Atovaquone/Proguanil

Component

Atovaquone. Proguanil (the parent biguanide, not cycloguanil) enhances atovaquone-induced collapse of parasite mitochondrial membrane potential in a murine malaria model [11]; no human mitochondrial data exist for proguanil in the sources read.

Direction

Excess drug effect on a mitochondrial target (functional inhibition). This is not a nutrient depletion, and no nutrient was measured (Hard Rule 2). The mitochondrial action is the drug's intended pharmacology in the parasite; here it is documented in human tissue at a much higher exposure.

Regulatory status

Labels state atovaquone selectively inhibits parasite mitochondrial electron transport [1][3]. Neither the DailyMed label nor the Health Canada monograph, as summarized, reports a human mitochondrial toxicity finding; trial laboratory abnormalities were limited to transaminase elevations [1].

Population evidence

ATOM window-of-opportunity study (NCT02628080), Clinical Cancer Research 2021 [6]: open-label, non-randomized, two cohorts, 46 recruited, 30 evaluable for hypoxia PET-CT (15 per cohort). Atovaquone 750 mg twice daily with fat-containing food for 7 to 16 days (median 12). Median plasma atovaquone 62.3 uM (IQR 50.2 to 94.2). Hypoxic volume fell more than 10% in 11 of 15 treated patients (73.3%) versus 2 of 15 (13.3%) untreated; median change -28.0% (95% CI -58.2 to -4.4) versus +15.5% (95% CI -6.5 to 35.5); adjusted geometric mean hypoxic volume 55% lower (95% CI 24% to 74%; p = 0.004). Only reported adverse event: grade 3 hypertension judged unrelated. Read from the author manuscript through a summarizing fetch tool.

Supporting, lower tier

AML pilot [8] (26 enrolled, 24 evaluable, daily dosing following Pneumocystis-prophylaxis guidance, up to 1,500 mg/day): oxygen consumption fell dose-dependently in primary AML cells exposed ex vivo (n = 17, p < 0.001), a laboratory measure on patient cells, not an in-patient measurement; no adverse events attributed to atovaquone. Human cancer cell lines [7]: oxygen consumption fell 78.7% to 90.0% at 10 uM after 2 hours across eight lines. Normal human fibroblasts showed no change in oxygen consumption [9].

Mechanism

Named: inhibition of cytochrome bc1 (complex III) [1][3][7][11].

Why orange, not red

One in-patient cohort with this drug measuring a surrogate (hypoxia PET), bounded by dose (1,500 mg/day against 250 mg atovaquone per marketed adult tablet [2][3]), duration (about 12 days) and tissue (tumor). The second human source is ex vivo, and a second ATOM publication listed on PubMed (PMID 36526815) is the same cohort, not an independent one. Red needs a second independent in-patient source.

Second-source search (repair 2026-10-10)

Opened and not independent or not in patients: the EJNMMI Research spatial analysis of the same ATOM trial (NCT02628080; 30 patients, 15 treated; treated mean tumor-to-blood ratio 1.11 to 1.01, hypoxic volume fall of at least 10% in 11 of 15) [27], which the article itself does not confirm as an identical patient set and which is in any case one trial; the ARCADIAN phase I protocol (NCT04648033, 20 evaluable patients, 450 to 750 mg twice daily, FMISO PET) states oxygen consumption is not measured in patients and reports no results [28]; a mouse xenograft study (atovaquone about 50 mg/kg/day in drinking water, MRI hypoxic fraction 53.4% lower at day 7 versus 85.5% higher with vehicle, 12 treated and 9 vehicle mice, in vitro oxygen consumption ablated at 10 uM) [29], which is Tier B; and a preclinical glioma abstract [30]. No second independent in-patient atovaquone study with a mitochondrial or oxygen-consumption surrogate was found, so the grade stays orange. Human bc1 is less sensitive to atovaquone than the parasite enzyme (external values of 70 and 460 nM cited in [25] against 3.0 nM for P. falciparum), which is consistent with the bounded, high-exposure scope.

