Metformin

Limits on what this shows

• Key negative/null findings: Not every participant on metformin showed blunting; some had a positive mitochondrial respiratory response despite treatment. This is a single-study finding; no independently confirmed replication trial was located for this dossier.
• At-risk subgroups: The cohort studied was specifically older adults undergoing a structured aerobic exercise program. The finding has not been established in younger adults, in patients who are not exercising, or as a resting-state mitochondrial effect.

Regulatory status

Not addressed by the FDA label. The label's Warnings and Precautions section covers lactic acidosis risk in renal or hepatic impairment and hypoxic states, not baseline mitochondrial capacity or exercise-adaptation capacity in a generally healthy labeled population.

Population evidence

Konopka et al., Aging Cell 2019 (PMID 30548390): randomized, double-blind, placebo-controlled trial, n=53 older adults (mean age approximately 62 years; placebo n=26, metformin n=27), 12 weeks of supervised aerobic exercise training. Metformin attenuated the exercise-induced increase in whole-body insulin sensitivity and VO2max, and abrogated the increase in skeletal-muscle mitochondrial respiration seen with training in the placebo group. The response was heterogeneous: the metformin group included both positive and negative responders, so the blunting was not universal.

Mechanism

Named at the tissue-adaptation level (blunted mitochondrial biogenesis/respiratory response to a training stimulus); the trial itself does not resolve whether this reflects direct complex I inhibition in skeletal muscle or an indirect, AMPK-mediated effect on the exercise-adaptation signaling pathway. No classic marker of mitochondrial toxicity (structural damage, membrane potential collapse) was reported at the therapeutic dosing used.

Counter-evidence held on record

Metformin's mild, reversible inhibition of mitochondrial complex I (see the Oxygen delivery/utilization and ATP production entries below, and the PK prescreen) is the drug's own intended pharmacological mechanism for lowering hepatic glucose output, not a nutrient deficiency. This entry is filed separately from those because Konopka's finding is a distinct, exposure-bounded human outcome, the blunting of adaptation to a specific exercise stimulus, and is not counted twice as evidence for the mechanism entries below (see the Overlap map). Framing caution for the consumer-facing surface Do not present this as evidence that metformin causes generalized muscle wasting or mitochondrial damage at rest; the finding is specific to the adaptive response to a defined aerobic training program in older adults, and resting mitochondrial function was not shown to be impaired.

Limits on what this shows

• Key negative/null findings: Not applicable in the depletion direction, because no depletion-direction Tier A or Tier B evidence was found to begin with; this absence is itself the finding recorded here.
• At-risk subgroups: Not established.

Regulatory status

Not addressed by the FDA label; no oxidative-stress marker or monitoring recommendation appears anywhere in the labeling reviewed.

Population evidence

No human randomized trial or meta-analysis stratified specifically to metformin (as opposed to diabetes treatment generally) reporting a quantified oxidative-stress biomarker outcome, such as malondialdehyde, F2-isoprostanes, or total antioxidant capacity, was located and independently verified for this dossier despite a dedicated search. This is recorded as a genuine evidence gap in the human literature actually reviewed here, not as reassurance that no such studies exist elsewhere.

Mechanism

Present at Tier B, and it runs the reducing (protective) direction rather than a depleting one. Neupane et al., Cureus 2026 (PMID 41948241), a systematic review focused mainly on age-related macular degeneration, describes metformin as cytoprotective in retinal pigment epithelial cells via reduced reactive oxygen species and Nrf2 pathway activation; this review's population (largely preclinical/AMD-specific models) does not generalize directly to a general type 2 diabetes population, and full text was not independently accessed beyond the listing for this dossier. Metformin's own mild complex I inhibition (see the Oxygen delivery/utilization entry) is separately hypothesized in the broader pharmacology literature to reduce electron leak and reactive oxygen species generation at the respiratory chain, consistent with, but not proof of, a net-protective direction.

Counter-evidence held on record

This is filed as a context flag running the opposite direction from a deficiency/burden claim, per the grading key's explicit provision for interactions running the other direction; it is not evidence that metformin eliminates oxidative stress in every tissue or patient, only that the directionally relevant literature located and reviewed here points toward protection rather than depletion. Framing caution for the consumer-facing surface Do not present metformin as an antioxidant supplement or therapy on the strength of this entry; the supporting evidence is preclinical/mechanistic and disease-specific (age-related macular degeneration), not a general human dosing or supplementation claim, and no oxidative-stress-related dose of anything is recommended anywhere in this dossier.

