Rivaroxaban

Limits on what this shows

• Key negative or null findings: The dose-response in the one directly verified study is non-monotonic: the highest tested concentration was protective, not more damaging, which undercuts even a simple "more drug, more mitochondrial harm" reading of this single in vitro dataset.
• At-risk subgroups: None identifiable from human or clinical data; the concentration-dependence described above has no established human correlate, so no patient subgroup can be flagged as at elevated mitochondrial-burden risk from this evidence.
• Dose/duration: Not applicable to any labeled human dose; the supportive finding used supratherapeutic in vitro concentrations only.

Regulatory status

Neither the FDA label nor the EMA SmPC names mitochondrial toxicity, mitochondrial myopathy, or any related finding for rivaroxaban.

Population evidence

Samiei et al., Drug Research 2019 (PMID 31499543): isolated rat kidney mitochondria exposed to rivaroxaban at 1.4, 2.8, and 5.6 mM. At 1.4 and 2.8 mM, reactive-oxygen-species generation rose along with mitochondrial swelling, membrane-potential collapse, and cytochrome c release. At 5.6 mM the direction reversed, with the abstract describing a protective effect against oxidative stress and mitochondrial toxicity at that highest concentration. Two further rivaroxaban-specific studies (Zekri-Nechar et al. 2022, PMID 36250331, mitophagy in high-glucose human coronary endothelial cells; Zamorano-Leon et al. 2020, PMID 32173474, mitochondrial-associated proteins in human abdominal aortic aneurysm tissue) were identified by title and listing but their full text/abstract could not be independently opened for this review; they are named to document that this line of inquiry exists, not cited for any specific finding.

Mechanism

Mechanism-screen class E (mitochondrial interference), present only as a laboratory finding: direct mitochondrial membrane and electron-transport-chain effects observed in isolated organelles exposed to rivaroxaban concentrations in the millimolar range. Therapeutic rivaroxaban plasma exposure is in the micromolar range, roughly three orders of magnitude below the concentrations that produced damage in this study.

Counter-evidence held on record

The isolated-mitochondria study is Tier B (in vitro, this drug specifically) but at concentrations roughly 1,000-fold above clinical plasma exposure, and its own highest tested concentration reversed the toxicity signal rather than extending it. No human, clinical, or even whole-animal mitochondrial-function study of rivaroxaban at therapeutic exposure was located. Framing caution: an isolated-organelle finding at millimolar concentrations is not evidence of mitochondrial harm at a therapeutic dose, and must not be presented to a consumer as a reason to take a mitochondrial-support supplement alongside rivaroxaban.

Limits on what this shows

• Key negative or null findings: The same human study found no significant change in total antioxidant capacity or in three of four other redox/vascular enzyme markers, so the single positive finding (MDA) did not generalize across the full panel measured in that trial.
• At-risk subgroups: Not applicable in the depletion sense. The population studied (Stage B heart failure) has elevated baseline oxidative stress from the underlying cardiac condition itself, which is a plausible confounder for any pre-post change unless compared against an untreated arm; this study had no control group.
• Dose/duration: 20 mg/day for two months in the one human study located; no data at other doses or longer duration for this parameter.

Regulatory status

Neither the FDA label nor the EMA SmPC names oxidative stress, antioxidant capacity, or any redox biomarker as an adverse effect or monitoring parameter for rivaroxaban.

Population evidence

Dilanchian et al., ARYA Atherosclerosis 2025 (PMID 40893721): a human interventional pre-post study, 39 Stage B heart-failure patients (ages 30–95) naive to rivaroxaban, given 20 mg/day for two months. Malondialdehyde (MDA) fell significantly (p<0.001). Total antioxidant capacity, homocysteine, paraoxonase-1, and arylesterase activity showed no significant change. This is the single Tier A source identified for this category: rivaroxaban specifically, in a human cohort, with a directly measured redox parameter.

Mechanism

Not a nutrient mechanism. The proposed pathway in this literature is anti-inflammatory and endothelial: rivaroxaban's PAR-1/PAR-2 signaling effects are reported in animal models (isoproterenol-induced cardiac injury, sunitinib- and cisplatin-induced organ injury, endothelial glycocalyx damage) to reduce reactive-oxygen generation secondary to reduced thrombin-receptor-mediated inflammation, not by supplying or sparing an antioxidant nutrient.

