Amoxicillin

Limits on what this shows

• Null findings: Sailer 2025 as above. The replicate count is very small (N = 2 to 4), so the null is weak. The Lemna minor plant study seen in the listing found amoxicillin alone not toxic (not used).

Direction

Possible drug toxicity (excess burden) in animal models; neutral in the one human-cell respiration test. Not a nutrient deficiency.

Regulatory status

Not mentioned in the label text read [1][2].

Population evidence

None in humans. Sailer 2025 [22] (Tier B, amoxicillin alone, 0.01 to 2 mg/mL, N = 2 to 4 biological replicates, permeabilized HEK 293T cells, high-resolution respirometry): no detectable effect on leak respiration, N-linked or NS-linked oxidative phosphorylation and no time-dependent effect; a slight inhibition of S-linked electron transfer capacity and a slight Complex IV decrease were noted, but the authors state amoxicillin "did not alter any measured respiratory parameters". Membrane potential was not assessed for amoxicillin (solvent interference); gentamicin and ciprofloxacin did affect respiration in the same system. Zebrafish embryo exposure study [24] (Tier B, abstract only): amoxicillin exposure increased mortality and malformations dose- and time-dependently and was accompanied by structural mitochondrial damage and reduced ATP; concentrations and magnitudes not retrieved. Rat liver mitochondria [25] (combination product): amoxicillin-clavulanate opened the permeability transition pore in vitro and in vivo at 30 mg/kg (3.4-fold; 10 mg/kg no effect) with cytochrome c release and raised ATPase activity. A prenatal mouse study reporting mitochondrial swelling in alveolar type II cells was seen as a listing line only [37]. Kalghatgi 2013 [23] (Tier C for amoxicillin): the abstract states quinolones, aminoglycosides and beta-lactams caused mitochondrial dysfunction and ROS overproduction in mammalian cells; ampicillin was the beta-lactam named in a press summary; amoxicillin was not tested and ampicillin-specific data were not read.

Mechanism

Named only in animal abstracts: in the zebrafish model a long noncoding RNA promoted ubiquitination-mediated degradation of pyruvate carboxylase, impairing mitochondrial metabolic function; in rat liver, pore opening with cytochrome c release [24][25]. Not shown in human tissue.

Cohort test

CLASS EXTRAPOLATION (aminopenicillin/beta-lactam) for Kalghatgi: amoxicillin was not in that cohort. The rat study used amoxicillin-clavulanate, so it is Tier B/C for amoxicillin alone. Capped at yellow; no human source exists, so orange is not available.

Counter-evidence held on record

Counter-evidence The only direct human-cell test is null and small. The positive findings are in embryos, rodents and a combination product at doses whose human relevance was not assessed, and the zebrafish and rat results come from abstracts only. Framing caution Theoretical and non-human. Do not present amoxicillin as mitochondrially toxic in people, and do not recommend mitochondrial or antioxidant supplements from this entry.

Limits on what this shows

• Null findings: Sailer 2025: no significant effect of amoxicillin on respiration in permeabilized human cells [22]. No human null for oxidative markers found.

Direction

Excess oxidative burden (drug effect in animal models). No antioxidant nutrient was measured, so this is not a nutrient depletion, and supplementation would not be a remedy supported by this entry.

Regulatory status

Not mentioned in the label text read [1][2].

Population evidence

No human study located (PubMed listing "amoxicillin oxidative stress", top 10 records: all animal, plant or invertebrate) [37]. Tier B animal evidence with amoxicillin itself: Karaman 2019 [26], mouse testis, amoxicillin lowered reduced glutathione and raised hydrogen peroxide, with antioxidant enzymes (superoxide dismutase, catalase, glutathione reductase, G6PD, glutathione-S-transferase) significantly affected (direction not stated in the abstract); dose, route and n not retrieved. Listing-level only, abstracts not opened: recurrent amoxicillin in mice (100 mg/kg every other week for 12 weeks) with colonic oxidative stress, DNA damage and dysbiosis; recurrent exposure with testicular oxidative DNA damage; 30 mg/kg daily in mice; zebrafish at 100 mg/L for 7 days with altered antioxidant enzymes; Nrf2 activation in zebrafish gut and kidney [26]. Combination product: in rat liver, amoxicillin-clavulanate at 30 mg/kg raised malondialdehyde and activated caspase-9 and caspase-3 (10 mg/kg had no effect on the pore endpoint) [25]. Class (Tier C): Kalghatgi 2013 [23] reports that clinically relevant doses of quinolones, aminoglycosides and beta-lactams caused ROS overproduction in mammalian cells and raised blood oxidative stress markers in mice; per a press summary the beta-lactam tested was ampicillin, and amoxicillin was not tested.

