Rechallenging a Drug That Already Injured His Liver, Because Nothing Else Treats His Disease
A patient whose liver fully recovered from isoniazid-induced injury needs isoniazid again, and no adequate alternative exists for his specific tuberculosis exposure. The debate isn't whether rechallenge is risky — it clearly is — it's whether the alternative is actually safer.
Desmond A., a 39-year-old man, immigrated eight years ago and now works as a home health aide, a job he says he's proud of because it lets him care for people the way someone once cared for his own grandmother. Eighteen months ago he was started on isoniazid for latent tuberculosis infection identified during a routine health screening, and developed drug-induced liver injury six weeks in — ALT rising to eight times the upper limit of normal, with resolution to baseline within two months of stopping the drug and no other identifiable cause found on workup at the time. His liver has been entirely normal on every check since.
He has now been identified as a close contact of a household member recently diagnosed with active, isoniazid-susceptible tuberculosis, placing him at meaningfully elevated risk of progression from latent to active disease if untreated. The instinct in a household-contact situation is to reach for isoniazid again, since that is the drug his contact's organism is known to be susceptible to. What that instinct skips is that susceptibility to isoniazid does not imply resistance to anything else. Menzies's nine-country randomized trial found four months of daily rifampin non-inferior to nine months of isoniazid for preventing active tuberculosis, with a higher completion rate and, specifically, fewer grade 3-to-5 hepatotoxic events — a rate difference of 1.2 percentage points in rifampin's favor. In a patient whose ALT has already gone to eight times normal on isoniazid, that is not a marginal preference. The genuine question left is therefore narrower than a rechallenge: whether a liver that has demonstrated susceptibility to one antitubercular drug can be assumed safe on another, and how closely to watch it while finding out.
Whether the alternative is actually the safer choice
I want to treat him, and quickly — untreated progression from latent to active tuberculosis in a household contact of an active case is a serious risk, and it's the thing driving my urgency here. But I'm not going to argue for rechallenging isoniazid to get there. Menzies settled the question I would once have raised: four months of rifampin is non-inferior for preventing active disease, and his contact's isolate being isoniazid-susceptible tells us nothing about rifampin. I'd start 4R. What I'd flag is that 3HP is off the table for him — it contains isoniazid — so the short rifapentine option most of my colleagues reach for first isn't available.
We agree on the drug, so let me push on the part we haven't settled. A documented DILI episode tells us something about him, not only about isoniazid: a patient who has mounted an eight-fold transaminase rise to one antitubercular agent is not the same baseline risk as a treatment-naive patient starting rifampin, even though rifampin's population-level hepatotoxicity is lower. Rifampin's own injury pattern is usually cholestatic and usually early, which is a different thing to watch for than the hepatocellular picture he actually had.
Treating the trial's population-level safety margin as though it transfers intact to him is the move I'd resist — Menzies enrolled patients without a prior drug-induced liver injury, so the very thing that makes him interesting is the thing his evidence base excluded.
Neither of you has raised what I think is the practical hazard, which isn't hepatic at all. Rifampin is a potent inducer, and he's a home health aide — if anything gets added to his regimen over these four months, from a statin to an antiretroviral to hormonal contraception in a partner's care plan he's managing, the induction is real and it is not intuitive to non-specialists. That needs to be documented somewhere he and his other prescribers will actually see it. On the liver: I'd still want a specific, pre-committed stopping rule with an exact ALT threshold and a fixed interval, decided today. His first episode ran six weeks before injury became apparent, and a vague intention to 'watch closely' is exactly the plan that drifts past where it should have stopped.
Four months of daily rifampin started, with a pre-committed stopping threshold of ALT at 3 times the upper limit of normal (rather than his prior episode's 8-fold rise) and biweekly LFT checks for the first two months. Isoniazid rechallenge was considered and rejected; 3HP was excluded because it contains isoniazid.
Not fully agreed: whether Menzies's safety margin can be assumed to hold in a patient his trial would have excluded. The hepatologist's concern about individual susceptibility was documented explicitly as an active, unresolved risk being accepted deliberately, not a disagreement talked past — the plan proceeds with that caution on record rather than as settled consensus.