ART Timing in TB/HIV Coinfection: Racing the CD4 Count
A CD4 count of 38 and a five-day-old TB diagnosis put the team on a clock most trial evidence has already read for this exact number -- the disagreement is about how much lead time his lungs deserve first.
R.M., a 34-year-old man, has worked the overnight shift stocking a regional grocery distribution warehouse for six years, a schedule he chose specifically so he could pick up his daughter from preschool most afternoons before his wife left for her own shift. He'd written off three months of night sweats and a cough that wouldn't clear as the warehouse's poor ventilation and a hard winter, until he dropped nearly twenty pounds without trying and his supervisor sent him to urgent care. A chest X-ray there showed a cavitary right-upper-lobe infiltrate; sputum came back AFB-smear-positive two days later, and an HIV test drawn as part of the same workup came back positive -- a diagnosis he had no reason to suspect and no symptoms he'd have attributed to it. He has no other chronic illness and, by his own account, hadn't seen a doctor in nearly a decade before this week.
His CD4 count returned at 38 cells/uL, with an HIV RNA of 610,000 copies/mL -- profound immunosuppression, not a borderline reading. He started standard four-drug rifampin-based TB therapy five days ago and has tolerated it without difficulty; his night sweats have already begun to ease. The team's actual question is when to layer antiretroviral therapy on top of it. SAPIT, the trial that first settled this question for very low CD4 counts, randomized patients to integrated versus sequential ART and stopped early on a mortality benefit; CAMELIA tested 2 weeks against 8 weeks specifically in patients with CD4 under 200 and found the earlier arm reduced death despite more IRIS. STRIDE is the trial that actually drew the line at 50 cells/uL -- its own subgroup analysis found the survival benefit of early ART concentrated entirely below that threshold, with no mortality advantage and a real IRIS cost above it. At 38 cells/uL, R.M. sits inside the exact population all three trials agree should not wait. His sputum smear was graded 3+, a heavy bacillary burden by the scale his team uses to track treatment response, consistent with the cavitary finding on imaging rather than a milder, paucibacillary presentation -- a detail that makes the pulmonologist's IRIS concern a genuine, case-specific consideration rather than a generic caution layered on top of the mortality argument the trials establish.
On the TB ward, day five of treatment
A CD4 of 38 places him exactly in the population these trials were built to answer. SAPIT randomized integrated versus sequential ART in patients with TB and stopped early on a mortality benefit for starting during TB treatment rather than after. CAMELIA went further and directly compared 2 weeks against 8 weeks in CD4-under-200 patients, and the 2-week arm survived better despite a higher IRIS rate. STRIDE is the one that actually drew the boundary at 50 cells/µL — its own subgroup analysis found the mortality benefit essentially entirely below that line, with no benefit and more IRIS above it. He isn't close to that line. Start ART within the first two weeks.
You're right that the mortality data point one direction here — I'm not arguing the trials disagree with you.
But his imaging is not incidental to this decision. A cavitary, high-burden infiltrate means a large mycobacterial antigen load, and that's precisely the substrate TB-IRIS feeds on — in his lungs, not somewhere I'd rather it happen. Even CAMELIA's own early arm wasn't day one; its median start was closer to day 14. I'd let five more days of rifampin work reduce his bacillary burden before we add a second drug regimen that's going to reconstitute his immune response against exactly the organism sitting in that cavity. He's also down nineteen kilos from his stated baseline — that's not incidental either. Real disease burden this extensive doesn't resolve as fast as a five-day fever trend suggests, and I don't want the early symptomatic improvement to make us treat the underlying antigen load as smaller than it actually is.
Whichever day the two of you land on, the harder problem is what he's actually started on. Rifampin is a potent inducer of UGT1A1 and CYP3A4, and standard-dose dolutegravir's levels fall substantially under that induction — the current DHHS approach is dolutegravir 50mg twice daily rather than once, specifically to compensate. If anyone reaches for a once-daily fixed-dose combination out of habit, they'll underdose him during the exact window everyone in this room agrees is the highest-stakes part of his treatment.
Agreed: ART starts on day twelve of TB treatment, splitting the difference the pulmonologist asked for without drifting toward the delayed timeline the trial data argues against — dolutegravir 50mg twice daily with tenofovir/lamivudine, continued alongside the unchanged four-drug TB regimen. Close inpatient monitoring for IRIS is planned through the first month regardless of which day ART began, since none of the three trials eliminate that risk, only shift its odds.