Phenytoin

Limits on what this shows

• Key negative/null findings: Korczyn et al. 1974, "Phenytoin and folate deficiency: failure to show dose dependency" (Sources [7]) - a genuine null/complicating finding recorded rather than omitted: folate lowering did not track phenytoin dose in a dose-dependent manner in this study, which qualifies but does not reverse the overall direction of effect.
• At-risk subgroups: Long-term/chronic users (enzyme induction is a cumulative effect); patients of reproductive age, given folate's additional relevance to neural-tube-defect risk in pregnancy; patients on phenytoin polytherapy with other inducing anticonvulsants.
• Counter-evidence on record: Korczyn 1974's dose-independence finding (above). Rivey, Schottelius & Berg 1984, "Phenytoin-folic acid: a review" (Sources [8]) is a review rather than new primary data and is cited for synthesis context, not as an independent Tier A primary source.

Regulatory status

FDA label, Warnings and Precautions: "Macrocytosis and megaloblastic anemia have occurred [and] these conditions usually respond to folic acid therapy" (Sources [1]).

Population evidence

Tier A, source 1: the label itself. Tier A, source 2: Linnebank et al. 2011, Ann Neurol (Sources [2]) - prospective cohort, n=2,730 antiepileptic-drug-treated patients, untreated epilepsy patients, and healthy controls; "treatment with carbamazepine, gabapentin, oxcarbazepine, phenytoin, primidone, or valproate was associated with lower mean serum folate levels," phenytoin directly named and in the cohort; low folate correlated with elevated homocysteine and MCV; supplementation normalized levels in 95% of treated patients within three months. Tier A, source 3: Xu et al. 2019, Medicine (Baltimore) (Sources [3]) - "Effects of phenytoin on serum levels of homocysteine, vitamin B12, folate in patients with epilepsy," phenytoin-specific title. Additional phenytoin-specific primary literature identified by title (mechanism confirmed, full abstract not independently re-verified this review; see access note): Hoffbrand & Necheles 1968, "Mechanism of folate deficiency in patients receiving phenytoin" (Sources [4]); Billings 1984, "Interactions between folate metabolism, phenytoin metabolism, and liver microsomal cytochrome P450" (Sources [5]); Maxwell et al. 1972, "Folate deficiency after anticonvulsant drugs: an effect of hepatic enzyme induction?" (Sources [6]).

Mechanism

Named and consistent across sources: hepatic CYP450 enzyme induction accelerates folate catabolism/turnover (Billings 1984 ties this directly to liver microsomal CYP450; Maxwell 1972's title frames the same question). This is the same induction mechanism stated on the label.

Counter-evidence held on record

This is a drug-induced catabolic effect, not simply low dietary folate intake, and correction requires more than diet alone. Folic acid supplementation itself lowers phenytoin serum levels by enhancing its hepatic clearance (Lewis et al. 1995, "Phenytoin-folic acid interaction," Sources [9]; Rivey et al. 1984, Sources [8]); starting or increasing folate repletion without phenytoin level monitoring risks breakthrough seizures. Any folate repletion in a phenytoin patient should be done with prescriber oversight and phenytoin level follow-up, not as unsupervised self-supplementation.

Limits on what this shows

• Key negative/null findings: None identified specific to phenytoin this review; the broader AED-bone-health literature (Bartl 2007, Sources [16]; Hamed 2016, Sources [17]) is class-level synthesis rather than a phenytoin-specific null finding.
• At-risk subgroups: Chronic/long-term users (per label); institutionalized or reduced-sun-exposure patients, a classic confounder for any AED-bone-health finding (immobility and reduced sun exposure independently lower vitamin D, so confounding by indication/institutionalization is presumed until excluded per Hard Rule 6 - the Krishnamoorthy and Moro-Alvarez titles specifically target phenytoin monotherapy, which mitigates but does not eliminate this concern); pediatric patients (Vestergaard 2015).
• Counter-evidence on record: Confounding-by-indication limitation stated above is the main qualifier on this entry; no contradicting Tier A evidence specific to phenytoin was identified.

Regulatory status

FDA label, Warnings and Precautions: enzyme induction "may enhance the metabolism of vitamin D and decrease vitamin D levels, which may lead to vitamin D deficiency, hypocalcemia, and hypophosphatemia"; chronic use "associated with decreased bone mineral density...and bone fractures" (Sources [1]).

