Diflunisal

Limits on what this shows

• Null findings: Nadanaciva 2013 [18] tested diflunisal in an isolated-mitochondria, hepatocyte and zebrafish panel but no diflunisal-specific result was retrievable. Mahmud 1996 [19] did not name diflunisal and is excluded. No human mitochondrial or clinical hepatotoxicity measure was sought or found.
• At-risk subgroups: Unknown. Renal impairment raises exposure (half-life prolonged up to ten-fold, glucuronides accumulate) [27], which would raise burden if the in vitro effect operates in vivo.

Regulatory status

No mitochondrial statement appeared in the label extracts [1].

Experimental evidence (Tier B, diflunisal tested)

McDougall 1983 [14]: diflunisal induced a progressive loss of respiratory control in tightly coupled rat liver mitochondria starting at 3.3 micromolar, with stimulated state 4 respiration on succinate or glutamate plus malate (abstract opening; later sentences were not retrieved, and the summary says similar effects were seen across animal species). Chavez 1985 [15]: diflunisal uncouples oxidative phosphorylation by carrying hydrogen ions through the inner membrane, short-circuiting protons, with decreased membrane potential and intramitochondrial calcium release. Horie 1998 [16]: in freshly isolated rat hepatocytes, diflunisal caused significant LDH leakage. Kasuya 2012 [17]: diflunisal and one other NSAID inhibited medium-chain and long-chain acyl-CoA synthetase activity in mouse liver and brain mitochondria, with paired IC50 values from 64.7 to 284.4 micromolar in liver and 57.3 to 114.0 micromolar for palmitoyl-CoA formation in brain (assignment of each value to diflunisal not resolvable).

Mechanism

Named: protonophore-like uncoupling of the inner mitochondrial membrane [14][15]. A class-level review notes that NSAIDs interact with phospholipids and uncouple oxidative phosphorylation, impairing GI barrier function; diflunisal was not named in the text seen [33].

Exposure relevance

A DHFR paper describes diflunisal Ki of 34 micromolar as well below peak plasma levels [23]. Because diflunisal is over 99% bound to plasma proteins [1], free concentrations reaching mitochondria are lower than total levels; this is a caution of this dossier, not a statement of the sources.

Hard Rule 2 check

This is the drug's own pharmacology, not a nutrient deficiency; direction is increased mitochondrial burden, with no nutrient implied.

Counter-evidence held on record

Counter-evidence and confounding All data are isolated organelles, cells or mouse and rat tissue; no human study links diflunisal at therapeutic doses to mitochondrial injury. Protein binding makes the effective concentration uncertain. The same experiments underlie the ATP entry and are not independent confirmations (see Overlap map). Framing caution for the consumer surface This is an in vitro pharmacology finding, not evidence of mitochondrial damage in people and not a reason to take mitochondrial or antioxidant supplements.

Limits on what this shows

• Null findings: The plasma finding is itself a null: no increase in redox-active plasma iron [21]. A PubMed listing for "diflunisal oxidative" returned a nickel-diflunisal complex paper (no abstract text retrieved) and no human oxidative stress study.
• At-risk subgroups: Unknown. The plasma experiment used iron-overloaded patients, a special population.

Regulatory status

No oxidative stress or antioxidant statement in the label extracts [1].

Experimental evidence (Tier B)

Sihn 2018 [21]: diflunisal efficiently carries iron into HeLa or HepG2 cells, inducing iron-catalyzed oxidative stress; it did not raise redox-active iron in plasma from iron-overloaded patients. Horie 1998 [16]: cytotoxicity (LDH leakage) in rat hepatocytes, without an oxidant marker in the excerpt.

Mechanism

Named for the cell finding: iron loading of cells with Fenton-type oxidant generation (the source's stated "iron-catalyzed" mechanism). Whether this occurs at therapeutic doses or without added iron was not established.

Counter-evidence held on record

Counter-evidence and confounding A single in vitro paper supports this entry and its relevance to a person on a normal diet is untested. The iron-delivery finding also feeds the Iron entry and is not independent of it. Framing caution for the consumer surface Not evidence that diflunisal lowers antioxidant capacity and not a reason to take antioxidant or iron supplements.