Counter-evidence held on record

The human study is single-arm with no placebo or untreated comparator, so the MDA reduction cannot be separated from regression to the mean, concurrent cardiac care, or the natural course of Stage B heart failure over two months. Multiple rat and mouse studies (isoproterenol cardiac injury, cisplatin and sunitinib organ injury, endothelial glycocalyx damage, TLR4/Myd88/NF-κB modulation in a depression model) report the same directional finding — reduced oxidative stress markers with rivaroxaban — but only titles and abstracts were reviewed for several of these and full text could not be independently opened for this review. Framing caution: this is not a nutrient-status finding in either direction. It documents rivaroxaban's own reported pleiotropic anti-inflammatory effect, not a change in any vitamin, mineral, or antioxidant nutrient level, and per hard rule 1 must never be presented to a consumer as a nutritional deficiency (or nutritional benefit) claim.

Limits on what this shows

• Key negative or null findings: ROCKET AF, the only large rivaroxaban-vs-warfarin head-to-head trial, reported GI bleeding only as a bleeding-safety endpoint; the FDA label's restatement of its adverse-reactions data includes no ferritin, iron-study, or nutrient-deficiency endpoint for either treatment arm.
• At-risk subgroups: Patients with pre-existing anemia, GI angiodysplasia, older age, concomitant antiplatelet or NSAID use, and renal impairment all independently raise both bleeding risk and baseline iron-deficiency prevalence — a clear confounding-by-indication and confounding-by-comorbidity picture (hard rule 6).

Regulatory status

Neither the FDA label nor the EMA SmPC lists iron deficiency, non-bleeding anemia, or a ferritin change as an adverse reaction distinct from bleeding itself.

Population evidence

A single-center Australian retrospective study of iron-deficiency-anemia patients on DOACs undergoing endoscopy (Ching et al., Intern Med J 2025, n=178) analyzed DOACs as a pooled class without a rivaroxaban-specific breakdown. A second paper whose title suggested per-drug stratification of DOAC GI-bleeding characteristics could not be opened for this review (access blocked); no statistic from it is used here.

Mechanism

Indirect only: chronic occult or overt GI blood loss depletes iron stores over time in any patient on any anticoagulant. Rivaroxaban has no independent mechanism affecting iron absorption, transport, or storage beyond whatever bleeding it causes.

Counter-evidence held on record

Active-comparator limitation: ROCKET AF compared rivaroxaban to warfarin, not to no anticoagulation, so even the observed bleeding-rate difference cannot be used to infer what either drug does to iron status relative to an untreated baseline. No placebo-controlled or untreated-comparator iron-status data exist for rivaroxaban. Framing caution: per hard rule 2, gastrointestinal bleeding is a clinical outcome, not a nutrient-deficiency diagnosis. Even in a patient who develops iron-deficiency anemia while on rivaroxaban, the correct clinical attribution is anticoagulant-associated GI blood loss requiring medical evaluation, not "rivaroxaban depletes iron." This entry documents that the question has been asked in the literature, not that it has been answered.

Limits on what this shows

• Key negative or null findings: Same as the original entry: ROCKET AF's adverse-reactions data include no ferritin or iron-study endpoint.
• At-risk subgroups: Same as the original entry: baseline anemia, older age, renal impairment, and concomitant antiplatelet or NSAID use.

Regulatory status

Unchanged from the original entry: no labeled iron-deficiency, ferritin, or iron-study finding for rivaroxaban.

Population evidence

No new source was found in this expansion beyond Ching et al. 2025 (PMID 39718804, DOACs pooled as a class, not rivaroxaban-stratified) already cited in the original yellow-tier entry. A dedicated iron-specific search for this category (ferritin, transferrin saturation, hepcidin, iron-deficiency anemia incidence) returned the same literature already on file; no rivaroxaban-stratified Tier A iron-status study exists.

Mechanism

Identical to the original entry: indirect only, via labeled GI bleeding, with no independent mechanism by which rivaroxaban affects iron absorption, transport, or storage.

