Atomoxetine

Limits on what this shows

• Key null/negative findings: At 1 and 10 µM, mitochondrial mass did not decrease (it increased at 10 µM) and autophagy activation was not observed; overt membrane-potential loss required 50 µM, roughly 10-fold the approximate 5 µM human plasma concentration the authors cite.
• At-risk subgroups: Same CYP2D6 poor-metabolizer plausibility note as the Oxidative stress entry; unstudied.

Regulatory status

Not addressed in the FDA label.

Population evidence

Same Corona et al, 2019 study (see Oxidative stress entry for design). Mitochondrial membrane potential was significantly decreased only at 50 µM. Mitochondrial mass increased at 10 µM and decreased at 50 µM, a nonmonotonic pattern. OXPHOS Complex II and Complex IV protein content decreased at 50 µM (Complex I, III, and V were not reported in the material reviewed). The autophagy marker ratio LC3-II/LC3-I increased at 20 and 50 µM and decreased at 10 µM. Cell death was 1 to 7.7% at 1 to 5 µM versus 32 to 35% at 20 to 50 µM.

Mechanism

Tier B, in vitro, drug-specific; biphasic and dose-dependent. The authors propose therapeutic concentrations may support normal or even favorable mitochondrial dynamics while supratherapeutic concentrations impair membrane potential and respiratory-complex content.

Counter-evidence held on record

No animal or human in vivo mitochondrial-function data (biopsy, imaging, or functional respiratory testing) were located for atomoxetine in this review. The clearest impairment findings required concentrations roughly 10-fold above typical therapeutic plasma levels. This is not evidence that standard clinical dosing impairs mitochondrial function, and no supplement or monitoring test is validated to detect or prevent this in patients.

Limits on what this shows

• Key null/negative findings: The lowest tested concentration (1 µM, below typical clinical plasma range) showed no significant ROS increase. Ascorbic acid co-treatment reduced high-dose toxicity, indicating the effect is at least partly redox-mediated and modifiable in this in vitro system.
• At-risk subgroups: None established in humans. CYP2D6 poor metabolizers reach 5- to 10-fold higher plasma concentrations than extensive metabolizers (see PK basis), which would place them further into the ROS-positive concentration range tested here; this is a plausible but entirely unstudied at-risk subgroup, not a confirmed finding.

Regulatory status

Not addressed in the FDA label; no labeled oxidative-stress or redox monitoring parameter.

Population evidence

Corona et al, 2019, Scientific Reports: differentiated human SH-SY5Y neuroblastoma (neuron-like) cells exposed to atomoxetine at 1, 5, 10, 20, and 50 µM for 7 days during differentiation. Cytosolic ROS (DHE fluorescence) increased significantly at 5, 10, 20, and 50 µM, with no significant increase at 1 µM. Mitochondrial ROS (MitoSOX) was elevated at 10, 20, and 50 µM. Co-treatment with 50 µM ascorbic acid attenuated cell death from the highest atomoxetine doses. The authors note human clinical plasma concentrations reach approximately 5 µM, within the ROS-positive range tested.

Mechanism

Tier B, drug-specific, in vitro. The paper does not fully elucidate the redox mechanism beyond the ROS measurements themselves, and explicitly describes a biphasic, dose-dependent response, with the authors suggesting therapeutic concentrations may act differently than supratherapeutic ones.

Counter-evidence held on record

No clinical human study of oxidative-stress markers (malondialdehyde, total antioxidant capacity, total oxidant status, superoxide dismutase, glutathione) in atomoxetine-treated patients was located despite searches across multiple phrasings and databases; this entry rests entirely on one in vitro paper. Do not use this finding to justify antioxidant supplementation. The ascorbic acid effect was a cell-culture rescue experiment at a defined in vitro concentration, not a human dosing study, and no dose recommendation follows from it.

Limits on what this shows

• Why no grade: Copper is mechanistically adjacent to this drug's own pharmacology: dopamine beta-hydroxylase, the enzyme converting dopamine to norepinephrine, is a well-established copper-dependent oxygenase. But that adjacency is two steps removed from atomoxetine's actual molecular action. Atomoxetine's direct target is the presynaptic norepinephrine transporter (NET, SLC6A2), a reuptake protein; dopamine beta-hydroxylase is a separate biosynthetic enzyme, upstream in the same catecholamine pathway, that atomoxetine does not bind or inhibit. The biochemical adjacency describes a pathway atomoxetine's downstream effects sit near, not a demonstrated effect of atomoxetine on copper status, and no source located in this review or its dedicated re-research pass tested the latter. One pharmacogenomic study (Fang et al, 2015) found dopamine-beta-hydroxylase gene variants moderate atomoxetine treatment response; per Hard Rule 4 this is the reverse direction from a depletion claim (a gene/pathway factor predicting drug response, not the drug affecting the nutrient), and it is disclosed here but not used to support any grade. A dedicated re-research pass for this revision also tested the adjacency at the class level, that is, whether any norepinephrine reuptake inhibitor, not atomoxetine alone, has been shown to alter dopamine-beta-hydroxylase activity or copper status; no such study was found either (see Tooling disclosure), so even a Tier C class-extrapolation yellow is not supported. Per Step 7, a purely theoretical, two-step mechanism with zero direct data at any tier does not clear the yellow bar, so this entry remains ungraded rather than upgraded on mechanism alone.