Limits on what this shows

• Key negative/null findings: None located; this is the only human study on metformin and copper status identified in the literature searched for this dossier, so there is no independent replication or contradiction to report either way.
• At-risk subgroups: Not established. The cross-sectional design cannot establish a dose-response or duration-response relationship, and the two groups were not shown to be matched on diabetes duration, renal function, or diet, so confounding by indication (more advanced or longer-duration diabetes being preferentially treated with metformin) has not been excluded.

Regulatory status

Not addressed by the FDA label at all; the label contains no copper-related language, monitoring recommendation, or warning.

Population evidence

Otowa-Suematsu et al., BMJ Open Diabetes Research & Care 2025 (DOI 10.1136/bmjdrc-2025-005255): cross-sectional study, n=189 adults with type 2 diabetes (93 metformin users, 96 non-users). Metformin users had significantly lower serum copper (16.0 vs 17.8 µmol/L, P<0.001), and metformin use was identified as an independent predictor of serum copper level. Beyond the topline comparison and P-value, full-text detail (exact adjusted covariates, effect size with confidence interval, and whether iron or zinc were also reported) could not be independently verified for this dossier and is not asserted here.

Mechanism

The study authors propose metformin's metal-binding chemistry (the biguanide structure is a recognized metal-ion ligand) as a candidate explanation; this is the authors' own working hypothesis as summarized in the available abstract, not an independently confirmed mechanism. No human tissue-copper or functional-copper-marker study (for example, ceruloplasmin or cytochrome c oxidase activity) was located to corroborate it.

Counter-evidence held on record

A single cross-sectional comparison cannot separate metformin's own effect from confounding by diabetes duration, renal function, or dietary pattern between users and non-users; the authors describe an association and a working chelation hypothesis, not a demonstrated causal or dose-response depletion. Framing caution for the consumer-facing surface Do not present this as an established copper-deficiency risk; it is one observational signal, not a replicated finding, and no functional copper-deficiency marker in metformin users has been reported anywhere in the literature reviewed. Copper is assessed here entirely independently of iron below; the two are never merged into one trace-mineral grade.

Limits on what this shows

• Key negative/null findings: No study located and independently verified for this dossier demonstrates reduced serum iron, reduced transferrin saturation, or a validated functional iron-deficiency marker attributable specifically to metformin exposure.
• At-risk subgroups: Not established for a depletion direction. If the hepcidin-suppression mechanism above is physiologically relevant in humans, any subgroup already prone to iron overload (for example, hereditary hemochromatosis carriers) would be the theoretical group to watch for an excess-, rather than deficiency-, direction effect, though no clinical data confirming this was located.

Regulatory status

Not addressed by the FDA label; no iron-related language appears in the labeling reviewed.

Population evidence

Behradmanesh et al., Iran Red Crescent Med J 2011 (PMID 22737516), is titled as a clinical study of metformin's effect on serum ferritin in women with polycystic ovary syndrome; the direction and magnitude of the reported effect could not be independently verified for this dossier beyond confirming the study's existence, topic, and PMID, and it is not used here to support a specific directional claim. Separately, Luque-Ramírez et al. (Diabetes Care 2007, DOI 10.2337/dc07-0642; and Journal of Clinical Endocrinology & Metabolism 2011, DOI 10.1210/jc.2010-2211) established that women with polycystic ovary syndrome carry elevated body iron stores as a consequence of insulin resistance and hyperinsulinism itself, mediated by reduced circulating hepcidin. Metformin was not an exposure tested in either of these two studies, so they describe the disease-state background against which any metformin-PCOS-iron study must be interpreted, not a metformin-specific finding.

Mechanism

Deschemin, Foretz, Viollet and Vaulont, Scientific Reports 2017 (DOI 10.1038/s41598-017-12976-2): in hepatocytes, metformin inhibited BMP6-induced hepcidin gene expression through an AMPK-independent pathway. Hepcidin is the master negative regulator of intestinal iron absorption and macrophage iron release; suppressing it is mechanistically expected to increase iron availability, not reduce it. This is Tier B (in vitro/mouse hepatocyte) evidence and has not been confirmed as a physiologically meaningful effect on human iron status.