Limits on what this shows

• Null findings: Controlled volunteer studies at 250 mg twice daily show little blood loss, and the week 2 result in Rider 1983 is null [10][7]. No ferritin, transferrin saturation, serum iron or iron deficiency anemia outcome was found for diflunisal in the channels tried (PubMed listings "diflunisal iron", "diflunisal iron deficiency" and "diflunisal ferritin" returned 429 and could not be completed; "diflunisal iron absorption" returned HTTP 404, consistent with zero hits but not confirmed; "diflunisal anemia" returned only the rabbit paper and a hypersensitivity case series; label extracts name no iron).
• At-risk subgroups: Older adults, women in long-term care (Lapane subgroup), prior ulcer or GI bleeding, anticoagulant or indomethacin co-use, longer exposure, and the higher volunteer dose (500 mg twice daily) [1][9].

Regulatory status

Label: upper GI ulcers, gross bleeding or perforation in approximately 1% of patients treated for 3 to 6 months and about 2% to 4% treated for one year; anemia statement as in the Oxygen delivery entry; platelet effects are quantitatively less than aspirin, shorter and reversible; iron itself is not mentioned [1].

Population evidence

SOS meta-analysis [11]: pooled RR for upper GI complications for diflunisal 4.37 (95% CI 1.07 to 17.81), a very wide interval, against ibuprofen 1.84 and piroxicam 7.43 in the same list. Perez-Gutthann 1997 [12]: nested case-control (1,377 cases, 10,000 controls) in which diflunisal had the highest odds ratio among the NSAIDs compared and ibuprofen the lowest; no diflunisal estimate was in the extract. Lapane 2001 [13]: nursing-home cohort of 125,516 residents, GI hospitalization RR 6.08 (95% CI 2.27 to 16.26) in women exposed to diflunisal. Volunteer fecal blood loss: De Schepper 1978 [7][8], 250 mg twice daily did not significantly increase 51Cr-measured blood loss, with 0.32 to 0.53 mL per day for diflunisal in the reconstructed abstract, while aspirin did; Green 1981 [9], 250 mg twice daily no platelet effect, 500 mg twice daily minimal inhibition of collagen-induced aggregation, moderately prolonged bleeding time and increased fecal blood loss; Rider 1983 [10], 36 volunteers for 14 days, diflunisal 1,000 mg/day showed a slight but significant rise at week 1 and no significant change at week 2, aspirin 4.0 g/day a significant rise. All of these measure bleeding or GI complications, not iron.

Mechanism

Present. Prostaglandin synthetase inhibition with reversible platelet inhibition (label) [1]. In vitro, diflunisal does not raise redox-active iron in plasma from iron-overloaded patients, but an Fe(II)-diflunisal complex or diflunisal efficiently carries iron into HeLa and HepG2 cells and induces iron-catalyzed oxidative stress [21]; the relevance to human iron status is unknown and the extract did not say whether the free drug or the preformed complex was responsible.

Hard Rule 2 and 6 checks

GI bleeding and anemia are outcomes, not iron deficiency. Confounding by indication and channeling is presumed in the observational studies; the SOS interval is wide, the Perez-Gutthann estimate was not read, and the Lapane result is a female subgroup. Aspirin, indomethacin and other NSAID results from sibling dossiers were not carried over.

Counter-evidence held on record

Counter-evidence and confounding Three Tier A observational sources point to elevated GI complication risk, while the volunteer fecal-blood-loss trials found little at 250 mg twice daily. The two do not conflict: volunteers were healthy and treated for days to two weeks, whereas the observational studies are older, sicker, longer-exposed populations. Neither measures iron stores. Framing caution for the consumer surface GI bleeding on diflunisal is a drug-toxicity pathway. Iron supplementation is not supported by any diflunisal iron-status data; unexplained anemia or black stool needs prescriber evaluation for the bleeding source.

Limits on what this shows

• Null findings: No copper status study for diflunisal was found in the channels tried (PubMed listings "diflunisal copper" returned 429 twice, then "diflunisal Cu(II)" succeeded; Crossref title queries returned only compendial records).
• At-risk subgroups: Unknown.

Regulatory status

No copper statement in the label extracts [1].

