cause mortality, frailty, physical decline, and dementia in elderly cohorts — consistently outperforming standard clinical risk factors in large prospective studies. This page covers the clinical serum interpretation in aging cohorts: reference ranges, mortality and frailty prediction, intervention-responsiveness, and current limitations. For receptor biology, upstream regulation, and molecular mechanism, see gdf15.

GDF-15 is also one of seven plasma-protein DNAm surrogates embedded in the grimage-2019 clock — its methylation-predicted level is one of the strongest individual contributors to that clock’s mortality-predictive power.

Identity

  • Protein page: gdf15 — gene biology, GFRAL receptor, UPRmt mechanism, metformin connection, therapeutic landscape
  • Canonical assay modality: Serum or EDTA-plasma immunoassay (ELISA or electrochemiluminescence; Roche Elecsys GDF-15 assay widely used in cohort studies)
  • Unit: pg/mL (= ng/L; the two units are numerically equivalent)
  • Pre-analytical considerations: Stable in serum/plasma; freeze-thaw cycles affect some assays; fasting status modestly influences levels; hemolysis should be avoided. No strong circadian variation reported.
  • Cross-reference: GDF-15 is embedded as a DNAm surrogate (proxy) in grimage-2019 (verified); GDF-15 is one of the 1,301-protein aging atlas analytes in lehallier-proteomic-clock-2019

Aging Trajectory

GDF-15 rises with age more consistently than almost any other circulating protein. In apparently healthy elderly individuals (n=429), Kempf et al. found a median serum GDF-15 of 762 pg/mL (IQR 600–959 pg/mL) 1. This population skewed older (mean age not stated but recruited from cardiology referral and general population); values in younger healthy adults are substantially lower.

In the Chinese multicenter reference-interval study (n=7,764 healthy adults across six cities), GDF-15 levels increased significantly with age in both males and females, with males consistently higher than females at matched ages 2. Age- and sex-specific reference intervals were established using non-parametric 2.5th–97.5th percentile methods, though exact decade-level values were not reported in the abstract.

In the Dallmeier 2016 KAROLA study of stable coronary heart disease patients (n=1,073; median age not specified in abstract), median baseline GDF-15 was 1,232 pg/mL — substantially higher than the healthy-elderly range above, consistent with cardiovascular disease burden driving GDF-15 elevation 3.

Approximate age-stratified ranges (healthy individuals, general population)

The table below synthesizes data from Kempf 2007 and the broader literature. These values are approximate summaries and should not be used for clinical reference without verification against assay-specific manufacturer reference intervals. needs-replication — assay-specific reference intervals vary; the Roche Elecsys and Quantikine ELISA differ in calibration.

Age groupApproximate median (pg/mL)Notes
20–40 yrs~400–600Healthy adults; data sparse
40–60 yrs~600–900Rises with metabolic aging, subclinical disease
60–75 yrs~900–1,300Community-dwelling; disease burden increasing
75–85 yrs~1,300–2,000Frailty threshold (~2,000) emerging
85+ yrsoften >2,000High disease burden; anorexia of aging range

The ~1,500 pg/mL upper reference limit applied in clinical vascular studies (Park 2026: healthy controls 1,016±219 pg/mL; 17/20 atherosclerotic disease patients exceeded 1,500) is consistent with the Kempf elderly median plus 2SD 1. dose-response-unclear — no single authoritative reference range for aging-population clinical use has been adopted across assay platforms.

Clinical Mortality Association

Wiklund 2010 — Swedish population cohort

In a Swedish population-based study (Uppsala Longitudinal Study of Adult Men + twin sub-cohort; n=876 males + 324 twins; up to 14 years of follow-up), serum MIC-1/GDF-15 was an independent predictor of all-cause mortality with an adjusted odds ratio of death of 3.38 (95% CI 1.38–8.26) for high vs. low GDF-15, independent of BMI and comorbidities 4. This paper established GDF-15 as a novel marker of all-cause mortality in aging men.

Fujita 2023 — Japanese community-dwelling older adults

In a Japanese community-dwelling cohort (n=1,801 older adults), higher serum GDF-15 quartile predicted total mortality with an HR of 1.98 (95% CI 1.09–3.59) for the highest vs. lowest quartile, adjusted for demographics and clinically relevant variables 5. Critically, the association became non-significant when further adjusted for cystatin C (HR 1.65, 95% CI 0.89–3.05) or β2-microglobulin (HR 1.69, 95% CI 0.91–3.12), suggesting GDF-15’s mortality prediction in this cohort is substantially confounded by kidney function biomarkers — a finding that meaningfully limits its interpretation as a pure aging signal.