Counter-evidence held on record

The single directionally relevant human study identified (Behradmanesh 2011) could not be verified in the direction needed to support either a depletion or a retention claim, so this entry does not assert a net direction for human iron status; it records that the one confirmed mechanistic pathway runs opposite to a depletion story, per the requirement to check interaction direction explicitly. Framing caution for the consumer-facing surface This entry should never be surfaced as an iron-deficiency claim; if it is surfaced at all, it should be framed as an open mechanistic question with a plausible retention-direction pathway, not a depletion risk, and no iron dose or iron-avoidance recommendation is made anywhere in this dossier. Iron is assessed here entirely independently of copper above; the two are never merged into one trace-mineral grade.

Limits on what this shows

• Key negative/null findings: Qin et al., Nature 2012 (PMID 23023125): a two-stage metagenome-wide association study, n=345, found type 2 diabetes itself is characterized by gut microbial dysbiosis (reduced butyrate-producing bacteria, increased opportunistic pathogens), a confounder that exists independent of any specific drug exposure documented in that study. This is the confounding-by-indication check required before crediting microbiome change to metformin alone: some dysbiosis in a metformin-treated cohort is attributable to diabetes itself, not the drug. Wu et al.'s randomized, placebo-controlled, within-patient design is what allows this entry to attribute a portion of the change specifically to metformin rather than to diabetes progression.
• At-risk subgroups: Not established at the individual level; the cited trial measured population-level community shifts, not a defined higher-risk subgroup.

Regulatory status

The FDA label documents gastrointestinal adverse effects (diarrhea, nausea, vomiting, flatulence, and related symptoms) as very common, dose-related, and more frequent at treatment initiation, consistent with a gut-level drug effect, but the label does not attribute this to microbiome alteration specifically and does not use microbiome language anywhere.

Population evidence

Wu et al., Nature Medicine 2017 (PMID 28530702, DOI 10.1038/nm.4345): a double-blind, placebo-controlled trial of approximately four months in individuals with treatment-naive type 2 diabetes, alongside cross-sectional cohort comparisons. Metformin had strong, measurable effects on gut microbiome composition. Fecal microbiota from metformin-treated donors, transferred into germ-free mice, improved the recipient mice's glucose tolerance relative to microbiota from before treatment, directly demonstrating that the altered microbiota is functionally sufficient to affect host glucose handling rather than a bystander marker. The paper also reports effects on bacterial metabolic pathways, including genes encoding metalloproteins and metal transporters. Exact participant counts by sub-cohort and specific taxon-level shifts (for example, named genera) could not be independently confirmed beyond the abstract for this dossier and are not asserted here.

Mechanism

Proposed and partly demonstrated: metformin's incomplete absorption leaves high concentrations in the gut lumen well before any systemic distribution, giving it direct, local access to gut bacteria independent of any systemic action. The germ-free-mouse transfer experiment is the strongest evidence that the resulting community shift, not just glycemic improvement, drives part of the therapeutic effect.

Counter-evidence held on record

Metformin's common, well-labeled GI adverse effects (diarrhea, nausea) are a clinical outcome, not a microbiome-composition measurement, and are not re-filed here as gut-microbiome evidence; they are noted only as a plausible downstream consequence consistent with a gut-level drug effect. Framing caution for the consumer-facing surface This entry describes a compositional and functional shift in the gut microbiome, not a nutrient deficiency in the traditional sense, and not proof of generalized "gut damage." The cited trial's own functional read-out, improved glucose tolerance in mice receiving post-treatment microbiota, points toward a therapeutically relevant shift rather than a uniformly adverse one, and it should not be marketed as pure harm.

Limits on what this shows

• Key negative/null findings: Neither foundational paper reports whole-body hypoxia, reduced arterial oxygen content, or impaired oxygen delivery in an intact human; the effect described is cellular/mitochondrial oxygen utilization in isolated liver-tissue preparations, not systemic oxygen transport.
• At-risk subgroups: The labeled lactic-acidosis risk, the clinical scenario where impaired oxidative handling becomes apparent, is concentrated in renal impairment, hepatic impairment, acute heart failure, sepsis, hypoxic states, and excess alcohol intake, per the existing PK prescreen's renal-elimination facts above; this subgroup information is already carried in the PK prescreen and is not restated as new evidence here.

Regulatory status

The FDA label's boxed warning on lactic acidosis is the closest labeled acknowledgment of this physiology: it describes lactate accumulation under conditions (renal impairment, hepatic impairment, hypoxic states, excess alcohol) that impair lactate clearance or increase its production, consistent with reduced oxidative disposal of pyruvate under those conditions. The label does not describe this in complex-I or oxygen-utilization terms and does not present it as relevant to a patient with normal renal function.

