Isoniazid

Limits on what this shows

• Null findings: Mbala 1998 [10]: 85 children on isoniazid-containing treatment randomized to B6 or placebo, no neurological or neuropsychiatric case in either group over 6 months of treatment plus 3 months after (B6 levels not measured). Centner 2014 [9]: 116 HIV-infected in-patients starting TB treatment with B6 supplementation; plasma B6 status was normal irrespective of neuropathy, while neuropathy developed at 26 per 100 person-months (95% CI 18 to 35) and was associated with slow acetylation (HR 2.5, 95% CI 1.1 to 5.9). Everyone was supplemented, so this shows that B6 status did not track neuropathy, not that depletion is absent. In the van der Watt cohort PLP deficiency was not associated with neuropathy [6]. A systematic review [7] concludes there is no firm evidence of a direct causal link between low B6 and neuropathy. The NIH Office of Dietary Supplements B6 fact sheet does not list isoniazid among interacting drugs [48].
• At-risk subgroups: Per the label: the malnourished, alcoholics, diabetics, slow inactivators, and people on higher doses [2]. Cohort context: children above 10 mg/kg/day [8], people with TB disease (baseline PLP low in 18 of 20 [4]) and HIV [6].

Direction

Depletion, circulating and functional (plasma pyridoxal phosphate). Peripheral neuropathy is an outcome, not a B6 measurement, and is not counted toward this grade (Hard Rule 2).

Regulatory status

Both labels read state in Clinical Pharmacology that pyridoxine (vitamin B6) deficiency is sometimes observed in adults with high doses of isoniazid, and the adverse-reaction list names "Pyridoxine deficiency, pellagra" [1][2]. Peripheral neuropathy is called the most common toxic effect, dose-related, and more frequent in the malnourished and in people predisposed to neuritis (alcoholics, diabetics), and in slow inactivators. The Dosage section states that concomitant pyridoxine is recommended in the malnourished and in those predisposed [2]. This is label wording, not advice from this dossier.

Population evidence

Visser 2004 [4]: 20 patients with pulmonary TB, plasma PLP before and after 1 week of therapy including isoniazid. Baseline PLP was already low in 18 of 20 (BMI low in 10). After 7 days PLP fell (P below 0.001) in all but one patient, who had inadvertently received pyridoxine, and the fall occurred while albumin rose (P below 0.001) and CRP fell (P below 0.01). Abstract read only, so PLP values and CIs were not retrieved, and the other regimen drugs are not named. van der Watt 2015 [6]: 159 HIV-infected adults, over 50 percent had PLP under 25 nmol/L despite B-complex use, those with a history of TB and neuropathy before ART were more likely to be deficient (P=0.029), and slow or intermediate NAT2 phenotypes impacted their PLP levels; isoniazid exposure was not quantified, so this is supportive context, not an isoniazid-attributed effect. Pellock 1985 [8]: 38 children tested after 2 to 18 months of isoniazid, pyridoxine status by Tetrahymena bioassay; 5 children (13 percent) were deficient, none with definite clinical signs, and children on more than 10 mg/kg/day had a higher incidence (numbers by dose not in the abstract). There was no unexposed comparator and no pre-treatment measurement. Two further direct measurements exist and could not be read: Standal 1974 on early changes in pyridoxine status of isoniazid-treated patients [13] and a 1967 whole-blood B6 study in TB patients on high-dose isoniazid [14].

Outcome data, supporting only

Neuropathy outcomes are shown for mechanism context and are not counted. A Madras trial report [5] found that a B-complex containing a low amount of pyridoxine fully prevented neuropathy at isoniazid 12.5 to 15.2 mg/kg, while neuropathy rose with isoniazid dose and was commoner in slow inactivators. In the DART trial cohort [11], current isoniazid use carried an adjusted HR of 1.59 (95% CI 1.02 to 2.47) for a new neuropathy episode (518 of 2,895 took isoniazid).

Mechanism

Present but only secondhand here. The Tier D encyclopedia entry names inhibition of pyridoxine phosphokinase [44]. The primary mechanism paper was not retrieved, and the one review opened states only that chronic isoniazid can produce neuropathy [16]. Treat the named mechanism as plausible and unverified in this review.