Mechanism

Proposed for bactericidal antibiotics generally (mitochondrial ROS); a named mechanism for amoxicillin in mammals was not read. Direct respirometry with amoxicillin found no respiratory effect in human cells [22], which argues against a mitochondrial route there, while the rat and zebrafish studies point to mitochondrial membrane involvement [25][24].

Cohort test

CLASS EXTRAPOLATION (aminopenicillin to amoxicillin) for the Kalghatgi result: amoxicillin was not in the cohort, ampicillin was. Capped at yellow. Animal doses and repeated-exposure schedules may exceed human exposure; relevance was not assessed.

Counter-evidence held on record

Counter-evidence The animal studies are mostly reproductive-toxicity or environmental-exposure designs; several are listing-level only here. The human-relevant direct test of mitochondrial function was null. Infection and inflammation raise oxidative markers in patients independent of the drug. Framing caution Theoretical animal and class signal only. Do not present as an expected effect in people, and do not recommend antioxidant supplements.

Limits on what this shows

• Null findings: The PubMed listing for amoxicillin with iron, ferritin, copper, zinc or chelation returned no human study of amoxicillin and a trace-mineral measurement [37].

Direction

Nutrient affecting the drug (in vitro), not the drug depleting the nutrient (Hard Rule 4). The "may increase the copper pool" statement is an in vitro inference, not a measurement.

Regulatory status

Copper is not mentioned in the label text read [1][3].

Population evidence

None in humans. A 2023 review of in vitro work states copper-mediated hydrolysis of ampicillin, amoxicillin and cephalexin at lower-than-physiological pH, a spectroscopic study (Bozic 2018, cited in the review, not opened) proposing amoxicillin-copper coordination through a side-chain amine and an amide nitrogen and suggesting the antibiotics "may increase the Cu pool in body fluids", reduced activity of copper-amoxicillin complexes against S. aureus and of copper-complexed amoxicillin against E. coli, an older report of improved activity against B. subtilis and E. coli, and the reviewers' statement that results are "inconsistent" [35]. No concentrations were given in the review.

Mechanism

Coordination chemistry is shown in vitro. Physiological relevance (plasma copper, ceruloplasmin binding, absorption or excretion) is not shown.

Counter-evidence held on record

Counter-evidence The primary in vitro studies were not opened and the review itself calls the results inconsistent. Copper status was not measured in anyone. Framing caution Not a depletion claim and not a reason to take copper. Any copper supplement would carry its own NIH upper limit; none is recommended and no dose appears in this dossier.

Limits on what this shows

• Why no grade: Dual condition met: no Tier B or C source was found to support even a yellow grade, and no circulating consumer claim on iron and amoxicillin was found to rebut. Hemolytic anemia is a separate outcome entry (Oxygen delivery/utilization), not an iron finding. This is a disclosed gap, not a finding of safety; re-check with PubMed access restored.

What was searched

FDA and DailyMed label text: iron appears only as a capsule colorant (iron oxides); no iron interaction is listed [1][3]. PubMed listing "amoxicillin AND (iron absorption OR ferritin OR copper OR zinc OR chelation)": ten records, none a human study of amoxicillin with an iron or ferritin outcome [37]. Drugs.com interaction page for amoxicillin with ferrous sulfate: the fetched page returned unrelated drugs and no usable statement. Retailer page for amoxicillin: names no nutrient [36]. The H. pylori eradication papers seen in a vitamin listing involve amoxicillin as treatment, not exposure.

Channel outcomes

Open-web search budget was exhausted before an iron-specific phrasing was run. Two further PubMed iron and hemoglobin phrasings ("amoxicillin iron deficiency anemia"; "amoxicillin serum iron OR hemoglobin OR anemia adverse") returned 429 and were not read. Europe PMC and OpenAlex search endpoints returned 429; Crossref title searches were fuzzy and returned nothing on topic. This is a mixed result: no record found on the channels that worked, plus tooling limits on others. A thin literature is likely but not established.