Webber 2024 — Health ABC + Cardiovascular Health Study (preprint; PMC11326340)

In the largest cohort analysis identified (n=2,677; Health ABC n=1,174 + CHS n=1,503; mean age 75.4±4.4 years), highest vs. lowest GDF-15 quartile was associated with 6:

OutcomeHR (highest vs. lowest quartile)
All-cause mortality1.81 (1.53–2.15)
Severe mobility disability2.13 (1.64–2.77)
Heart failure2.09 (1.66–2.64)
Atherosclerotic CVD1.56 (1.22–1.98)
Coronary heart disease1.47 (1.17–1.83)
Dementia (CHS only)3.50 (1.97–6.22)

These HRs are from a preprint (medRxiv; no peer-reviewed journal publication identified as of 2026-05-09) and have been verified against the preprint PDF. needs-replication — peer-reviewed publication pending; findings should be confirmed once journal-reviewed version is available.

Dallmeier 2016 — KAROLA stable coronary heart disease cohort

In n=1,073 patients with stable coronary heart disease, both elevated baseline GDF-15 and 12-month change in GDF-15 predicted CVE and total mortality. Adjusted HR for elevated baseline GDF-15: 1.68 (95% CI 1.08–2.62) for cardiovascular events and 1.73 (1.02–2.94) for death. Median 12-month change was −16.7% — GDF-15 fell modestly over a year in this disease population 3. The decline may reflect regression to the mean or treatment effects rather than a generic aging trend.

Summary extrapolation table

DimensionStatus
GDF-15 mortality correlation in humans?yes — multiple large cohorts; all-cause mortality, CVD, frailty
Pathway conserved across populations?yes — consistent across Swedish, Japanese, Korean, US cohorts
GDF-15 causally drives mortality in MR?partial — some MR evidence; see Mendelian Randomization section

Frailty and Sarcopenia Association

In a cross-sectional Korean community-dwelling cohort (Korean Frailty and Aging Cohort Study; n=929, aged 70–84 yrs), sarcopenic individuals had median GDF-15 of 1,221 pg/mL vs. 1,019 pg/mL (non-sarcopenic; p<0.001); highest tertile (≥1,245 pg/mL) associated with OR 1.96 (95% CI 1.16–3.33) for sarcopenia vs. lowest (<885 pg/mL). However, the prospective component (n=788, 2-year follow-up) found no significant association with incident sarcopenia, raising questions about directionality 7.

The MAPT study (n=1,096 community-dwelling adults, mean age 75.3±4.4 years; 63.9% female) found high GDF-15 (≥1,500 pg/mL) cross-sectionally associated with frailty (Fried phenotype ≥3 criteria; high vs. low: OR=3.56, 95% CI 1.58–8.03 in Model 3, adjusted for age, sex, BMI, MAPT group, education, and IL-6) 8. Longitudinally, high GDF-15 predicted incident frailty over 4 years in the minimally adjusted model (HR=1.69, 95% CI 1.03–2.78, Model 1: age and sex only), but this association became non-significant in the fully adjusted model (HR=1.40, 95% CI 0.85–2.33, Model 4: additional comorbidities). This attenuation limits the strength of the prospective claim.

In mobility-limited older adults (n=429 BIOFRAIL study; mean age 79.6±6.2 years; 64% female; frailty by Clinical Frailty Scale ≥5), median GDF-15 was 2,252 pg/mL in frail vs. 1,438 pg/mL in non-frail (p<0.001); optimal frailty cut-off 2,047 pg/mL (AUC 0.681, 95% CI: 0.623–0.739) 9. For sarcopenia (EWGSOP2), median GDF-15 was 1,916 pg/mL vs. 1,569 pg/mL in non-sarcopenic (p=0.035), but discriminatory accuracy was poor (AUC 0.577).