Counter-evidence held on record

Counter-evidence and confounding Confounding by indication is presumed: active TB, poor intake and inflammation each lower PLP, 18 of 20 patients were low before the first dose, and the 38-child study had no baseline or control group [8]. The fall despite improving CRP argues against inflammation as the sole cause, but the cohort is 20 people over 7 days and the regimen drugs are not separated, so isoniazid alone is not isolated. Why not red: the two direct cohorts agree in direction and the label supports it, but the attributable size is not quantified in either (no PLP values retrievable; 13 percent deficient and none symptomatic), the label itself says only "sometimes", the named mechanism comes from a Tier D page, and the two other direct isoniazid B6 studies are unread. A systematic review finds no firm causal link between low B6 and neuropathy [7], and in a supplemented cohort B6 status was normal regardless of neuropathy [9]. Pyridoxine itself can cause neuropathy: a slow acetylator on isoniazid worsened until the vitamin was withdrawn [12]. Framing caution This is a bounded circulating and functional signal, not a statement that any person on isoniazid needs a B6 supplement. Excess pyridoxine is itself a neuropathy hazard [12], so supplementation would be the risk, not the remedy, outside prescriber direction. Neuropathy must not be described as B6 deficiency.

Limits on what this shows

• Null findings: No human study measuring blood or urinary niacin metabolites in isoniazid-exposed people was retrieved, so a null could not be sought on a measured parameter.
• At-risk subgroups: Maize-dependent and food-insecure diets, HIV, underweight, heavy alcohol use, lactation, and the food-scarcity season [23]; poorly nourished TB patients [24].

Direction

Depletion, functional (niacin deficiency syndrome). Pellagra is a clinical deficiency syndrome diagnosed without a niacin assay, so this is outcome-level evidence of a nutrient deficiency state. That is why it cannot reach red (nutrient not directly measured).

Regulatory status

Pellagra appears in the adverse-reaction lists of both labels read [1][2]. Niacin itself is not discussed.

Population evidence

Nabity 2022 [23]: matched case-control, Malawi, 197 patients with pellagra and 781 controls matched 4 to 1 on sex and age, analysed by conditional logistic regression. Isoniazid exposure: adjusted OR 42.6 (95% CI 13.3 to 136.6). Other independent covariates: HIV infection, food insecurity, underweight, excess alcohol use, and lactation in women. Median time from isoniazid start to rash was 5 months (IQR 3 to 7) in the food-scarcity season versus 9 months (IQR 8 to 11) at harvest (HR 7.2, 95% CI 3.2 to 16.2). Patients who stopped isoniazid and took multi-B vitamin treatment improved within 30 days (clinical management reported by the study). Ishii 1985 [24]: 8 pellagra cases among 106 TB necropsies, symptoms appearing during isoniazid in all eight, death 4 to 16 weeks later, with no unexposed comparison group. Further case reports appeared in a PubMed listing (titles only, not opened, not relied on).

Mechanism

Named by the study authors, not independently verified here: isoniazid interferes with niacin synthesis [23] and inhibits the conversion of tryptophan to niacin [24]. In a mouse model, isoniazid with a low-niacin diet produced pica (nausea-like behaviour) and lowered urinary niacin and tryptophan-pathway metabolites (1-methyl-2-pyridone-5-carboxamide, kynurenic acid, xanthurenic acid); niacin supplementation prevented the pica [25]. A proposed upstream link to the B6 entry is disclosed in the Overlap map; it is not established by a source here.

Counter-evidence held on record

Counter-evidence and confounding The OR is large but rests on a clinical diagnosis in a setting with several strong pellagra risk factors that also predict isoniazid prescribing (HIV, food insecurity). The study adjusted for them, yet residual confounding cannot be excluded in an observational design [23]. The necropsy series has no comparator [24]. Pellagra risk in people with adequate protein and niacin intake is not addressed by either source. Framing caution This is a bounded finding in undernourished and food-insecure populations on prolonged isoniazid. It is not a niacin-deficiency claim for people on isoniazid in general, and this entry carries no supplement recommendation.