Counter-evidence held on record

Counter-evidence and confounding The cohort had lung cancer, was not randomized and used no placebo. Hypoxia PET is an indirect readout, and the authors state oxygen consumption was not measured in patients [6]. Atovaquone is more than 99% protein bound [1], so plasma totals overstate free drug. No data were retrieved on plasma concentrations or mitochondrial markers at malaria-prophylaxis exposure. A nonclinical review reports no gastrointestinal, cardiovascular or central autonomic effects of either drug at clinically relevant concentrations (abstract only) [12]. Framing caution This entry describes oncology-range exposure, about six tablets' worth of atovaquone per day. It is not a finding of mitochondrial injury at malaria doses and not a reason to add any supplement. No coenzyme Q10 or other antioxidant recommendation follows from it.

Component

Atovaquone.

Direction

Excess oxidant production in vitro (a drug effect). Not an antioxidant nutrient deficiency.

Population evidence

None in humans. In vitro [10]: H460 and FaDu cells, 30 uM atovaquone increased MitoSOX and MitoPY1 signals and reduced glutathione at 4 and 24 hours; 10 uM did not; the SOD mimetic MnTBAP rescued the ROS signal and N-acetylcysteine abolished platinum sensitization. Human MCF7 cells [9]: oxidative stress induced, with a stated 1 uM IC-50 for cancer stem cells, which the authors say is more than 50 times below average human serum concentration (no numeric serum value given in the text read). The Ashton study [7] reports no ROS measurements.

Concentration context

Median plasma atovaquone in ATOM was 62.3 uM at 750 mg twice daily [6], above the 30 uM threshold, but atovaquone is more than 99% protein bound [1] and culture media contain less protein. Plasma concentrations at malaria-prophylaxis dosing were not retrieved.

Mechanism

Candidate: mitochondrial superoxide generation (the signal was rescued by the SOD mimetic MnTBAP [10]); the source of the electron leak was not tested in the text read, and nothing was measured in patients.

Counter-evidence held on record

Counter-evidence and confounding Cancer cell lines are more redox-fragile than normal tissue, and one study reported normal fibroblasts were unaffected for oxygen consumption [9]. Framing caution In vitro tier only. It does not support any antioxidant recommendation.

Limits on what this shows

• Why no grade: Dual condition met: no Tier B or C evidence on copper and no circulating copper claim for this drug was found. A disclosed gap, not a finding of safety. Kept separate from Iron (Hard Rule 7).

What was searched

Web search "atovaquone OR proguanil serum zinc OR copper OR ceruloplasmin chelation metal binding". Results opened: a Biochemical Journal 2004 study of copper transport in P. falciparum infected red cells (parasite biology; does not mention the combination), and a phase I trial of a different copper-binding drug, ATN-224 (no relation to the combination). One result on metal-atovaquone synthetic complexes (an inorganic chemistry paper) was not opened. PubMed listing "(atovaquone OR proguanil) AND (ferritin OR iron deficiency OR zinc OR copper)": 18 results, first 10 shown, none about host copper status; they were synthetic complexes, nanoplatforms, an Aspergillus keratitis paper and tropical anaemia reviews.

Channel outcomes

The first 10 of 18 PubMed hits were read; hits 11 to 18 were not displayed. Web search budget was exhausted before further phrasings could be tried. DailyMed, Health Canada and HPRA labels as summarized contain no copper statement. Genuine gap on the channels that worked, with a residual tooling limit (hits 11 to 18, no further search phrasings).

Repair re-search (2026-10-10)

Web searches: "atovaquone proguanil zinc OR copper serum levels malaria patients treated" (nine links, none about copper or zinc status by title; none opened) and "atovaquone-proguanil plus iron OR ferrous sulfate interaction absorption" (drug-reference pages and an NCBI Bookshelf long-term-effects page, none opened). Europe PMC REST, OpenAlex search and a Crossref keyword listing errored or returned 429; PubMed listings were rate-limited and the fetch tool asked not to retry. No Tier B or C source for atovaquone, proguanil or the combination and this mineral was found. Genuine gap on the channels that opened, with a continuing tooling limit.

Limits on what this shows

• Why no grade: Dual condition met: no Tier B or C evidence on iron and no circulating iron claim was found. Disclosed gap with a residual tooling limit as for Copper; re-check if PubMed hits 11 to 18 or further phrasings become available.