Limits on what this shows

• Null findings: Vrieze 2014 [11]: 20 men with metabolic syndrome, amoxicillin 500 mg three times daily for 7 days versus vancomycin; the abstract states "Amoxicillin did not affect any of these parameters" (fecal diversity, bile acids, insulin sensitivity). Reijnders 2016 [12]: 57 obese prediabetic men, 7 days; amoxicillin did not change gut microbiota composition at 7 days or at 8 weeks versus placebo (secondary summaries; full text not read). Reijnders D. is an author of the Vrieze paper, so the two trials are related groups and may not be fully independent. Zaura also found no diversity change.
• At-risk subgroups: Longer courses (7 versus 3 days in one cohort; 3 months in another), repeated courses, and people with an already disturbed gut. One of six volunteers in the earliest study had not recovered by 2 months. Infants were studied only with a combination exposure.

Direction

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

Regulatory status

Label: "Treatment with antibacterial agents alters the normal flora of the colon leading to overgrowth of C. difficile"; C. difficile-associated diarrhea "has been reported with use of nearly all antibacterial agents"; "Amoxicillin may affect the intestinal flora, leading to lower estrogen reabsorption" [1][2]. Diarrhea is among the most common reactions (more than 1%) [3].

Population evidence

De La Cochetiere 2005 [7]: 6 healthy adults, amoxicillin 500 mg three times daily for 5 days, TTGE of dominant fecal 16S rRNA. Similarity to own day 0: 93 to 99% on day 1, average 73% on day 3 (range 62 to 82%), 74% on day 4 (range 46 to 94%), 88% on day 30 (65 to 95%), 89% on day 60 (66 to 98%). Five of six were at 87% or higher by day 60; one stayed below 70% (66.4%) at 2 months. A fingerprinting method, not a diversity index. Zaura 2015 [8]: placebo-controlled, amoxicillin arm 15 randomized, 14 completed. Shannon diversity and OTU richness: "no significant effect". Composition differed from placebo at week 1 only and was lost by month 1; 706 of 4,968 predicted functions (14%) differed at week 1 (272 up, 434 down); resistance-associated functions rose (9 up, 4 down). The butyrate-producer loss in that paper was reported for clindamycin and ciprofloxacin, not amoxicillin; taxon-level amoxicillin data (Table S1) were not read. Abeles 2016 [9]: healthy adults, amoxicillin or azithromycin versus a placebo household member, followed 6 months. Gut and oral diversity changed significantly after antibiotics; 7 days of amoxicillin generally reduced diversity more than 3 days. Abstract only: n and effect sizes not retrieved. The systematic review attributes persistent Lachnospiraceae reduction at 6 months to this study [13]. Dhariwal and Petersen 2023 [10]: 20 patients, 3 months of amoxicillin or placebo, shotgun metagenomics. Placebo stable; amoxicillin shifted diversity and composition and reduced short-chain fatty acid-producing species, largely recovered by 9 months; total resistance gene load and diversity rose and stayed higher at 9 months (beta-lactam resistance most affected). Abstract only: no effect sizes. Systematic review [13]: 10 studies of amoxicillin alone. Enterobacteria increased in 6; Bifidobacterium results conflicted (decreases in two studies recovered by about 3 to 4 weeks, an increase in another); most studies with follow-up returned to baseline within 2 to 4 weeks; 3 studies still showed alterations from 42 days to 6 months.

Mechanism

Present and named in the label: antibacterial activity against susceptible commensal bacteria, with overgrowth of resistant organisms (Enterobacteria in 6 reviewed studies) [1] [13].

Cohort test

Raymond 2016 [17] (cefprozil only) and Anthony 2022 [18] (azithromycin, levofloxacin, cefpodoxime) were named in the assignment; amoxicillin was not in either cohort, so no result from them is used. MacPherson 2018 [14] (amoxicillin-clavulanate, n = 70: lower alpha diversity, Enterobacteriaceae up, Lachnospiraceae and Ruminococcaceae down, back toward baseline one week after stopping) and the infant intravenous amoxicillin plus ceftazidime study [15] are combination exposures: Tier C for amoxicillin alone and not counted. A mouse study with 3, 7 and 14 day courses [16] (Tier B) found diversity change that mostly recovered, slower after 14 days.