Limits on what this shows

• Null and contradicting findings: Tostmann 2010 [20]: 81 Tanzanian patients with pulmonary TB, median 25-OH-D rose from 91 to 101 nmol/L over 2 months of treatment (median change +6.0, IQR -0.7 to 25.0, P=0.001) and PTH also rose; the authors suggest diet and sunlight. The regimen is not itemized in the abstract. Toppet found no 25-OH-D change on isoniazid alone [19].
• At-risk subgroups: Low sunlight exposure, marginal baseline vitamin D, children, and longer courses. These are context inferences, not findings of the retrieved studies.

Direction

Depletion of circulating vitamin D metabolites through inhibited hydroxylation (functional), not dietary deficiency. Fracture, osteomalacia and bone density are outcomes and are not graded here (Hard Rule 2).

Regulatory status

Not mentioned in the Chartwell label [2]; the Teva label sections retrieved did not mention it [1].

Population evidence

Brodie 1981 [17]: 8 healthy volunteers, isoniazid 300 mg daily for 14 days. 1,25-(OH)2D fell 47 percent (P below 0.01) after a single dose and stayed reduced. 25-OH-D fell in every subject and to below the normal range in 6 (P below 0.01). Serum calcium and phosphate fell (P below 0.001) and PTH rose 36 percent (P below 0.01). Means and CIs were not in the abstract. Toppet 1988 [19]: children with asymptomatic TB; in 9 given isoniazid alone, 1,25-(OH)2D decreased after 3 months with no significant change in 25-OH-D (magnitudes not in the abstract). Brodie 1982 [18], a combination study (rifampicin 600 mg plus isoniazid 300 mg, 14 days, 8 healthy volunteers): 25-OH-D fell 34 percent (P below 0.01), 1,25-(OH)2D fell 23 percent (P below 0.05), PTH rose 57 percent, but calcium and phosphate did not fall; in 9 TB patients 25-OH-D fell further by 6 months (P below 0.01) while 1,25-(OH)2D and PTH showed no significant change at 6 months. The isoniazid-alone contribution cannot be separated from rifampicin there. Williams 1985, a long-term patient study [22], could not be opened.

Mechanism

Present. Isoniazid inhibited hepatic mixed-function oxidase activity in volunteers (reduced antipyrine and cortisol oxidation), and the authors infer matching inhibition of hepatic 25-hydroxylase and renal 1-alpha-hydroxylase [17]. In rat liver homogenates isoniazid at 1, 5 and 10 mmol/L left 81, 53 and 35 percent of 25-hydroxylase activity, and 120 mg/kg intraperitoneally lowered activity by 26 percent in vivo [21]. No retrieved source addresses CYP24 (24-hydroxylase).

Counter-evidence held on record

Counter-evidence and confounding The strongest data are a 14-day healthy-volunteer study of 8 people and a 9-child subgroup; neither was designed to show a clinically meaningful deficiency. Patients with TB start with low vitamin D and recover with food and sunlight, as the Tanzanian cohort shows, which can mask or reverse a drug effect [20]. The combination study is confounded by rifampicin, a known vitamin D-lowering drug [18]. Framing caution This is a finding about circulating metabolites in short or mixed-regimen studies. It is not evidence that an individual on isoniazid is vitamin D deficient and carries no supplement recommendation.

Limits on what this shows

• Null findings: No fall with rifampicin plus isoniazid in healthy volunteers [18]. No patient serum calcium measurement was retrieved.
• At-risk subgroups: Not defined by a retrieved source.

Direction

Depletion of circulating calcium and phosphate, secondary to the vitamin D finding above. Not a dietary deficiency.

Regulatory status

Calcium appears in the Chartwell label only as an inactive ingredient; no calcium or phosphate effect is described [2].