What was searched

Web search "atovaquone proguanil iron deficiency OR ferritin OR hepcidin OR serum iron effect" (hits were hepcidin in pregnancy and malaria iron papers without the drug as the studied exposure, listed but not opened). Web search "atovaquone iron homeostasis ferroptosis OR iron chelation OR iron overload human cells" (generic ferroptosis reviews, not opened). The same PubMed listing as Copper. Opened labels and monograph carry no iron interaction statement.

Channel outcomes

Same as Copper. Anemia as a label adverse event is handled in the Oxygen delivery/utilization and Folate entries and is not iron deficiency (Hard Rule 2).

Repair re-search (2026-10-10)

Web searches "atovaquone-proguanil hemoglobin change OR haematological recovery treatment trial ... day 28 haemoglobin", "atovaquone-proguanil prophylaxis randomized trial haemoglobin ..." and "atovaquone-proguanil ferritin OR iron status OR hepcidin OR serum iron malaria treatment cohort". Opened: (a) the Zambian malarial-anaemia trial of atovaquone-proguanil versus sulphadoxine-pyrimethamine (arm sizes unclear in the abstract, roughly 128 per arm by failure-rate arithmetic; enrolment on packed cell volume below 21%; treatment failure 8% versus 22%, OR 3.34, 95% CI 1.54 to 7.21 for the comparator) [31], whose abstract reports no haemoglobin, ferritin or iron values and which is a disease-treatment trial with an active comparator, so it supports no iron grade; (b) the preprint of a study of iron markers in 55 controlled human malaria infection volunteers and 171 Malaysian patients [32], where atovaquone-proguanil is named only as an example rescue treatment, no iron result is stratified by drug, and there was no treated-uninfected group (excluded cohort). Hepcidin, ferritin and soluble transferrin receptor changes there are infection effects and cannot be attributed to this drug (Hard Rule 6). Anemia remains an outcome, not iron deficiency (Hard Rule 2). Genuine gap for drug-attributable iron data; the published eBioMedicine full text and the Zambian full text were not opened.

Limits on what this shows

• Why no grade: Dual condition met: the cohorts opened do not include atovaquone-proguanil, so they are neither Tier B nor usable Tier C support for a risk claim, and I do not extrapolate from artemisinin or doxycycline studies to a bc1 inhibitor. No circulating microbiome claim for this drug was traced. Diarrhea is recorded as an adverse event in the Dietary intake entry, not as a microbiome finding. An effect on gut protozoa is conceivable given the drug's antiprotozoal activity but unmeasured.

What was searched

Web searches "atovaquone proguanil gut microbiome 16S travelers malaria prophylaxis fecal microbiota", "atovaquone-proguanil antimalarial prophylaxis effect on gut microbiota composition study", "atovaquone gut microbiota dysbiosis mice OR patients 16S Pneumocystis prophylaxis microbiome" and a proguanil/cycloguanil intestinal-flora query. PubMed listing "atovaquone gut microbiota": 1 result, Israel 2025 [18] (abstract via OpenAlex): a TriNetX cohort of more than 100,000 atovaquone-proguanil users reports lower digestive cancer incidence (HR 0.49 to 0.53), with a separate fecal metagenomic analysis of 156 individuals for Toxoplasma gondii abundance; the abstract does not state that the metagenomic sample was drug-exposed, so it shows no drug effect on the microbiota. Opened cohorts that excluded this drug: French soldiers on doxycycline [19], murine artemisinin combinations [20], Kenyan infants on artemether-lumefantrine [21], travelers [22].

Channel outcomes

PubMed listing for the combined term was rate-limited once (429) and not repeated in that form. Web search budget was exhausted. The Gut Microbes full text was not opened. Genuine gap for exposure-stratified microbiome data, plus tooling limits.