Experimental evidence (Tier B, chemistry)

Fountoulaki 2011 [22]: copper(II) diflunisal complexes with nitrogen-donor ligands, crystal structures with deprotonated diflunisal coordinated through carboxylato oxygens, with DNA and serum albumin binding and electrochemical study. PubMed listing "diflunisal Cu(II)" also showed later complex papers (2024, 2025) that were not opened.

Mechanism

Metal coordination is a property of the molecule; whether it changes copper absorption, distribution or excretion in vivo was not tested.

Hard Rule 7 note

Entered independently of Iron; the iron-carrier finding belongs to the Iron entry. Copper is not merged with it.

Counter-evidence held on record

Counter-evidence and confounding Synthetic complexes made in organic solvents say little about physiology; this grade reflects the rule that a Tier B source forces at least yellow, not a measured copper effect. Framing caution for the consumer surface A chemistry observation. It does not indicate copper loss or a need for copper.

Limits on what this shows

• Null findings: PubMed listings "diflunisal microbiome" and "diflunisal microbiota" each returned HTTP 404 (consistent with zero hits, not confirmed); "diflunisal bacteria" and "diflunisal antibacterial" returned 429. No fecal 16S, diversity, dysbiosis or C. difficile study of diflunisal was found. This is partly tooling-limited.
• At-risk subgroups: Unknown.

Regulatory status

No microbiome statement in the label extracts. Diarrhea, constipation and nausea are listed adverse reactions above 1%, which are symptoms, not microbiome measures; little or no diflunisal appears in feces [1].

CLASS EXTRAPOLATION

Cell Host Microbe 2024 [31]: in clinical cohorts and aspirin-treated mice, aspirin suppressed Parabacteroides goldsteinii (opening of the abstract only). Reviews of NSAID enteropathy describe microbiota as a critical factor [32] and NSAID GI damage as involving mitochondrial uncoupling and barrier impairment [33]; none names diflunisal in the text seen. Cohort test: aspirin only, diflunisal not named; Tier C, capped at yellow.

Mechanism

Not established for diflunisal. A mouse study reports diflunisal anti-bacterial and anti-inflammatory activity against S. aureus [34], which is not a gut observation.

Counter-evidence held on record

Counter-evidence and confounding The class source is an aspirin study; diflunisal is nearly completely absorbed, so luminal exposure is mainly during transit. The microbiome entry is an unknown, not an observed effect. Framing caution for the consumer surface Theoretical class-level risk only; not a reason to take probiotics on the basis of this entry.

Limits on what this shows

• Null findings: The randomized trial in familial amyloid polyneuropathy (130 patients, 2 years, diflunisal versus placebo) is the only placebo-controlled long exposure retrieved, but only its abstract was read and no hemoglobin, platelet or renal data from it were retrieved [5]. Healthy-volunteer fecal blood loss studies at 250 mg twice daily found no significant increase [7][8].
• At-risk subgroups: Older adults with ATTR cardiomyopathy and reduced renal function; people on warfarin (label: prolonged prothrombin time) or with prior GI bleeding; combination with indomethacin (label: fatal GI hemorrhage) [1].

Regulatory status

Label: anemia is sometimes seen in patients receiving NSAIDs including diflunisal, possibly from fluid retention, occult or gross GI blood loss, or an incompletely described effect on erythropoiesis; hemoglobin or hematocrit should be checked in long-term users with signs of anemia; hemolytic anemia and thrombocytopenia are listed among adverse reactions occurring in fewer than 1 in 100 patients [1]. A second label pass did not return the hemoglobin sentence [2]; the first is relied on.

Population evidence

Huang 2025 [6]: systematic review and meta-analysis, 12 studies (Peiro-Aventin 2024, Castano 2012, Quarta 2015, Siddiqi 2022, Choi 2020, Choi 2022, Gilad 2021, Mints 2019, Hanson 2018, Ikram 2018, Koyama 2015, Lohrmann 2020), 539 patients, mean age about 70, follow-up about 12 months, all on diflunisal. Post-treatment versus baseline within the diflunisal group: hemoglobin MD -0.32 g/dL (95% CI -0.57 to -0.07, I2 0%, p 0.01); eGFR MD -5.55 mL/min/1.73 m2 (CI -9.75 to -1.34, I2 0%); platelets MD -11.61 x 10^3 per microliter (CI -20.12 to -3.11); creatinine increase not significant (p 0.07). Pooled discontinuation 24% (CI 15% to 36%, I2 72%); reasons listed as renal impairment 21%, GI impairment 13%, bleeding 6%, fluid retention 6%, with the denominator unclear (the four figures sum to 46%, above the pooled 24%, so they cannot all be percentages of all patients). Number of studies behind each of the hemoglobin, eGFR and platelet analyses was not stated in the text read.