Counter-evidence held on record

See the original "GI bleeding and iron status" entry above for the full counter-evidence record (active-comparator limitation of ROCKET AF, confounding by comorbidity). It is not repeated in full here to avoid the appearance of two independent findings where one exists. Overlap control: this entry and the original GI-bleeding-iron entry are the same underlying finding, counted once in the overlap map below. Do not treat these as two independent pieces of evidence when this dossier is summarized downstream.

Limits on what this shows

• Key negative or null findings: Not applicable; nothing was measured to be negative or null. This is an absence of any study, not a study that found no effect.
• At-risk subgroups: Cannot be identified without a mechanism or a population study to bound one.
• Why no grade: Both conditions for UNGRADED are met: (a) no Tier B or Tier C evidence exists for rivaroxaban or for any direct factor Xa inhibitor class member, and (b) no circulating consumer claim linking rivaroxaban to copper was found to rebut. Per the locked grading key, UNGRADED carries no governance state and must be printed as a gap, not folded into a reassuring "no effect" statement.

Regulatory status

Neither the FDA label nor the EMA SmPC mentions copper.

Population evidence

None. PubMed and Crossref searches combining rivaroxaban with copper returned no relevant result; the only matches were coincidental (a contraceptive-safety review mentioning copper IUDs in anticoagulated patients, and unrelated case reports of Wilson's disease, a hereditary copper-metabolism disorder, in patients who happened to also be taking rivaroxaban for an unrelated indication). Neither bears on rivaroxaban's effect on copper status.

Mechanism

Absent. None of the nine mechanism-screen classes (A–I) applied in this protocol point toward a copper-specific pathway for rivaroxaban; there is no absorption, chelation, renal-wasting, or utilization mechanism connecting factor Xa inhibition to copper handling.

Counter-evidence held on record

None exists to hold, in either direction. This is distinct from the red/orange "no findings" entries above, which document a search that found nothing supporting harm; here, no search (Tier A through D) surfaced anything at all, including no consumer claim to audit. Framing caution: absence of evidence is not evidence of absence. This entry must never be rendered to a consumer as "copper status is unaffected by rivaroxaban" — the correct statement is that copper status in relation to rivaroxaban has not been studied.

Limits on what this shows

• Key negative or null findings: No rivaroxaban-specific human gut-microbiome study (16S sequencing, metagenomics, or dysbiosis marker) was located in this review; the only drug-specific data are from one disease-specific mouse model, which limits generalization to a normal-physiology human user.
• At-risk subgroups: Confounding by indication applies directly here: the HSOS model itself substantially perturbs gut microbiota and intestinal barrier function independent of any drug, so the finding cannot be generalized to rivaroxaban users without liver sinusoidal injury.

Regulatory status

Neither the FDA label nor the EMA SmPC names microbiome, gut flora, or intestinal bacterial composition as an effect of rivaroxaban.

Population evidence

Liu et al., Frontiers in Microbiology 2025 (PMID 40919202, DOI 10.3389/fmicb.2025.1607131): a mouse model of monocrotaline-induced hepatic sinusoidal obstruction syndrome (HSOS). Rivaroxaban treatment improved liver-injury markers and increased beneficial taxa (including Allobaculum and Pediococcus ) while decreasing others; the protective effect was abolished by concurrent antibiotic treatment, indicating the gut microbiota is mechanistically necessary for rivaroxaban's benefit in this specific disease model, not evidence rivaroxaban itself is harmful to microbiome composition.

Mechanism

Mechanism-screen class G (microbiome disruption affecting endogenous synthesis) was tested and, in the one drug-specific study located, ran in the protective/favorable direction rather than the disruptive direction the class was screened for. A separate, reverse-direction Tier C source (Chen et al. 2022, Frontiers in Pharmacology, PMID 35401180, title: "Changes in the Gut Microbiota May Affect the Clinical Efficacy of Oral Anticoagulants") points the other way: microbiota composition affecting anticoagulant drug efficacy, not the drug affecting the microbiome. Full text of the Chen review could not be independently opened for this review and it is not cited for any specific statistic.

