Metronidazole

Direction

Possible excess mitochondrial burden in a definable genetic subgroup; protective in a rat NSAID model. Liver failure is a drug outcome, not a mitochondrial measurement (Hard Rule 2).

Regulatory status

Contraindicated in Cockayne syndrome; "Severe irreversible hepatotoxicity/acute liver failure with fatal outcomes have been reported" [1][2].

Subgroup outcome (Tier B, case series)

Wilson 2015 [25]: from the Cockayne Syndrome Natural History study, 8 patients (8% of the cohort) developed acute liver failure after metronidazole and 3 died 6 to 11 days after the first dose; 2 also had neurologic deficits; no patient with Cockayne syndrome took metronidazole without a serious adverse effect. No mitochondrial parameter was measured in people.

Animal and cell evidence (Tier B)

Hernandez Herrera 2025 [26]: ercc6-deficient zebrafish larvae were killed by metronidazole while nearly all wild-type controls survived; mutant embryos consumed more oxygen, read by the authors as abnormal mitochondrial function; the SOD mimetic MnTBAP restored oxygen consumption; no ATP was reported. Xiao 2019 [24]: in primary human neurons and cell lines, metronidazole induced apoptosis at clinically relevant concentrations; the mitochondrial complex impairment, membrane potential loss and "energy crisis" are attributed to tigecycline, not metronidazole. Leite 2001 [27]: in rats, metronidazole pretreatment significantly reduced the indomethacin-induced rise in state 4 oxygen consumption and the fall in respiratory control ratio in isolated liver mitochondria (protective direction; doses not retrieved).

Cohort test: set aside

A 2025 respirometry study in human HEK 293T cells tested gentamicin, ciprofloxacin and amoxicillin only [18]; metronidazole was not tested, so it gives no information here.

Mechanism

Reductive activation under low oxygen [21] is the established chemistry; why Cockayne cells are vulnerable is unresolved in the retrieved sources (the zebrafish authors propose ROS-driven mitochondrial dysfunction).

Counter-evidence held on record

Counter-evidence and confounding The subgroup is very rare and the contraindication is absolute; for everyone else the mitochondrial data are in vitro, animal or run in the opposite direction. Findings here overlap with the Oxidative stress and ATP entries. Framing caution A contraindication and a safety flag, not a nutrient finding. No mitochondrial or antioxidant supplement is implied.

Direction

Possible excess oxidative burden, conflicting between models.

Regulatory status

Labels contain no oxidative stress statement; they report carcinogenicity in rodents and mutagenicity in some in vitro assays [2].

Human cell evidence (Tier B)

Xiao 2019 [24]: azithromycin, clindamycin and tigecycline caused oxidative damage in primary human neurons and cell lines; metronidazole induced apoptosis but "did not increase ROS production", which the authors read as a ROS-independent mechanism (no concentrations or values in the record).

Animal evidence (Tier B)

Hassan 2020 [28]: cerebellar oxidative stress markers significantly higher in rats after 60 days at 400 mg/kg/day. A mouse myocardial study (Chemosphere 2024) appeared in a PubMed listing by title only and was not opened. Hernandez Herrera 2025 [26]: the SOD mimetic MnTBAP restored oxygen consumption in metronidazole-sensitive ercc6 mutant zebrafish; ROS were proposed, not measured in the abstract.

Human measurement

No human oxidative marker (MDA, antioxidant enzyme, total antioxidant capacity) in metronidazole users was found in the first 20 results of the PubMed phrasing "metronidazole oxidative stress" (145 hits).

Mechanism

Reductive activation under low oxygen [21]; whether this produces redox stress in oxygenated human tissue is not established by the sources read.

Counter-evidence held on record

Counter-evidence and confounding The rodent dose is far above human exposure and the one human-cell study points the other way; the Cockayne zebrafish result concerns a rare genotype. Framing caution No antioxidant supplement is recommended on this evidence.

Direction

Neutral and indication-linked in humans; nutrient-to-drug in vitro (Hard Rule 4). Kept separate from Copper (Hard Rule 7).

Regulatory status

Labels name no iron interaction; iron appears only as iron oxide colorant in the capsule [2] (not mentioned in the tablet label summary [1]).

Human evidence (Tier A, single arm)

Monajemzadeh 2008 [35]: 102 children aged 5 to 11 years with Giardia lamblia, metronidazole 15 mg/kg/day for 5 days; blood drawn before and 40 to 45 days after. Iron-deficiency anemia was present in 26.5% at baseline. After treatment hemoglobin, hematocrit, serum ferritin, serum iron to TIBC ratio and TIBC rose and red-cell distribution width fell (all P < 0.0001). No untreated comparator; before and after means not in the abstract record. Hard Rule 6: an improvement after the drug is started cannot be attributed to a drug-caused deficit.