Mechanism

Present but unresolved. The label names fluid retention, GI blood loss and an effect on erythropoiesis [1]. Bleeding pathway: [11][12][13]. Hemolysis: pregnant rabbits at 40 and 60 mg/kg/day developed severe hemolytic anemia after a marked drop in erythrocyte ATP, a single 180 mg/kg dose produced anemia through gestation day 15, and the anemia and ATP depletion were unique to the rabbit among the species examined [20]. Reduced renal function (eGFR fell in the same meta-analysis) is a candidate contributor, which is this dossier's inference, not a statement in the source. The meta-analysis authors gave no mechanistic discussion of the hemoglobin fall and did not separately quantify GI bleeding, iron deficiency or hemodilution.

Hard Rule 2 and 6 checks

Hemoglobin is the measured parameter of this category, but a hemoglobin fall is an outcome and is not iron deficiency (Rule 2). Confounding is not excluded: the design has no control group, the populations are older cardiac amyloidosis patients whose disease, fluid status and renal function change over time, and the authors flag retrospective bias [6]. The orange grade rests on Tier A, drug-specific, measured and consistent (I2 0%) data bounded to one population and about a year. An editor who regards uncontrolled before/after data as insufficient should move this entry to yellow.

Counter-evidence held on record

Counter-evidence and confounding The data are pre/post within treated patients; disease progression, age, fluid shifts and reduced renal function could each lower hemoglobin without a direct drug effect on red cells. The rabbit hemolysis and ATP fall were, per the abstract, unique to the rabbit among the species examined [20], so it is weak support for a human effect. Hemolytic anemia in people appears in the label only as a rare (under 1 in 100) adverse reaction. Framing caution for the consumer surface This is not an iron or vitamin deficiency and not a reason to start iron. A low hemoglobin on diflunisal calls for a prescriber to look for blood loss, kidney function and fluid status, with iron indices measured first.

Limits on what this shows

• Null findings: None specific to diflunisal retrieved. PubMed listings for "diflunisal mitochondria" and "diflunisal oxidative" were read; no ATP assay in humans or intact tissue was among them.
• At-risk subgroups: Unknown.

Regulatory status

No ATP or energy statement in the label extracts. Fatigue or tiredness appears among adverse reactions above 1% incidence, which is not an ATP measurement [1].

Experimental evidence (Tier B)

Clark 1984 [20]: severe maternal hemolytic anemia in rabbits at 40 and 60 mg/kg/day following a dramatic decrease in erythrocyte ATP. McDougall 1983 [14]: loss of respiratory control (the uncoupling signature) in rat liver mitochondria; the abstract text read broke off before the state 3 result, so an ATP synthesis rate was not retrieved. Horie 1998 [16]: the cytotoxic NSAIDs and diphenylamine as a group extensively decreased hepatocellular ATP, with diflunisal reported as causing significant LDH leakage; an ATP value for diflunisal alone is not in the excerpt.

Mechanism

Hepatocyte and mitochondrial ATP: uncoupling [14][15]. Erythrocyte ATP in rabbits: mechanism not stated in the text read. Mature red cells have no mitochondria (general physiology, not from a retrieved source), so that finding cannot be a mitochondrial uncoupling effect.

Hard Rule 7 and 2 notes

Entered separately from Mitochondrial burden and Oxygen delivery/utilization; it shares sources with both (see Overlap map). A drug effect on ATP is not a nutrient deficiency.

Counter-evidence held on record

Counter-evidence and confounding Group-level ATP statements do not isolate diflunisal; the rabbit erythrocyte ATP depletion was, per the abstract, unique to the rabbit among the species examined and came from 40 to 60 mg/kg/day dosing [20]. Framing caution for the consumer surface These are laboratory findings. They do not show reduced cellular energy in people taking diflunisal, and fatigue should not be attributed to ATP loss.