Population evidence

Brodie 1981 [17]: serum calcium and phosphate fell (P below 0.001), with PTH up 36 percent, in 8 healthy volunteers on isoniazid 300 mg daily for 14 days (values not in the abstract). Brodie 1982 [18]: with rifampicin added, calcium and phosphate did not fall. A 1982 paper on calcium metabolism during rifampicin and isoniazid therapy appeared in a PubMed listing but could not be opened.

Mechanism

Present, shared with the vitamin D entry: reduced hydroxylation of vitamin D lowering calcium absorption, with PTH rising in response [17].

Counter-evidence held on record

Counter-evidence and confounding One study of 8 volunteers over 14 days supplies the calcium signal, the magnitude is not in the abstract, and it is the same source as the vitamin D entry, so it adds no independent confirmation. Could rise to orange if the Brodie 1981 full text shows a meaningful fall or the unopened patient calcium paper confirms it. Framing caution Do not read as calcium deficiency or a bone-loss claim. Fracture and bone density are outcomes and are not graded. No calcium supplementation is implied.

Limits on what this shows

• Null findings: Folate rose with treatment in Line 1971 [27], and was already common before treatment [26][27], which is the Hard Rule 6 pattern (similar or worse before the drug).
• At-risk subgroups: Poorly nourished TB patients [26].

Direction

Possible depletion direction, but not attributable to isoniazid on the evidence retrieved.

Regulatory status

Folate is not mentioned in the Chartwell label; the Teva label sections returned did not mention it [2][1].

Population evidence

Klipstein 1967 [26]: two patients developed megaloblastic anemia on isoniazid and cycloserine. In 120 TB patients serum folate was subnormal in 7 of 12 inadequately nourished and 5 of 24 well-nourished untreated patients, in 15 of 29 on isoniazid plus cycloserine, and in 2 of 55 on other combinations. Line 1971 [27]: 33 TB patients, 45 percent had low serum folate and 55 percent megaloblastic erythropoiesis before treatment; after 3 months of chemotherapy serum folate rose significantly (regimen not named in the abstract). Evans 1971 on folate and treatment [28] could not be opened.

Mechanism

Absent for isoniazid. In vitro, isoniazid depressed growth of L. casei only at concentrations far above pharmacologic doses, while cycloserine did so at about three times its usual therapeutic concentration, and pyrazinamide not at all [26]. The authors conclude the deficiency is mainly dietary and the mechanism unknown.

Counter-evidence held on record

Counter-evidence and confounding Confounding by indication is presumed and is supported by the data: folate deficiency was frequent before treatment and improved on treatment. Cycloserine is a competing explanation for the one drug-linked signal [26]. Framing caution This is a context flag, not a folate-depletion claim for isoniazid. No folate or folic acid recommendation is implied.

Limits on what this shows

• Null findings: No change in B12 absorption was detected [27].
• At-risk subgroups: Not defined.

Direction

Depletion direction on a single cohort mean, not attributable to isoniazid.

Regulatory status

Not mentioned in the labels read [2].

Population evidence

Line 1971 [27]: 33 TB patients, mean B12 425 to 348 pg/mL after 3 months, no overall change in B12 absorption. The abstract does not name the regimen; isoniazid appears only via MeSH indexing in the OpenAlex record, so the isoniazid-in-cohort test is partial. Klipstein 1967 [26]: B12 was normal in the two anemic patients and in all 36 untreated patients, and indeterminate or subnormal in 17 of the patients on drug therapy.

Mechanism

Absent. No source retrieved links isoniazid to B12 handling.

Counter-evidence held on record

Counter-evidence and confounding The regimen in the 1971 cohort probably included other drugs; the abstract names none of them. A fall in a mean with unchanged absorption cannot be assigned to a drug. Framing caution Monitoring context only. It is not a B12-depletion claim for isoniazid.

Limits on what this shows

• At-risk subgroups: Not defined.

Direction

Food (or nutrient) affecting the drug, not the drug depleting a nutrient. Filed as a timing interaction per Hard Rule 4.

Regulatory status

Teva label: isoniazid should not be administered with food, and bioavailability is significantly reduced when taken with food; tyramine- and histamine-containing foods (cheese, red wine, some fish are named) should be avoided [1]. Drugs.com consumer information says certain foods may need to be avoided [45].