In vitro (Tier B)

Yamaki 2022 [36]: Fe3+ increased C. difficile growth and toxin gene expression and significantly raised the metronidazole minimum inhibitory concentration; deferoxamine reduced iron-stimulated toxin production. The PubMed listing for "metronidazole iron" (217 hits, first 20 seen) shows many bacterial and protozoal studies tying iron homeostasis to metronidazole resistance (titles only, not opened); one title concerns iron malabsorption in giardiasis (not opened).

Mechanism

None for a drug effect on human iron. In bacteria, iron-sulfur proteins and iron homeostasis relate to nitroimidazole activation and resistance (titles and [36] only).

Counter-evidence held on record

Counter-evidence and confounding The giardiasis cohort shows infection-linked iron deficiency improving with treatment; it says nothing about an iron-lowering drug effect and shares its source with the Oxygen delivery entry. Framing caution No iron supplement is recommended or discouraged by this entry. The in vitro finding concerns bacteria in a dish, not dietary iron.

Direction

Nutrient-to-drug (reverse of depletion; Hard Rule 4). Kept separate from Iron (Hard Rule 7).

Regulatory status

Labels name no copper interaction [1][2].

Evidence (Tier B/C, in vitro review)

Bozic Cvijan 2023 [37]: metronidazole is among the antibiotics listed as studied with Cu ions; direct Cu-metronidazole interactions were described "nearly 40 years ago"; a Cu(II) complex binding through the imidazole nitrogen is reported to have higher stability and protection from inactivation by several enzymes; a 2022 study reported that copper complexes of metronidazole or metronidazole benzoate had better biological activity (no assay detail given). The review reports no synergy or antagonism and no human data.

Human measurement (Tier A, mixed exposure)

Olivares 2003 [33]: serum copper measured before and 3 months after treatment in 25 children with giardiasis treated with tinidazole and metronidazole; per the abstract wording (as summarized) only magnesium differed significantly in that group, so copper did not. No values given; infection and a second nitroimidazole confound.

Searches

PubMed "metronidazole copper" (73 hits): the first 20 were coordination chemistry, nanoparticle, sensor and environmental degradation papers; none measured copper in people. The question whether copper reduces nitroimidazole activation was not answered by any retrieved source; the only direct statement is the better activity of the complexes in vitro.

Mechanism

Complexation chemistry only.

Counter-evidence held on record

Counter-evidence and confounding Review-level evidence from in vitro work; no assay detail retrieved for the 2022 complex study. Framing caution No change to dietary or supplemental copper is implied.

Limits on what this shows

• At-risk subgroups: Children (the second single-drug trial and one multi-drug cohort were pediatric); healthy adults for the first trial; people taking metronidazole inside combination regimens (H. pylori regimens, three-drug volunteer regimen); repeated or prolonged courses are not studied for metronidazole.

Direction

Disturbance of the gut microbial community (a risk entry). It is not a nutrient deficiency and implies no supplementation (Hard Rules 2 and 5).

Regulatory status

Neither label read contains a gut-flora, microbiome or C. difficile statement [1][2]. They state the drug treats serious infections caused by susceptible anaerobic bacteria and that candidiasis "may present more prominent symptoms during therapy" [1].

Population evidence, single drug, trial 1 (Tier A, adults)

Lewis 2005 [7]: healthy volunteers randomized to metronidazole for 5 days, oligofructose for 12 days or both; stool on days 0, 5, 12 and 19. In the 16 SRB-positive volunteers, fecal sulphate-reducing bacteria fell with metronidazole (95% CI 2.19 to 3.36, P < 0.0001) and the fall persisted when it was combined with oligofructose (95% CI 2.32 to 4.01, P < 0.0001); sulphate reduction rates fell with metronidazole whether or not SRB were detected (95% CI 29.0 to 56.1, P < 0.0001); fecal anaerobes fell with metronidazole; short-chain fatty acid concentrations fell with metronidazole (95% CI -103.7 to -19.6, P = 0.007) and rose with oligofructose. Fecal sulphate and hydrogen sulphide did not change with metronidazole. Abstract only (reconstructed and paraphrased): total volunteers, dose, raw counts and units were not retrieved.

Population evidence, single drug, trial 2 (Tier A, children)

Gotfred-Rasmussen 2021 [10]: children colonized with Dientamoeba fragilis given metronidazole (n = 48) or placebo (n = 48), plus 70 parasite-negative controls; 275 stool samples before treatment, 2 weeks (T2) and 8 weeks (T5) later; ribosomal DNA sequencing. Increases in Hungatella, Sutterella and Streptococcus at 2 weeks were linked to metronidazole exposure rather than to colonization. Eight genera differed between children who lost the parasite and those who kept it; seven of those were back at pretreatment levels by 8 weeks. The authors conclude metronidazole had a short-term effect on some genera. Abstract only (reconstructed and summarized): dose, diversity indices and p-values were not retrieved.