Population evidence

No human study measuring resting oxygen consumption, tissue oxygenation, or a validated oxygen-utilization biomarker specifically in metformin users was located and independently verified for this dossier. The population-level human data available (Konopka 2019, filed under Mitochondrial burden above) addresses adaptation to a defined exercise stimulus, not resting oxygen delivery or utilization, and is not re-counted as independent evidence in this entry.

Mechanism

Named and demonstrated at Tier B, in isolated rat hepatocytes and mitochondrial preparations, not human tissue: El-Mir et al., Journal of Biological Chemistry 2000 (DOI 10.1074/jbc.275.1.223), and Owen, Doran and Halestrap, Biochemical Journal 2000 (DOI 10.1042/bj3480607), independently showed metformin (as dimethylbiguanide) inhibits complex I of the mitochondrial respiratory chain, reducing respiration in isolated hepatocytes. This is the biochemical basis proposed for metformin's rise in cellular AMP relative to ATP, which activates AMPK and suppresses hepatic glucose output. Per this dossier's governing rule, a drug's own demonstrated mechanism of therapeutic action is recorded here as pharmacology, not filed as an adverse nutrient-oxygen-delivery deficiency.

Counter-evidence held on record

This entry exists to document that the mitochondrial-oxygen-utilization mechanism behind metformin's own glucose-lowering action was actively considered against the "oxygen delivery/utilization" category and explicitly not filed as a deficiency, per the rule that a drug's own known clinical or pharmacological effect is not a nutrient-deficiency finding. Framing caution for the consumer-facing surface Do not present this entry as evidence of tissue hypoxia or impaired oxygen delivery in a typical patient with normal renal function; the supporting data are non-human, and the one labeled clinical scenario where oxidative handling becomes relevant (lactic acidosis) is a rare, renal/hepatic-impairment-associated boxed warning already governed by the drug's own dosing contraindications, not a dietary or supplement-addressable deficiency. This entry is assessed entirely independently of ATP production below; the two are never merged into one mitochondrial grade.

Limits on what this shows

• Key negative/null findings: Neither foundational mechanistic paper reports a generalized, whole-body ATP deficiency state; the effect described is localized to hepatocyte mitochondrial preparations under experimental conditions, and the drug's clinical utility itself depends on this being a mild, reversible, therapeutic-range effect rather than a toxic one.
• At-risk subgroups: Same renal/hepatic-impairment and hypoxic-state subgroup already carried in the PK prescreen and the lactic-acidosis boxed warning; not restated as new evidence here.

Regulatory status

Not addressed by the FDA label in ATP or bioenergetic terms; the label's boxed lactic-acidosis warning is the closest labeled acknowledgment of impaired oxidative energy metabolism, and only in the specific impaired-clearance/hypoxic scenarios already described in the PK prescreen and the Oxygen delivery/utilization entry above.

Population evidence

No human study measuring tissue ATP content or a validated cellular-energy-charge biomarker specifically in metformin-treated patients at therapeutic dosing was located and independently verified for this dossier.

Mechanism

Named, Tier B, non-human: El-Mir et al. (2000) and Owen et al. (2000), cited above under Oxygen delivery/utilization, describe reduced respiratory chain flux at complex I in isolated hepatocyte and mitochondrial preparations, which is mechanistically upstream of a rise in AMP/ATP and ADP/ATP ratios. A wider pharmacology literature debates whether this complex I effect, versus other proposed targets, is the operative mechanism at the much lower drug concentrations actually achieved in human plasma; this dossier attempted to retrieve that specific concentration-dependence literature and the attempts were blocked by source access limits, so the debate is noted here without asserting its resolution, per the citation-integrity rule.

Counter-evidence held on record

This entry is the clearest application of the rule that a drug's own demonstrated mechanism of action is not filed as a nutrient deficiency: reduced hepatic ATP synthesis, within a narrow and reversible range, is literally how metformin lowers blood glucose, and grading it as a deficiency finding would misrepresent the drug's basic pharmacology to a reader. Framing caution for the consumer-facing surface Do not present this entry as evidence that metformin causes fatigue, muscle weakness, or a generalized energy deficiency; no human outcome data of that kind were located, and any fatigue reported by a patient on metformin is far better explained, and independently evidenced in this dossier, by the red-tier vitamin B12 finding above than by a direct ATP effect. This entry is assessed entirely independently of Oxygen delivery/utilization above; the two are never merged into one mitochondrial grade.