The Swelling Is Real, and So Is What Happens If ART Stops
Everyone in the room agrees the inflammation is real and worsening. What's contested is whether treating it around a continued, working antiretroviral regimen is still the safer path, or whether his imaging has already crossed into territory that argues for pulling ART back.
A.M., a 38-year-old man, works installing home security systems, a job that has him climbing ladders and working in attics most days — physical work he was still doing, if more slowly, right up until this admission. He was diagnosed with disseminated tuberculosis and profound HIV-associated immunosuppression eight weeks ago, with CNS involvement confirmed on his initial imaging as several small parenchymal tuberculomas. He started TB therapy immediately and began ART three weeks later, a figure his own team arrived at. No guideline supplies it: guidance for central nervous system tuberculosis advises deferring ART well past the two-week window that mortality data support for pulmonary disease, precisely because this site's IRIS risk is higher. Three weeks split the difference between that caution and a CD4 count nobody wanted to leave untreated any longer. For the first ten days after starting ART he seemed to be improving — more alert, eating again, walking the hallway with assistance. His wife, who has been at the bedside most days, says that stretch was the first time in two months he'd seemed like himself.
Over the past four days that trajectory has reversed. He has developed a new, severe headache, intermittent vomiting, and today's repeat MRI shows his tuberculomas have enlarged, with new surrounding edema and early effacement of the fourth ventricle — findings consistent with a textbook paradoxical presentation of CNS TB-IRIS, but a genuinely concerning trajectory on imaging rather than a mild one. His HIV RNA remains undetectable and his CD4 count has climbed to 210 cells/µL, real evidence the ART is doing exactly what it's supposed to do immunologically — the same reconstitution now also appears to be driving the inflammatory process narrowing his fourth ventricle. Meintjes' randomized trial of adjunctive corticosteroids specifically for paradoxical TB-IRIS found real benefit without excess harm, but that trial's population did not include patients with his degree of new mass effect and early pressure signs on imaging — a genuine gap between what the strongest available evidence supports and what his own scan now shows.
Neuro-ICU, week three of TB treatment and ART
His CD4 recovery to 210 and undetectable viral load are exactly the immune reconstitution ART is supposed to produce, and that same reconstitution is what's driving this. Meintjes' randomized trial of adjunctive prednisone specifically for paradoxical TB-IRIS found real symptomatic benefit without a signal of harm. Worth naming directly: those first ten good days weren't a false start, they were the reconstitution actually working — which is exactly why I don't want to read the reversal as ART failing him. It's the same process now running somewhere less forgiving. I'd continue ART unchanged and start high-dose corticosteroids to treat the inflammation directly, rather than give up the ground his immune system has already regained.
I want to be direct that I'm not confidently arguing for interruption — I'm genuinely torn, which is different from disagreeing with you outright.
But Meintjes' trial population doesn't include what I'm looking at on today's MRI — new mass effect and early effacement of the fourth ventricle, not just inflammatory markers or mild lesion growth. CNS-involving IRIS is the one form of this syndrome where the real failure mode, herniation, is severe enough that some expert guidance still names ART interruption as appropriate for life-threatening presentations specifically. I think we're close enough to that line that steroids alone might not be a safe bet to make without a backup plan already decided.
I don't think this has to be decided as a single binary choice today. Start high-dose corticosteroids now, with ART continued, but with concrete, prespecified criteria: frequent neurologic checks, and if there's no meaningful improvement in his headache, vomiting, or repeat imaging within 48 to 72 hours, ART interruption becomes the next step immediately, not a fresh debate. That gives the infectious disease physician's approach the chance to work while giving the neurologist's concern a real, time-boxed trigger rather than leaving it as an open worry.
Agreed on the immediate plan — high-dose corticosteroids started now, ART continued unchanged — with an explicit, written 48-72 hour checkpoint rather than an open-ended wait-and-see. Not resolved, and deliberately left as a named point of genuine disagreement: whether that checkpoint should trigger automatic ART interruption if imaging hasn't improved, or whether a second steroid escalation should be tried first — the neurologist and infectious disease physician did not converge on this, and agreed to revisit it directly against the actual 48-hour imaging rather than settle it hypothetically today.