Population evidence

Tier A, source 1: the label. Tier A, source 2: Krishnamoorthy et al. 2010, Natl Med J India (Sources [10]) - "Early predisposition to osteomalacia in Indian adults on phenytoin or valproate monotherapy," a phenytoin-monotherapy-specific cohort, not class-extrapolated. Tier A, source 3: Moro-Alvarez et al. 2009, Eur Neurol (Sources [11]) - "Bone disease induced by phenytoin therapy: clinical and experimental study," phenytoin-specific title. Supporting class-level/mechanistic context (Tier B/C, not independently counted toward the red threshold): Kulak et al. 2004, "Bone mineral density and serum levels of 25-OH vitamin D in chronic users of antiepileptic drugs" (Sources [12]), cohort of chronic AED users with phenytoin among the drugs studied per title, though the phenytoin-specific subgroup size was not independently re-confirmed by abstract this review; Xu et al. 2006, "Intestinal and hepatic CYP3A4 catalyze hydroxylation of 1α,25-dihydroxyvitamin D3: implications for drug-induced osteomalacia" (Sources [13]) - mechanistic, CYP3A4-inducer-class level, not phenytoin-isolated, marked as class extrapolation; Andress et al. 2002 (Sources [14]) and Vestergaard 2015 (Sources [15]), both class-level AED bone-health reviews/cohorts.

Mechanism

Named: hepatic (and per Xu 2006, intestinal) CYP450 induction, principally via CYP3A4, accelerates catabolic hydroxylation of 25-OH and 1,25-(OH)2 vitamin D, lowering circulating active vitamin D and driving secondary hypocalcemia/hypophosphatemia and, with chronic exposure, decreased bone mineral density.

Counter-evidence held on record

Decreased bone mineral density and fracture are clinical outcomes of chronic vitamin D/calcium disturbance, not themselves evidence of vitamin D deficiency (Hard Rule 2) - the graded claim here is the measured drop in circulating vitamin D and the label-stated hypocalcemia/hypophosphatemia, not the fracture outcome itself. Any vitamin D repletion dose should stay under the NIH Office of Dietary Supplements adult tolerable upper intake level (4,000 IU/day) and be directed by a prescriber who can pair it with calcium/phosphate/ALP monitoring, not taken as an unsupervised high-dose regimen.

Limits on what this shows

• Key negative/null findings: McNinch & Tripp 1991, a two-year British Isles prospective surveillance study of hemorrhagic disease of the newborn (Sources [23]), is general-population HDN surveillance in which anticonvulsant exposure is one of several risk factors; phenytoin is not isolated as its own cohort in the title, so this source is marked class/general-context rather than counted toward the red threshold.
• At-risk subgroups: Infants of mothers on phenytoin (or other enzyme-inducing anticonvulsants) during pregnancy, specifically in the first 24 hours of life if prophylaxis is missed.
• Counter-evidence on record: The mechanism is explicitly stated by the label to be preventable with prophylactic vitamin K, which is itself supportive of, not contradictory to, the causal claim (a validated, falsifiable, correctable marker per Hard Rule 3).

Regulatory status

FDA label, Warnings and Precautions: "A potentially life-threatening bleeding disorder related to decreased levels of vitamin K-dependent clotting factors may occur in newborns exposed to phenytoin in utero. This drug-induced condition can be prevented with vitamin K administration to the mother before delivery and to the neonate after birth" (Sources [1]).

Population evidence

Tier A, source 1: the label. Tier A, source 2: Keith, Gundberg & Gallop 1980, "Phenytoin therapy and hemorrhagic disease" (Sources [18]), a phenytoin-specific title. Tier A, source 3: Allen, Ogden, Bentley & Jung 1980, "Fetal hydantoin syndrome, neuroblastoma, and hemorrhagic disease in a neonate" (Sources [19]) - hydantoin is phenytoin's drug class name and this case report is phenytoin-specific. Supporting era-matched case literature (Tier A/B, title-confirmed, not independently re-opened for full abstract this review): Evans, Forrester & Discombe 1970, "Neonatal haemorrhage following maternal anticonvulsant therapy" (Sources [20]); Stevenson & Gilbert 1970, "Anticonvulsants and hemorrhagic diseases of the newborn infant" (Sources [21]). Harden et al. 2009, AAN/AES practice parameter on women with epilepsy (Sources [22]), a clinical-practice-guideline synthesis, Tier A as guideline-level evidence.

Mechanism

Named: hepatic enzyme induction crosses the placenta and enhances fetal/neonatal catabolism of vitamin K-dependent clotting factors (II, VII, IX, X), producing a coagulopathy distinct from classical hemorrhagic disease of the newborn from simple vitamin K insufficiency, but corrected by the same remedy.

Counter-evidence held on record

This finding applies specifically to maternal/fetal/neonatal exposure, not to vitamin K status in non-pregnant adult phenytoin users, where no comparable depletion signal was identified this review. Vitamin K has no NIH Office of Dietary Supplements tolerable upper intake level established, but maternal and neonatal vitamin K prophylaxis is a clinical decision for the treating obstetric/pediatric team, not a self-directed supplementation choice.