Mechanism Context

GDF-15 does not drive mortality or frailty directly — it is a systemic reporter of integrated stress that rises in proportion to the burden of multiple hallmarks simultaneously. The upstream mechanisms are detailed in gdf15; the clinical interpretation consequence is:

  • GDF-15 elevation = weighted sum of mitochondrial-dysfunction (via UPRmt → ATF4 → GDF-15), cellular-senescence (SASP cytokine milieu activating NF-κB → GDF-15), and chronic-inflammation (NF-κB and HIF-1α transcription of GDF-15)
  • Because GDF-15 integrates multiple upstream stressors, it has high prognostic value but low specificity for any individual hallmark
  • Its appetite-suppressive effect (via brainstem GFRAL) means that very high GDF-15 in frail elderly may actively worsen anorexia of aging — making GDF-15 not purely a bystander biomarker but a potential active contributor to the sarcopenia-anorexia-cachexia spiral

no-mechanism — whether reducing GDF-15 (via anti-GFRAL/anti-GDF-15 antibodies) in elderly individuals with elevated GDF-15 would improve functional outcomes is unknown. Trials exist for cancer cachexia (ponsegromab; see gdf15 for details) but not aging per se.

Mendelian Randomization Status

Status: partial — MR evidence exists but is primarily for cardiovascular endpoints rather than aging per se.

Chen 2026 (UK Biobank, n=53,026; median 14.5-year follow-up) conducted two-sample MR and identified GDF-15 as a “broad-spectrum mediator” with mediation effects across 9 cardiovascular outcomes (median proportion mediated 25.7%), with the strongest observational association for heart failure (P=1.21×10⁻¹⁶⁶) 10. MR identified GDF-15 among 225 proteins with causal cardiovascular disease associations. Note: this paper is published in Protein & Cell (peer-reviewed; advance access August 2025), not a preprint.

MR evidence for GDF-15 → all-cause mortality specifically (independent of CVD mediating path) is not established. The instruments used in cardiovascular MR are primarily pQTLs from UK Biobank; aging-specific MR for GDF-15 as a driver of biological aging pace has not been published as of 2026-05-09. needs-replication — a dedicated MR study of GDF-15 pQTLs against longevity/mortality in an ancestry-diverse cohort is needed.

Intervention-Responsiveness

Metformin — paradoxical elevation

Metformin at therapeutic doses elevates serum GDF-15, as established mechanistically in Coll et al. 2020 (intestinal GDF-15 induction; see gdf15 for full discussion). This creates a clinical interpretation challenge: a patient on metformin who shows elevated GDF-15 does not have an elevated stress signal per se — the elevation is pharmacologically induced. Kolnes 2026 confirmed that exercise-related GDF-15 research must be interpreted separately from metformin’s drug effect 11.

The metformin-induced GDF-15 elevation may actually be mechanistically beneficial (anorexigenic effect → weight reduction) rather than harmful. This dissociation undermines GDF-15 as an uncontrolled biomarker in populations on metformin.

Clinical implication: Always note metformin use when interpreting serum GDF-15 in aging populations. dose-response-unclear — the magnitude of GDF-15 elevation varies by dose and duration of metformin treatment; population-level reference adjustments have not been established.

Acute exercise — transient increase

Kleinert et al. 2018 (n=7 healthy males; 1-hour cycling at 67% VO₂max) showed acute exercise raised plasma GDF-15 by 34% during exercise (p<0.001) and 64% above resting levels at 120 minutes post-exercise (p<0.001); skeletal muscle was not the source (a-v difference absent) 12. The source is likely liver, heart, or kidney responding to metabolic/hemodynamic exercise stress.

Interpretation: acute GDF-15 elevation post-exercise is a transient stress signal and should not be used for aging-risk assessment within hours of exercise. Whether chronic exercise training (as opposed to a single bout) lowers resting GDF-15 — as one would predict if exercise reduces the mitochondrial stress and inflammation driving GDF-15 — is not robustly established. contradictory-evidence — some training studies report modest reduction in resting GDF-15 in older adults; no well-powered RCT establishes a directional consensus.

Caloric restriction — context-dependent increase

Dostálová et al. 2009 (n=17 obese non-diabetic women; 2-week very low calorie diet) found that VLCD increased serum MIC-1/GDF-15 in obese subjects 13. This parallels the metformin finding: caloric restriction as an acute or severe dietary intervention appears to induce GDF-15 via ISR/ATF4 mechanisms. Whether modest chronic caloric restriction (the geroprotective intervention) in normal-weight individuals has the same effect is unclear. contradictory-evidence — the CR + GDF-15 relationship has not been characterized in the CALERIE-2 cohort (which would be the most relevant human dataset).

Senolytics — preliminary decrease

Some senolytic interventions have been associated with reduced GDF-15, consistent with the mechanistic expectation that clearing senescent cells (which contribute to SASP-driven GDF-15 elevation) would reduce circulating levels. Evidence is from small studies and not yet from well-powered RCTs. needs-replication unsourced — a verified primary citation for senolytic-driven GDF-15 reduction should be added on verification pass.