Repair re-search (2026-10-10)

Web searches: "atovaquone-proguanil gut microbiota malaria chemoprophylaxis Malarone 16S rRNA travelers" (results were vivax prophylaxis, CDC and pharmacology pages; none about microbiota); "atovaquone prophylaxis gut microbiome OR intestinal microbiota stem cell transplant OR HIV Pneumocystis trimethoprim-sulfamethoxazole comparison 16S" (transplant microbiome papers); "Malarone OR atovaquone-proguanil travellers faecal microbiota OR stool microbiome OR ESBL colonization" (a travel-microbiota review that the host blocked, and other pages, none opened). Opened: an autologous stem cell transplant pilot, 21 patients, 16S sequencing, drugs measured ciprofloxacin, fluconazole and acyclovir, atovaquone and proguanil not analyzed [33] (excluded cohort); and the Nature screen of more than 1,000 marketed drugs against 40 gut bacterial strains in vitro, in whose text read atovaquone and proguanil are not named [34]; its per-drug supplementary table was not reached, so an in vitro growth effect cannot be confirmed or excluded. Two further web searches (proguanil or atovaquone in mice with 16S; non-antibiotic drug screens naming these drugs) were refused because the shared web search budget was used up. PMC transplant and travel papers were bot-challenged. Genuine gap for exposure-stratified data, plus a continuing tooling limit.

Limits on what this shows

• Null findings: No hematologic laboratory signal in the label trial laboratories [1]. G6PD deficiency data were not retrieved from a primary source. Repair search: the randomized trial of atovaquone-proguanil versus sulphadoxine-pyrimethamine in Zambian children with malarial anaemia (packed cell volume below 21% at entry) reported treatment failure (transfusion, quinine, persistent anaemia or death within 14 days) of 8% with atovaquone-proguanil versus 22% with sulphadoxine-pyrimethamine (OR 3.34, 95% CI 1.54 to 7.21 for sulphadoxine-pyrimethamine) [31]. This is a disease-treatment trial with an active comparator; the abstract read reports no haemoglobin values, so it neither shows nor rules out a drug effect on haemoglobin.

Component

Atovaquone for the tumor oxygenation finding; proguanil/cycloguanil for the cytopenia signal (see Folate).

Direction

Reverse or neutral for the oxygenation finding (more oxygen left in tissue, an intended-mechanism consequence, so not a deficiency). Anemia and cytopenia are outcomes and not iron, B12 or folate deficiency (Hard Rule 2).

Regulatory status

DailyMed [1]: trial laboratory abnormalities limited to transaminases; post-marketing neutropenia, anemia, pancytopenia. Health Canada monograph [2]: blood tests may show fewer red cells, white cells and, in severe kidney disease, platelets; red-cell counts fell transiently in dogs; methemoglobinemia in one overdose patient who also took an unspecified dose of dapsone (confounded by dapsone). HPRA SmPC [3]: anaemia listed.

Population evidence

ATOM [6]: hypoxic volume 55% lower with atovaquone (95% CI 24% to 74%; p = 0.004); the authors infer reduced oxygen consumption and did not measure it. No hemoglobin, hematocrit or arterial oxygen data were retrieved for atovaquone-proguanil users. The prophylaxis cohort in [16] (300 soldiers, 6 months) reports symptoms only.

Mechanism

Atovaquone lowers tissue oxygen consumption through complex III inhibition [6][7]. No named mechanism for a hemoglobin fall other than the cycloguanil marrow effect (Folate entry).

Counter-evidence held on record

Counter-evidence and confounding The tumor result comes from one cohort in lung cancer patients and the shared ATOM evidence is also used under Mitochondrial burden (see Overlap map). Framing caution Not an oxygen-delivery deficiency. The tumor finding is a pharmacologic effect at a much higher dose, and anemia in a returning traveler has many causes, including malaria itself.

Component

Atovaquone.

Direction

Null for ATP in the one cell-line measurement; decreased UTP is a pyrimidine-synthesis effect (DHODH), not a vitamin or mineral.

Population evidence

None in humans. Ashton 2016 [7]: FaDu cells treated 24 hours, no significant change in ATP or GTP, UTP significantly lower, CTP lower but not significantly; oxygen consumption fell 78.7% to 90.0% at 10 uM across eight lines. AML pilot [8]: extracellular acidification rate not notably changed while oxygen consumption fell in primary AML cells. Fiorillo [9]: ATP not reported in the text read.

Mechanism

Named: complex III inhibition with compensatory glycolysis (Warburg-type shift reported in MCF7 cells [9]).

Counter-evidence held on record

Counter-evidence and confounding Cancer cells are glycolytically flexible. A normal-tissue or in-patient ATP measurement was not found. Framing caution No claim of low energy or ATP depletion follows from this entry.