Limits on what this shows

• Key negative/null findings: None identified this review.
• At-risk subgroups: Same chronic/long-term users as the Vitamin D entry; pediatric patients, given calcium's role in bone accretion during growth.
• Counter-evidence on record: Single-source limitation for clinical (non-in-vitro) quantification is the explicit reason this entry is not graded red.

Regulatory status

FDA label, same Warnings and Precautions sentence as the Vitamin D entry: enzyme induction "may enhance the metabolism of vitamin D and decrease vitamin D levels, which may lead to vitamin D deficiency, hypocalcemia, and hypophosphatemia" (Sources [1]).

Population evidence

Tier A, label-reported laboratory/clinical association from the same mechanism as Vitamin D. No independent magnitude, sample size, or confidence interval for calcium specifically was located in the accessible label text, and no second Tier A clinical source quantifying a phenytoin-specific calcium effect was confirmed this review - the reason this entry is bounded to orange rather than red, consistent with the Overlap map's disclosure that it shares its primary evidence with Vitamin D.

Mechanism

Named, secondary to the Vitamin D entry: reduced active vitamin D impairs intestinal calcium absorption, producing hypocalcemia; von Borstel Smith et al. 2007, "Effects of phenytoin and carbamazepine on calcium transport in Caco-2 cells" (Sources [24]), is supportive in-vitro mechanistic (Tier B) evidence for a direct effect on intestinal calcium transport as well, independent of the vitamin D pathway.

Counter-evidence held on record

This tracks with the vitamin D mechanism above, not an independent dietary calcium problem. Calcium supplementation decisions should follow correction of the underlying vitamin D status and stay under the NIH Office of Dietary Supplements adult tolerable upper intake level (2,500 mg/day for ages 19-50; lower/higher by age group), under prescriber guidance rather than as independent self-supplementation.

Limits on what this shows

• Key negative/null findings: None identified this review beyond the ambiguity already noted in the Linnebank abstract regarding which drugs specifically lowered B12 within the combined cohort.
• At-risk subgroups: Same long-term/chronic users as the Folate entry; patients with elevated homocysteine on testing (a validated, falsifiable functional marker per Hard Rule 3, directly measured in both Xu 2019 and Linnebank 2011).
• Counter-evidence on record: The ambiguity in the Linnebank 2011 abstract about which specific drugs drove the B12 effect is the explicit reason this entry is bounded to orange rather than red, despite two Tier A, phenytoin-indexed sources.

Regulatory status

Not separately addressed on the accessible label text; the label's folic-acid statement does not extend to B12 specifically.

Population evidence

Tier A, source 1: Xu et al. 2019, Medicine (Baltimore) (Sources [3]) - "Effects of phenytoin on serum levels of homocysteine, vitamin B12, folate in patients with epilepsy," phenytoin-specific title measuring B12 directly. Tier A, source 2: Linnebank et al. 2011 (Sources [2]) - n=2,730 cohort; abstract confirms "several [antiepileptic] drugs lowered vitamin B12 levels, while valproate surprisingly increased them," with phenytoin among the drugs studied in this cohort, though the abstract-level text retrieved this review does not isolate which specific drugs drove the B12-lowering subset separately from the folate-lowering list. Kathiravan, Kavitha & Shanthi 2021, Sci Rep (Sources [25]) and Allen 1970, "Folate and vitamin B12 in epilepsy" (Sources [26]) are additional phenytoin-indexed, title-confirmed sources not independently re-opened for full abstract this review.

Mechanism

Named but indirect: B12 and folate share one-carbon metabolism, and B12 lowering in this population is plausibly secondary to, or co-occurring with, the folate-catabolism mechanism rather than a separately named CYP450 pathway specific to B12.

Counter-evidence held on record

This is a measured serum biomarker finding, not a report of clinical B12-deficiency neuropathy or anemia. B12 has no NIH Office of Dietary Supplements tolerable upper intake level established (water-soluble, no observed toxicity), but repletion should still be confirmed by a serum B12/homocysteine panel and directed by a prescriber, particularly since B12 and folate status are usually assessed together in this population.

Limits on what this shows

• Key negative/null findings: The study's own authors explicitly state: "It remains to be determined whether a reduction in serum carnitine values in patients receiving anticonvulsant therapy is of clinical consequence" - a null/undetermined clinical-significance finding recorded directly rather than omitted.
• At-risk subgroups: Pediatric patients specifically (the only population studied); patients on phenytoin polytherapy combinations, where the magnitude trended larger in some groups than phenytoin monotherapy alone.
• Counter-evidence on record: Polytherapy-inclusion limitation: several of the eight groups combine phenytoin with phenobarbital, valproic acid, or carbamazepine, so the phenytoin-monotherapy-specific effect size cannot be fully isolated from this single study alone. No adult data were located.