Population evidence, multi-drug regimens containing metronidazole (Tier A for the regimen, not separable)

Haak 2018 [9]: 13 healthy men (10 treated, 3 untreated); 7 days of ciprofloxacin 500 mg every 12 h plus vancomycin 500 mg every 8 h plus metronidazole 500 mg every 8 h; 16S sequencing. Inverse Simpson index day 0 versus day 9: median 40.9 versus 5.3 (P = 0.001); observed taxa 163 versus 85 (P = 0.048); butyrate-producing genera fell from a median of 27% to 0.3% (P < 0.001); Bacteroides, Subdoligranulum and Faecalibacterium fell and Streptococcus and Lactobacillus overgrew. At day 49 butyrate producers were back to baseline but the inverse Simpson index was still lower (18.3, P = 0.011); at 8 to 31 months (median 12) composition was comparable to baseline but unweighted UniFrac still differed (P = 0.016). Seven of the 10 treated men also received an LPS infusion, which the authors say did not appear to affect recovery. Liou 2019 [11]: 1,620 adults with H. pylori randomized to 14-day triple therapy, 10-day concomitant therapy or 10-day bismuth quadruple therapy. Diversity fell at 2 weeks after all three regimens. The summary read states it returned to baseline by week 8 and, at 1 year, only after triple therapy, with the other two regimens not fully recovered at 1 year (its week 8 wording is ambiguous). By standard regimen definitions the concomitant and bismuth quadruple regimens contain metronidazole and the triple regimen does not (the retrieved summary did not list the drugs; verify against the full text). Faecal E. coli antibiotic resistance rose temporarily after triple and concomitant therapy and was back to baseline by week 8 and at 1 year. Abstract only, summarized. Zhou 2020 [13]: 22 penicillin-allergic children with H. pylori and 23 H. pylori-negative asymptomatic controls (the abstract summary does not make clear whether controls were also treated); 14 days of omeprazole, clarithromycin, metronidazole and bismuth; stool 16S at weeks 0, 2, 6 and 52. Alpha diversity fell at week 2 (P < 0.05; the Chao1 index in patients was not significant, P = 0.16), some changes persisted to week 6 and all were restored by week 52; Bacteroidetes fell, Proteobacteria rose, Bacteroides, Faecalibacterium, Phascolarctobacterium and Blautia declined and Escherichia-Shigella, Klebsiella, Enterococcus and Streptococcus rose. Abstract only (reconstructed); some p-values could not be matched to taxa. Jakobsson 2010 [12]: 6 adults, 3 treated with metronidazole 400 mg plus clarithromycin 250 mg plus omeprazole 20 mg twice daily for 7 days. Fecal Actinobacteria (Bifidobacterium, Collinsella) fell sharply, Shannon diversity fell, and the erm(B) macrolide-resistance gene rose 3 to 5 orders of magnitude, still more than 1,000-fold higher at 1 year and evident at 4 years. The authors report no p-values (too few patients) and infer that the Actinobacteria decline is "probably due to clarithromycin", an inference they did not test.

Mechanism

Present and named. Metronidazole is a prodrug reduced under low oxygen [21], which is why obligate anaerobes are the target, and the label indication is infection by susceptible anaerobes [1]. Selective loss of obligate anaerobes is the plausible route to the butyrate-producer loss in [9] and to the fall in fecal short-chain fatty acids measured in [7]; the link between taxa and metabolites is inferred from co-movement, not proven causal in the abstracts read.

Mixed and older single-drug data

Krook 1979 [8]: 6 patients with Crohn's disease and 3 healthy volunteers given metronidazole, culture-based; overall fecal flora "showed no significant change during treatment" in the summary read, yet patients who responded clinically had drastically changed flora, and metronidazole-resistant Gram-negative rods persisted in one non-responder. A tiny, old, culture-based study with dose and counts not retrieved; it does not contradict the direction of the two randomized studies and is not treated as a substantial contradicting Tier A source.

Cohort test: sources set aside

Dethlefsen and Relman 2011 [15]: 3 volunteers, ciprofloxacin only, metronidazole not mentioned. Zaura 2015 [16]: clindamycin, ciprofloxacin, minocycline and amoxicillin; metronidazole not in the text read. Maurice 2013 [17]: abstract names no antibiotic; article not opened. Cui 2022 [14]: a minocycline plus metronidazole H. pylori regimen whose abstract frames findings around infection and does not isolate a drug effect. None of these is used as evidence for metronidazole.

Clinical consequence (outcome, not a nutrient)

C. difficile infection and antibiotic-associated diarrhea are outcomes. Neither label read states a C. difficile warning [1][2], and the PubMed phrasing run for metronidazole with C. difficile and fecal microbiota returned treatment guidelines and reviews, not a cohort quantifying infection after a metronidazole course. No incidence figure for metronidazole was retrieved.