Clinical consequence (outcome, not a nutrient)

C. difficile infection is an outcome. Meta-analyses pool penicillins (community-associated pooled OR 2.71, 95% CI 1.75 to 4.21 [19]) or penicillin combinations (hospital-acquired OR 1.45, 1.05 to 2.02 [20]) and do not report amoxicillin alone. Probiotics lowered antibiotic-associated diarrhea across antibiotics (63 RCTs, 11,811 participants, RR 0.58, 0.50 to 0.68, NNT 13 [21]); amoxicillin was not stratified.

Why orange, not red

At least one Tier A source with amoxicillin in the cohort shows a real shift, but the effect is bounded in time and size, the diversity measure was null in one cohort that did report it, and two randomized trials report no composition change. Red also needs a consistent, clinically meaningful magnitude, which no source gives as a number for amoxicillin.

Counter-evidence held on record

Counter-evidence and confounding Short courses in healthy volunteers recovered quickly in several cohorts; the Vrieze and Reijnders nulls and the Zaura diversity null argue that the effect depends on population and measure. Several included studies are small (n = 6, 14, 20), use fingerprinting or predicted functions, or report only abstract-level results here. Infection itself, diet and co-medication confound patient studies. 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 short course causes lasting harm. A new severe or persistent diarrhea during or after a course needs clinical assessment for C. difficile (the label carries a CDAD warning).

Limits on what this shows

• Null findings: None located; the absence of a cohort is an evidence gap, not reassurance.
• At-risk subgroups: Not defined for amoxicillin from the sources read.

Direction

Drug toxicity (idiosyncratic immune red-cell destruction). Not iron deficiency, not an oxygen-carrying nutrient deficit (Hard Rule 2). Kept separate from the Iron and ATP entries (Hard Rule 7).

Regulatory status

Label adverse reactions, hematologic and lymphatic: "Anemia, including hemolytic anemia, thrombocytopenia, thrombocytopenic purpura, eosinophilia", usually reversible on discontinuation [1][2][3].

Population evidence

Case report [28]: a 25-year-old woman on amoxicillin-clavulanate developed jaundice and hemolytic anemia; direct antiglobulin test positive for anti-C3 and negative for IgG, cold agglutinin and Donath-Landsteiner tests negative; improved after stopping and a short steroid course; the authors judged amoxicillin the most likely cause. A review of drug-induced immune hemolytic anemia calls it rare and says the drugs most often linked are now cefotetan, ceftriaxone and piperacillin; penicillin-type covalent binding to red-cell proteins is the most accepted mechanism [27]. No incidence figure for amoxicillin was found and no hemoglobin or hematocrit comparison of amoxicillin users with non-users was located.

Mechanism

Immune: drug-dependent antibodies after hapten-type binding (class mechanism from the review; not tested for amoxicillin specifically).

Counter-evidence held on record

Counter-evidence The case involved a combination product and elevated liver enzymes; a single case cannot establish frequency. The review does not list amoxicillin among the commonest culprits. Framing caution Report new jaundice, dark urine or unexplained pallor and fatigue during a course to a clinician. Do not read this as an iron or oxygen-delivery deficiency; iron supplements are not indicated by this entry.

Limits on what this shows

• Null findings: The respiration null above. Kept independent of Mitochondrial burden and Oxygen delivery/utilization (Hard Rule 7); the shared origin is recorded in the Overlap map.

Direction

Reduced ATP in an embryo toxicity model (drug toxicity direction); neutral at the respiration surrogate in human cells.

Regulatory status

Not mentioned in the label text read [1][2].

Population evidence

None in humans. Zebrafish embryo study [24] (Tier B, abstract only): structural mitochondrial damage and reduced ATP levels with amoxicillin exposure; magnitude and concentrations not retrieved. Sailer 2025 [22]: oxidative phosphorylation and electron transfer capacity unaffected in permeabilized HEK 293T cells; ATP was not a reported endpoint. Rat liver amoxicillin-clavulanate raised mitochondrial ATPase activity (an enzyme-activity readout, not ATP content) [25].

Mechanism

Proposed in the zebrafish abstract: impaired pyruvate carboxylase and mitochondrial metabolic function. Not shown in humans.

Counter-evidence held on record

Counter-evidence ATP content was measured only in an embryo model (abstract level); the human-cell surrogate is null; a respiration surrogate in a permeabilized cell model does not measure cellular ATP. Framing caution No energy-depletion claim is supported for people. Do not recommend ATP-support or mitochondrial supplements from this entry.