Population evidence

Peloquin 1999 [43]: randomized four-period crossover, 14 healthy volunteers, single isoniazid 300 mg doses. Fasting mean Cmax 5.53 (SD 2.92) microg/mL, Tmax 1.02 h, AUC 20.16 (SD 12.45) microg x h/mL. A high-fat meal cut Cmax by 51 percent (2.73 microg/mL), nearly doubled Tmax (1.93 h) and cut AUC by 12 percent (17.72 microg x h/mL). An aluminum-magnesium antacid did not significantly change Cmax, Tmax or AUC (Cmax 5.62 microg/mL, AUC 20.27 microg x h/mL). The authors advise isoniazid on an empty stomach whenever possible.

Separation interval

Neither the label sentences nor the Peloquin abstract give an hour figure, so no interval is asserted here. The prescriber or pharmacist and the product label set the schedule.

Mineral and vitamin co-administration

An aluminum-magnesium antacid, a magnesium-containing product, produced no significant change in isoniazid exposure [43]. No label statement linking a vitamin or mineral supplement to isoniazid absorption was found in the label text read [1][2].

Mechanism

Present: reduced and delayed absorption with a high-fat meal [43]. The detailed mechanism is not given in the retrieved text.

Counter-evidence held on record

Counter-evidence and confounding Single-dose crossover in healthy volunteers with combined drugs given alongside, so the meal effect on steady-state treatment in patients is not shown. Only a high-fat meal and one antacid were tested [43]. Framing caution A timing instruction about the drug, not a nutrient deficiency, and not a reason to skip meals or change diet without the prescriber.

Limits on what this shows

• Why no grade: Dual condition met: no Tier B or C source was found to support even a yellow grade, and no circulating claim of isoniazid magnesium depletion was traced to an opened source to rebut. This is partly a genuine gap and partly tooling-limited (two targeted PubMed queries were refused), not a finding of safety. Re-check with direct PubMed queries when access returns.

What was searched

The Chartwell label was read for magnesium and does not mention it [2]; the Teva label sections returned did not mention it [1]. PubMed listing "tuberculosis treatment serum zinc copper magnesium" returned 2 unrelated records (a 1982 neuroinfection trace-element paper and a Russian trace-element paper on a herbal co-treatment, titles only). PubMed queries "isoniazid magnesium" and "isoniazid hypomagnesemia" were refused with 429 and returned nothing. The one magnesium-relevant record opened is a pharmacokinetic null in the nutrient-to-drug direction: an aluminum-magnesium antacid did not change isoniazid exposure in 14 volunteers [43], which is a timing question, not a depletion one. PubMed listing "isoniazid serum magnesium" returned 5 records, none on magnesium status. Wikipedia, Drugs.com and MSKCC pages name no magnesium claim [44][45][46].

Channel outcomes

PubMed HTML listing: three queries returned irrelevant results and two were rate-limited (429). OpenAlex single-record lookups worked but require known PMIDs. The shared web-search budget was exhausted. PMC and Springer pages: reCAPTCHA or 429.

Limits on what this shows

• Why no grade: Dual condition met: no retrieved source supports a zinc effect of isoniazid even at yellow, and no circulating zinc claim was traced to an opened source. The two trace-element cohorts that could settle it were unreadable, so this is a tooling-limited gap as well as a literature gap, and not a finding of safety.

What was searched

The Chartwell label does not mention zinc [2]. PubMed listing "isoniazid zinc copper serum" returned 2 records: a Brazilian cohort of serum iron, copper and zinc during TB treatment (abstract null, not opened [42]) and a multidrug-resistant TB copper-zinc paper (isoniazid-resistant, not relied on). Wiid 2004 reports that total antioxidant status rose with therapy and correlated with zinc, but does not itemize the regimen and does not show a drug-caused fall [39]. Qi 2021 on trace elements across 6 months of therapy could not be opened [41].

Channel outcomes

OpenAlex abstract fields were null for both trace-element cohort papers, and the Springer page returned 429. PubMed listing queries returned few records. Web search was unavailable.