Tooling disclosure

Original pass: searched via the Crossref API across multiple phrasings ("atomoxetine copper," "atomoxetine copper dopamine beta hydroxylase," "atomoxetine serum copper zinc trace element children"), the PubMed search-listing page, and drugs.com's atomoxetine-specific food-interaction and side-effect pages; none surfaced a copper claim or a drug-specific copper-status study. Dedicated re-research pass for this revision: Crossref was queried again with the same three phrasings plus two new ones, "atomoxetine trace element serum children zinc copper" and the class-level "norepinephrine reuptake inhibitor dopamine beta hydroxylase activity plasma copper"; all five queries returned cleanly (no rate-limiting) and surfaced zero atomoxetine-specific or NRI-class copper findings, only unrelated trace-element literature (HIV/diarrhea, iron-deficiency anemia, breast cancer) and unrelated DBH-activity literature (schizophrenia, congestive heart failure, cold-pressor stress, none with a reuptake-inhibitor drug in the cohort). PubMed's search-listing page returned one usable result set this session, surfacing only the already-known Fang et al 2015 DBH-variant study; a follow-up PubMed query was then rate-limited. The Europe PMC REST API, Semantic Scholar Graph API, and OpenAlex API each returned persistent HTTP 429 responses on every attempt this session, including after paced retries of 60 to 100 seconds between calls, and the proxy explicitly flagged Europe PMC as not worth retrying further this session; this matches a known, confirmed organization-level access constraint on these hosts rather than a query-specific failure. DailyMed's search page also returned a 429 this session, and an Unpaywall search-endpoint query returned a 422 (malformed-request) response rather than a content-based block. General web search (WebSearch) was unavailable for the entire re-research pass because this session's web-search query budget was already exhausted before any atomoxetine-copper query could run. Given five independently successful Crossref queries (spanning direct drug-nutrient, trace-element-panel, and class-level mechanistic phrasings) plus one successful PubMed query, all returning clean negative results, this is treated as a genuine literature gap rather than a tooling-limited one, notwithstanding the confirmed blocks on the other four channels.

Limits on what this shows

• Why no grade: The amphetamine and methylphenidate-family literature includes discontinuation trials linking ferritin to treatment continuity and add-on iron-supplementation trials in iron-deficient children, but every such trial located in this series enrolled patients on a stimulant, not on a nonstimulant NET inhibitor, and none included atomoxetine. Because atomoxetine's mechanism (norepinephrine-transporter inhibition, with no direct dopamine-releasing or presynaptic-catecholamine-synthesis action the way stimulants have) is pharmacologically distinct from that class, this dossier does not carry the stimulant class's iron findings forward. No independent atomoxetine-iron study, and no consumer claim linking atomoxetine specifically to iron, was located.

Tooling disclosure

Original pass: searched via the Crossref API ("atomoxetine iron ferritin ADHD," "atomoxetine iron ferritin") and PubMed-adjacent channels; both queries returned only unrelated general iron-metabolism and ferritin-biology literature with zero atomoxetine-specific hits. Dedicated re-research pass for this revision: Crossref was queried again with "atomoxetine ferritin iron deficiency," returning 20 results (dissertations and clinical papers on ferritin cutoffs, iron-deficiency anemia diagnosis, iron absorption biology), none mentioning atomoxetine or any ADHD medication, reconfirming the original null result independently. A PubMed fetch for "atomoxetine AND ferritin" was rate-limited (HTTP 429) before returning results this session. The Europe PMC REST API, Semantic Scholar Graph API, and OpenAlex API each returned persistent HTTP 429 responses across paced retries this session (60 to 100 seconds between attempts), the same organization-level access constraint documented in the Copper entry's tooling disclosure. WebSearch was unavailable for this pass, its session-wide query budget having already been exhausted. The independently reconfirmed, clean negative Crossref result is treated as sufficient to support a genuine gap rather than a channel-specific artifact, consistent with this dossier's explicit instruction not to extrapolate the stimulant-class iron/ferritin literature onto this nonstimulant NET inhibitor.

Limits on what this shows

• Key null/negative findings: The study used a chronic-stress disease model, not healthy animals, so it demonstrates atomoxetine attenuating stress-induced gut-barrier injury, not a baseline effect on an undisturbed gut; whether atomoxetine affects microbiome composition or barrier function in the absence of stress is untested. A second located study (an astaxanthin/ADHD-rat model using atomoxetine as a positive control) explicitly reported no gut-microbiome data at all for its atomoxetine arm, a confirmed reporting gap rather than a null finding.
• At-risk subgroups: None established in humans.

Regulatory status

Not addressed in the FDA label.