GDF-15 as a GrimAge Component

GDF-15 is one of seven plasma-protein DNAm surrogates embedded in grimage-2019. The DNAm-based proxy of GDF-15 expression is among the strongest individual contributors to GrimAge’s mortality-predictive composite. This means:

  1. Biological mechanisms elevating actual serum GDF-15 also tend to elevate GrimAge acceleration
  2. GrimAge can detect GDF-15-driven mortality risk even without a direct GDF-15 assay
  3. The two measurements are complementary: serum GDF-15 is direct, actionable (can identify metformin confounding, etc.); GrimAge DNAm-GDF-15 proxy integrates methylation changes upstream of the protein level

In the lehallier-proteomic-clock-2019 proteomic aging atlas, GDF-15 was among the proteins with the strongest age-upregulation signal across the human lifespan, particularly in the 4th and 6th decade waves of aging-related protein change. unsourced — the specific wave assignment for GDF-15 in Lehallier 2019 should be confirmed against the primary source.

Reference Range Summary

For clinical aging interpretation (not diagnostic cut-offs):

Interpretation zoneApproximate serum GDF-15Clinical context
Normal (young-adult baseline)<600 pg/mLHealthy 20–40-yr-old
Age-typical (60–75 yrs)600–1,300 pg/mLCommunity-dwelling; varies by sex and comorbidity
Elevated (borderline)1,300–2,000 pg/mLAssociated with frailty, metabolic disease
High (frailty/disease range)2,000–3,000 pg/mLFrailty threshold ~2,000 in BIOFRAIL; HF risk zone
Very high>3,000 pg/mLAdvanced disease, cancer, heart failure, severe mitochondrial disease

Confounders that raise GDF-15 regardless of biological aging:

  • Metformin use (pharmacological ISR induction)
  • Acute/recent exercise (transient; resolves within hours)
  • Cancer (tumor-derived GDF-15)
  • Heart failure (hemodynamic stress)
  • Chronic kidney disease (reduced clearance + renal production)
  • COPD (hypoxia-driven)
  • Severe obesity (metabolic stress)
  • Pregnancy (placental production — extremely elevated)

Specificity for aging-related mitochondrial stress and senescence is low without exclusion of these confounders. GDF-15 is a sensitive but non-specific aging signal.

Limitations and Gaps

  1. Specificity problem — GDF-15 rises with cancer, heart failure, CKD, COPD, pregnancy, and acute illness. It cannot be interpreted as an aging biomarker without careful exclusion of these conditions. It is a non-specific stress integrator, not an aging-specific marker. contradictory-evidence

  2. Metformin confounding — Therapeutic metformin elevates GDF-15 via intestinal ISR. Any aging study that does not stratify by metformin use will conflate drug-induced GDF-15 elevation with aging-related elevation. dose-response-unclear

  3. Assay heterogeneity — Different immunoassay platforms (Roche Elecsys, Quantikine ELISA, proximity extension assay in Olink panels) do not produce identical absolute values. Reference ranges are assay-specific; the values above are approximations across platforms. needs-replication

  4. MR evidence gaps — Causal MR evidence for GDF-15 → biological aging pace (as opposed to cardiovascular endpoints) is not established as of 2026-05-09. pQTLs for GDF-15 exist in UK Biobank proteomics, but a dedicated MR aging study has not been published. needs-replication

  5. Intervention directionality unclear — Exercise and CR both appear to acutely increase GDF-15 (ISR activation), while senolytics may decrease it (SASP reduction). Net chronic effect of these interventions on resting GDF-15 — and whether a reduction is biomarker-good or biomarker-bad — has not been systematically characterized. contradictory-evidence

  6. Sex and menopausal status — Males consistently have higher GDF-15 than age-matched females in most cohorts, though the gap narrows at older ages. Pregnancy dramatically elevates GDF-15 (placental production). Sex-specific reference ranges and aging trajectories require dedicated analysis. dose-response-unclear

  7. Causal vs. bystander ambiguity — Whether elevated GDF-15 drives worse outcomes (via anorexia, muscle wasting through GFRAL-dependent mechanisms) or is purely a bystander of upstream stress is unresolved. See gdf15 for the therapeutic landscape around this question.