Regulatory status

Not addressed on the accessible FDA/DailyMed label text.

Population evidence

Tier A, single source: Hug et al. 1991, J Pediatr (Sources [27]) - observational study, 471 pediatric patients across eight treatment groups (four monotherapy including phenytoin, four polytherapy including phenytoin combinations). Full abstract confirmed this review: total carnitine declined 23-48% across groups, free carnitine 9-45%, short-chain fatty-acid carnitine 46-64%, long-chain fatty-acid carnitine 6-29%; "the means of all four carnitine constituents were significantly reduced in all treatment groups (except for free carnitine in four groups)"; approximately 12% of phenytoin-treated patients showed reduced total carnitine specifically.

Mechanism

Not separately named in the retrieved abstract beyond the general anticonvulsant-therapy association; plausibly related to hepatic enzyme induction affecting carnitine biosynthesis/renal handling, consistent with phenytoin's broader induction profile, but this specific mechanistic link was not independently confirmed this review.

Counter-evidence held on record

This is a biochemical finding in children whose clinical relevance the original study explicitly leaves open - it is not evidence of carnitine-deficiency symptoms (hypoglycemia, cardiomyopathy, hypotonia) in phenytoin-treated patients generally. No NIH Office of Dietary Supplements tolerable upper intake level exists for carnitine; any pediatric carnitine testing or supplementation decision belongs with the treating pediatric neurologist, not as self-directed supplementation.

Limits on what this shows

• Key negative/null findings: None independently confirmed this review; this absence itself reflects incomplete verification rather than a confirmed null result.
• At-risk subgroups: Not independently confirmed this review.
• Counter-evidence on record: None independently confirmed this review. The grading is capped at yellow specifically because direction, magnitude, and population evidence were confirmed only at the title/metadata level, not via full abstract or primary-text review, a tooling limitation disclosed per Step 4 rather than smoothed over.

Regulatory status

Not addressed on the accessible FDA/DailyMed label text.

Population evidence

Tier A by cohort (phenytoin-indexed titles), title-level confirmation only, full abstracts not independently re-opened this review: Palm & Hallmans 1982 (Sources [32]); Taylor & Ghose 1986 (Sources [33]); Kuzuya, Hasegawa, Shimizu & Nabeshima 1993, "Effect of anti-epileptic drugs on serum zinc and copper concentrations in epileptic patients" (Sources [37]); Sarangi et al. 2014 (Sources [34]). Liu et al. 1998 (Sources [28]) measured Cu/Zn-SOD, which requires zinc as a cofactor, as part of its trace-element panel. Animal data (Tier B): Palm et al. 1985 (cat), Palm et al. 1986 (rat) (Sources [35], [36]).

Mechanism

Not independently confirmed this review; plausible routes include altered intestinal absorption, hepatic sequestration, or urinary loss under chronic anticonvulsant therapy, consistent with the trace-element-handling disturbance documented for copper above, but the specific chain for zinc was not verified from primary text this review.

Counter-evidence held on record

Direction is not established with confidence this review. No zinc supplementation recommendation follows from a yellow-tier entry under this protocol; a revision should re-open each title-confirmed source above by full abstract before this entry is reconsidered for a higher or more directionally specific grade. See the Overlap map - this entry shares evidence with Copper and, via Cu/Zn-SOD, with Oxidative stress.

Limits on what this shows

• Why no grade: Dual condition met on the evidence located: no Tier B/C evidence was found to support even a yellow grade, and no circulating claim specific to phenytoin was confirmed (biotin appears in some generic multi-drug anticonvulsant-depletion consumer lists, more often associated with valproate's carboxylase-inhibition mechanism than with phenytoin specifically, but this review could not independently verify the content or prevalence of any phenytoin-specific biotin claim before being rate-limited).

Channels tried

PubMed search-listing page for the combined query "phenytoin AND (biotin OR carnitine)" returned 48 total results; the biotin-relevant hits reviewed were limited to an unrelated analytical-chemistry paper using phenytoin as a hapten in a biotin-labeling reagent, with no drug-nutrient content. No biotin-specific phenytoin mechanism, case report, or class-extrapolation source was identified among the results reviewed. Tooling constraint disclosed: this review's remaining WebFetch budget was exhausted by session-wide rate limiting (repeated HTTP 429 responses, confirmed independently via the agent-proxy status log's HTTP 403 policy-denial entries against pubmed.ncbi.nlm.nih.gov and related hosts) before a dedicated, biotin-only query (separate from the combined biotin/carnitine query above) or a dedicated consumer-claim-landscape search could be run. This is disclosed explicitly so a future reviewer treats this as a probable gap compounded by a genuine tooling limitation, not a fully exhaustive negative result, and re-runs a biotin-only query first.