Searches (red hunt)

PubMed "metronidazole gut microbiota human" (253 hits, first 50 reviewed) and "metronidazole fecal microbiota 16S treatment effect" (54 hits, first 50) surfaced the Haak, Jakobsson, Gotfred-Rasmussen, Zhou, Cui and Liou records and many mouse, dog, cat and horse studies (Tier B, not used). A third phrasing, "metronidazole faecal flora volunteers" (6 hits), surfaced Lewis 2005 and Krook 1979. Two phrasings aimed at healthy-volunteer and placebo trials were rate limited and not read, so further single-drug trials may exist.

Why red

Red hunt result. Two independent single-drug controlled human studies (one randomized, one placebo-controlled), in different populations (healthy adults, children with a parasite), measured gut bacteria directly [7][10]; both show loss or shift of gut bacteria in the expected direction, one adds measured functional change (lower short-chain fatty acids and sulphate reduction), the mechanism is named, and no substantial contradicting Tier A source was found. The multi-drug cohorts [9][11][13][12] are consistent and show larger and longer changes, but they are supportive context only and are not counted toward red, because metronidazole cannot be separated from the partner antibiotics (Jakobsson's authors attribute the main decline to clarithromycin). The single-drug effects are modest and time-limited (genus-level and mostly gone by 8 weeks in children), which is why the entry text states recovery and the framing caution below; that bounded size is a limit on interpretation, not a reason for a lower tier for expected pharmacology.

Counter-evidence held on record

Counter-evidence and confounding All sources are small or abstract-level here; the Lewis total number of volunteers and dose were not retrievable. The multi-drug sources include other antibiotics, a proton pump inhibitor (three of them) and in two cases the H. pylori infection itself, all of which change the gut community. The single-drug trial was in children with a parasite and may not transfer to adults. No null Tier A source was found for metronidazole, but none was searched beyond the PubMed phrasings listed in Sources. Framing caution This is a disturbance risk, not a deficiency. No probiotic, fiber or vitamin supplement is recommended by this entry, and nothing here says a standard course causes lasting harm. Severe or persistent diarrhea during or after a course needs clinical assessment.

Direction

Reverse (protective) in the one rat surrogate study; otherwise unmeasured.

Regulatory status

Labels contain no ATP or energy-metabolism statement [1][2].

Surrogate evidence (Tier B)

Leite 2001 [27]: isolated rat liver mitochondria, indomethacin model; metronidazole pretreatment reduced the rise in state 4 oxygen consumption and the fall in respiratory control ratio; ATP not mentioned in the abstract. Hernandez Herrera 2025 [26]: oxygen consumption was the only metabolic readout; no ATP in the abstract. Xiao 2019 [24]: "energy crisis" described for tigecycline only.

Searches

PubMed "metronidazole ATP" (57 hits; first 20 were bacterial and protozoal resistance papers, drug interaction papers and one on ecto-nucleotidases of Trichomonas, none a mammalian ATP measurement) and "metronidazole mitochondrial" (73 hits; first 20 reviewed).

Mechanism

None established for ATP.

Counter-evidence held on record

Counter-evidence and confounding Graded yellow rather than ungraded only because a Tier B oxidative-phosphorylation surrogate exists. It shares all its evidence with the Mitochondrial burden entry. Framing caution Not a finding of reduced energy production. "Low energy" claims for metronidazole have no ATP data behind them.

Direction

Neutral and indication-linked. Not an oxygen-delivery deficit (Hard Rule 2). Kept separate from ATP production (Hard Rule 7).

Regulatory status

Labels list reversible neutropenia (leukopenia) and, rarely, reversible thrombocytopenia and recommend leukocyte counts before and after therapy [1][2]; no anemia, hemolysis, methemoglobin or hemoglobin statement.

Human evidence (Tier A, single arm)

Monajemzadeh 2008 [35]: hemoglobin and hematocrit rose after metronidazole in 102 children with giardiasis (P < 0.0001; means not retrieved); see the Iron entry for design and limits.

Searches

PubMed "metronidazole hemolysis" (48 hits; first 20 contained no relevant metronidazole red-cell study) and "metronidazole anemia hemoglobin" (35 hits; H. pylori anemia papers where eradication is the exposure, the Monajemzadeh study, and a title on thrombotic thrombocytopenic purpura with metronidazole vaginal gel, not opened).

Mechanism

None identified.

Counter-evidence held on record

Counter-evidence and confounding This is a context flag from a single-arm study of children with a high baseline rate of iron-deficiency anemia (26.5%) [35]; it adds no independent evidence beyond the Iron entry. Framing caution Not an oxygen-delivery deficiency. Unusual pallor or fatigue during a course is a clinical question, not a supplement signal.