Cross-references

  • gdf15 — protein biology, receptor, UPRmt mechanism, metformin connection, therapeutic landscape (anti-GDF-15 antibodies)
  • grimage-2019 — GDF-15 DNAm surrogate is a component; mortality prediction architecture
  • lehallier-proteomic-clock-2019 — GDF-15 in the 1,301-protein aging proteome
  • mitochondrial-dysfunction — upstream driver of GDF-15 via UPRmt → ATF4
  • cellular-senescence — SASP-driven NF-κB induction of GDF-15
  • chronic-inflammation — NF-κB and HIF-1α drive GDF-15; reciprocal signaling
  • metformin — metformin elevates GDF-15 via intestinal ISR; Coll 2020
  • senolytics — preliminary evidence senolytics reduce GDF-15 (unverified)

Footnotes

Footnotes

  1. doi:10.1373/clinchem.2006.076828 · Kempf T et al. · Clinical Chemistry 53(2):284–291, 2007 · observational (cross-sectional reference population) · n=429 apparently healthy elderly + n=153 CHF patients · median serum GDF-15 762 ng/L (IQR 600–959 ng/L) in healthy elderly · archive: pending download · DOI confirmed via PubMed PMID 17185363 ↩ ↩2

  2. doi:10.2147/JIR.S523980 · Chen K, Fan X, Wu J et al. · J Inflammation Res 2025 · observational (multicenter reference interval study) · n=7,764 healthy Chinese adults (6 cities) · GDF-15 increases with age (P<0.001); males significantly higher than females; age- and sex-specific 2.5th–97.5th percentile reference intervals established · archive: not_oa · PMID 40703640 ↩

  3. doi:10.1373/clinchem.2016.256206 · Dallmeier D et al. · Clinical Chemistry 62(8):1091–1101, 2016 · observational (prospective cohort — KAROLA study) · n=1,073 stable coronary heart disease patients · median baseline GDF-15 1,232 pg/mL; HR CVE 1.68 (1.08–2.62), HR death 1.73 (1.02–2.94) for elevated GDF-15; median 12-month GDF-15 change −16.7% · PMID 27197673 · archive: not checked ↩ ↩2

  4. doi:10.1111/j.1474-9726.2010.00629.x · Wiklund FE et al. · Aging Cell 9(6):1057–1064, 2010 · observational (prospective cohort) · n=876 males + 324 twins; up to 14-year follow-up · adjusted OR death 3.38 (95% CI 1.38–8.26) for high GDF-15 · model: Swedish population-based (Uppsala Longitudinal Study of Adult Men + twins) · PMID 20854422 · archive: download failed (bronze OA returned 403); DOI confirmed via Crossref; n and OR confirmed from PubMed abstract · 100th citation percentile in archive (273 citations) no-fulltext-access ↩

  5. doi:10.1093/gerona/glad105 · Fujita Y, Ito M et al. · J Gerontol A Biol Sci Med Sci 2023 · observational (prospective cohort) · n=1,801 community-dwelling Japanese older adults · HR total mortality 1.98 (95% CI 1.09–3.59) highest vs. lowest GDF-15 quartile (adjusted for demographics + clinical variables); became NS when further adjusted for cystatin C (HR 1.65, 0.89–3.05) or β2-microglobulin (HR 1.69, 0.91–3.12) — association largely explained by renal function · archive: not_oa · PMID 37190783 ↩

  6. doi:10.1101/2024.08.07.24311629 · Webber K, Patel S, Kizer JR, Eastell R, Psaty BM, Newman AB, Cummings SR · medRxiv preprint 2024 (not peer-reviewed as of 2026-05-09; no published journal version identified) · observational · n=2,677 (Health ABC n=1,174 + CHS n=1,503; mean age 75.4±4.4 yrs) · mortality HR 1.81 (1.53–2.15); severe mobility disability HR 2.13 (1.64–2.77); HF HR 2.09 (1.66–2.64); ASCVD HR 1.56 (1.22–1.98); CHD HR 1.47 (1.17–1.83); dementia HR 3.50 (1.97–6.22) for highest vs. lowest GDF-15 quartile · PMID 39148825 · PMC PMC11326340 · archive: PDF downloaded and verified against primary source ↩

  7. doi:10.1093/gerona/glab201 · Kim M, Walston J, Won CW et al. · J Gerontol A Biol Sci Med Sci 77(3):528–535, 2022 · cross-sectional + 2-yr prospective (Korean Frailty and Aging Cohort Study) · n=929 cross-sectional; n=788 prospective; aged 70–84 yrs; eGFR <60 excluded · median GDF-15 1,221 pg/mL (sarcopenic) vs. 1,019 pg/mL (non-sarcopenic; p<0.001); highest tertile (≥1,245 pg/mL) OR sarcopenia 1.96 (1.16–3.33) vs lowest (<885 pg/mL); prospective 2-yr incident sarcopenia: NS · archive: not_oa · PMID 34255062 ↩