Counter-evidence held on record

The only drug-specific evidence located runs opposite to a "risk" framing and is confined to a specific hepatic disease model with antibiotic-confirmed microbiota-dependence; it cannot be extrapolated to a general claim that rivaroxaban changes gut flora in an average patient. No dysbiosis, diarrhea-pattern, or GI-microbiome-specific adverse-event signal appears in the FDA label's adverse-reactions data. Framing caution: this entry must not be read either as "rivaroxaban harms the microbiome" or as "rivaroxaban improves gut health" for a general user; the only data available are model-specific and mechanistic, not a general-population finding in either direction.

Limits on what this shows

• Key negative or null findings: No rivaroxaban-specific hemoglobin or oxygen-delivery endpoint exists in the label or in any study located for this review, positive or negative.
• At-risk subgroups: Same subgroup as the bleeding/iron entries: patients with pre-existing anemia, older age, renal impairment, or concomitant antiplatelet/NSAID use, for whom bleeding-driven anemia (and thus reduced oxygen-carrying capacity) is most plausible if it occurs.

Regulatory status

The FDA label's ROCKET AF adverse-reactions table (cited above) reports GI bleeding rates but no hemoglobin, hematocrit, oxygen-saturation, or VO2-type endpoint specific to rivaroxaban.

Population evidence

None found that directly measures oxygen delivery or utilization in rivaroxaban-treated patients. A mouse study (Imano et al. 2018, J Pharmacol Sci, PMID 30055890) examined rivaroxaban's effect on cardiac remodeling under experimentally imposed intermittent hypoxia; this tests whether rivaroxaban modifies the cardiac response to externally induced low oxygen, not whether rivaroxaban itself changes oxygen delivery, and is noted only to document that it was reviewed and found off-target for this category.

Mechanism

Indirect only: chronic GI blood loss lowers hemoglobin, which lowers blood oxygen-carrying capacity, exactly the same underlying mechanism already documented in the iron entries above. No independent mechanism by which factor Xa inhibition itself alters erythropoiesis, hemoglobin oxygen affinity, or mitochondrial oxygen utilization was identified.

Counter-evidence held on record

This entry rests entirely on a mechanistic bridge from the already-documented GI-bleeding signal (hard rule 2 of the base protocol: a clinical outcome, bleeding, is not itself a nutrient or parameter measurement). No independent oxygen-delivery data exist for rivaroxaban in either direction. Framing caution: this is a theoretical extension of a labeled bleeding signal, two mechanistic steps removed (bleeding→anemia→reduced oxygen-carrying capacity), not a direct or measured finding, and must never be presented to a consumer as evidence rivaroxaban impairs oxygen delivery.

Limits on what this shows

• Key negative or null findings: The same non-monotonic reversal noted under mitochondrial burden applies directly to this parameter: the highest concentration tested did not extend the SDH decline seen at lower concentrations.
• At-risk subgroups: None identifiable; no human or clinical correlate of this in vitro finding exists.
• Dose/duration: Not applicable to any labeled human dose.

Regulatory status

Neither the FDA label nor the EMA SmPC names ATP production, energy metabolism, or electron-transport-chain function as an effect of rivaroxaban.

Population evidence

Samiei et al. 2019 (PMID 31499543, the same study cited under mitochondrial burden): succinate dehydrogenase (complex II of the electron transport chain) activity fell at 1.4 and 2.8 mM rivaroxaban and did not extend that decline at 5.6 mM, mirroring the same reversal seen in the ROS and membrane-potential measurements from that study. No PubMed search combining rivaroxaban with ATP, mitochondrial respiration, or oxidative phosphorylation returned any additional result.

Mechanism

Mechanism-screen class E, the same entry as mitochondrial burden: complex-II inhibition would, if it occurred at a clinically relevant concentration, blunt oxidative ATP synthesis. It was observed only at millimolar concentrations roughly three orders of magnitude above therapeutic plasma exposure.

Counter-evidence held on record

Identical limitation to the mitochondrial-burden entry: supratherapeutic in vitro concentration, non-monotonic dose-response, no whole-animal or human confirmation. This entry is graded separately per this expansion's requirement that oxygen delivery/utilization and ATP production not share a single grade, but it draws on the same underlying dataset as mitochondrial burden and is not additional independent evidence; see the overlap map below. Framing caution: this in vitro complex-II finding must not be presented to a consumer as evidence that rivaroxaban lowers cellular energy production at a therapeutic dose.