Population evidence

Lapmanee et al, 2025, International Journal of Molecular Sciences: male rats under 14-day repeated restraint stress (2 hours/day), randomized to atomoxetine 10 mg/kg, escitalopram 10 mg/kg, cannabidiol 20 mg/kg, CBD-lipid-nanoparticles 20 mg/kg, stressed vehicle, or unstressed control. Versus stressed vehicle, atomoxetine significantly reduced serum FITC-dextran (an intestinal-permeability marker, p<.001), restored short-chain fatty acid levels (p<.01), restored polyamine levels (p<.01), and increased hippocampal and colonic tight-junction proteins ZO-1 and occludin (p<.01 to p<.001). Atomoxetine did not fully restore osteocalcin, a bone marker outside this category's scope.

Mechanism

Not established beyond a general gut-brain-bone axis framing by the study's authors. No 16S sequencing or other microbial-composition data were reported, only permeability, short-chain fatty acid, polyamine, and tight-junction protein markers; this is gut-barrier-function evidence, not community-composition evidence.

Counter-evidence held on record

Single animal study, one dose, one stress paradigm, no microbial-composition sequencing performed. The direction found is reassuring rather than concerning and should not be over-read as atomoxetine protecting gut health in general use, only as evidence against a disruption narrative in this specific stress model. No probiotic, prebiotic, or short-chain-fatty-acid supplement recommendation follows from this entry.

Limits on what this shows

• Key null/negative findings: The source paper explicitly does not report ATP, oxygen-consumption, or respiration-rate data; this entry should not be read as a directly measured energy-output finding.
• At-risk subgroups: None established.

Regulatory status

Not addressed in the FDA label.

Population evidence

Same Corona et al, 2019 source. Complex II (succinate dehydrogenase, feeding electrons into the respiratory chain) and Complex IV (cytochrome c oxidase, the terminal electron acceptor immediately upstream of ATP synthase) protein content both decreased at 50 µM. A targeted re-check of the paper confirmed no ATP assay, oxygen-consumption-rate (Seahorse-type) measurement, or phosphocreatine/energy-charge data were reported anywhere in it. No other located source measured ATP or cellular energy output with atomoxetine at any tier.

Mechanism

Tier B, indirect and inferential. Complex II and IV are structural components of the electron-transport chain that feeds ATP synthase; reduced protein content of these complexes is consistent with, but does not directly demonstrate, reduced ATP-synthesis capacity.

Counter-evidence held on record

This is the weakest-evidenced of the three mitochondrial-adjacent entries, an inference from protein content rather than a functional energy measurement; see Overlap map. No clinical fatigue or "low energy" complaint on atomoxetine should be attributed to impaired ATP production on the basis of this entry. The drug's labeled fatigue/somnolence adverse effects have not been mechanistically linked to this in vitro finding by any source located here, and no supplement is recommended.

Limits on what this shows

• Why no grade: The FDA label documents increased heart rate (mean +5.0 beats/min in extensive metabolizers, +9.4 beats/min in CYP2D6 poor metabolizers) and blood pressure, with 5 to 10% of patients showing potentially clinically important changes (heart rate ≥20 beats/min or blood pressure ≥15 to 20 mmHg). These are direct consequences of increased synaptic norepinephrine from NET inhibition, atomoxetine's intended mechanism of action, not a hemoglobin, red-cell-mass, or arterial-oxygenation finding. No case report, cohort study, or label safety table describing anemia, altered hematocrit, or impaired tissue oxygen delivery attributable to atomoxetine was located, and no consumer claim linking atomoxetine to oxygen-carrying capacity or "low energy from anemia" was found to rebut as white.

Tooling disclosure

Original pass: searched via the Crossref API across two independent phrasings ("atomoxetine hemoglobin erythropoiesis oxygen," "atomoxetine anemia hemoglobin case report"), both returning only unrelated general hematology literature. The FDA label's cardiovascular and postmarketing safety sections were also reviewed directly and contain no anemia or hematocrit signal. Dedicated re-research pass for this revision: Crossref was queried again with "atomoxetine hemoglobin hematocrit erythropoiesis," returning 20 results (capillary blood collection methods, anemia in kidney disease, pregnancy and postmenopausal hematology, dengue hemorrhagic fever), none mentioning atomoxetine, independently reconfirming the original null result. A PubMed fetch attempting "atomoxetine hemoglobin OR hematocrit OR anemia" returned a result set that PubMed's automatic term handling did not restrict to atomoxetine (general anemia and erythrocyte literature with no atomoxetine hits); a corrected single-term retry was rate-limited (HTTP 429) before it could complete. The Europe PMC REST API, Semantic Scholar Graph API, and OpenAlex API each returned persistent HTTP 429 responses across paced retries this session, the same organization-level constraint documented in the Copper entry's tooling disclosure, and DailyMed's search page also returned a 429 on this pass. WebSearch was unavailable, its session-wide query budget already exhausted. The independently reconfirmed, clean negative Crossref result plus the direct FDA label review are treated as sufficient to support a genuine gap.