  8. doi:10.1002/jcsm.70182 · Sánchez-Sánchez JL, Rolland Y, Lucas A, Guyonnet S, Vellas B, de Souto Barreto P; for the MAPT/DSA Group · J Cachexia Sarcopenia Muscle 2026, 17:e70182 (peer-reviewed) · observational secondary analysis (MAPT RCT cohort — longitudinal) · n=1,096 (mean age 75.3±4.4 yrs, 63.9% female); 4-yr follow-up; frailty by Fried phenotype · frailty OR (high ≥1,500 vs low <1,500 pg/mL GDF-15) 3.56 (1.58–8.03) cross-sectional (Model 3, adjusted); incident frailty HR 1.69 (1.03–2.78) prospective minimally adjusted (Model 1: age+sex) — attenuated to HR 1.40 (0.85–2.33) fully adjusted (Model 4: comorbidities added; p=0.187, NS) · archive: PDF downloaded and verified against primary source · PMID 41560403 ↩

  9. doi:10.1007/s11357-025-01946-6 · Hansen P, Nygaard H, Praeger-Jahnsen L, Schultz M, Dela F, Aagaard P, Ryg J, Suetta C · GeroScience 2025, 48:1955–1966 (published online 25 November 2025; peer-reviewed) · observational cross-sectional (BIOFRAIL study, Copenhagen University Hospital) · n=429 mobility-limited older adults (mean age 79.6±6.2 yrs; 64% female; 24.9% frail by Clinical Frailty Scale ≥5; 15.6% sarcopenic by EWGSOP2) · frailty median GDF-15 2,252 pg/mL vs. non-frailty 1,438 pg/mL (p<0.001); AUC 0.681 (95% CI: 0.623–0.739) for frailty; frailty cut-off 2,047 pg/mL (Youden’s index); sarcopenia median 1,916 pg/mL vs. 1,569 pg/mL (p=0.035); AUC 0.577 for sarcopenia (poor discriminatory ability) · archive: PDF downloaded and verified against primary source · PMID 41286529 ↩

  10. doi:10.1093/procel/pwaf072 · Chen YL, Wang JJ, You J, Cheng JY et al. · Protein & Cell 2026, 17:231–247 (advance access published 6 August 2025; peer-reviewed) · MR + observational (UK Biobank n=53,026; median 14.5-yr follow-up, IQR 13.8–15.3) · Cox proportional hazard models: 3,089 significant protein-CVD associations; GDF-15 among proteins with broadest CVD association (10 outcomes, AUC 0.78 cumulative importance score) · two-sample MR (IVW): 225 proteins causally linked to CVDs; GDF-15 identified as broad-spectrum mediator across 9 CVD outcomes (median proportion mediated 25.7%); heart failure P=1.21×10⁻¹⁶⁶ observational · archive: PDF downloaded and verified against primary source · PMID 40927895 ↩

  11. Kolnes et al. Frontiers in Endocrinology 2026 (PMID 41928890) — a review on metformin and GDF-15 — notes that exercise itself increases circulating GDF-15, citing Kleinert 2018 and Sabaratnam 2024 (acute exercise in T2DM). Confirms the exercise-GDF-15 dissociation from metformin-GDF-15 elevation as a confound to interpret in any mixed intervention study. ↩

  12. doi:10.1016/j.molmet.2017.12.016 · Kleinert M, Clemmensen C, Sjøberg KA et al. · Molecular Metabolism 9:96–101, 2018 · in-vivo (human exercise study) · n=7 healthy males; 1-hr cycling at 67% VO₂max · GDF-15 rose 34% during exercise (p<0.001); 64% above resting at 120-min post-exercise (p<0.001); skeletal muscle not the source (no a-v difference) · archive: pending · PMID 29398617 ↩

  13. doi:10.1530/EJE-09-0417 · Dostálová I, Roubícek T, Bártlová M et al. · Eur J Endocrinol 161(3):397–404, 2009 · observational + intervention (2-wk VLCD) · n=17 obese non-diabetic women; n=14 obese T2DM; n=23 healthy lean controls · VLCD significantly increased serum MIC-1/GDF-15 in obese non-diabetic group; T2DM group: NS; specific numeric changes not stated in abstract · archive: not_oa · PMID 19515791 ↩

Graph View

Backlinks