Acute PCP: The Same Question, the Opposite Answer
The same ART-timing question that argues for waiting in cryptococcal meningitis argues for moving quickly here — a direct, deliberate contrast built on the same underlying trial logic landing in opposite places.
T.J., a 29-year-old man, spends most weekends coaching a youth flag-football league he helped found three years ago, something he'd been reluctant to miss even as a dry cough crept in over three weeks and slowly turned into real shortness of breath climbing the bleachers. He came to the emergency department after a night of fever and an oxygen saturation his roommate measured on a borrowed pulse oximeter at 88% — low enough that his roommate drove him in rather than waiting for morning. A chest CT showed diffuse ground-glass opacities without focal consolidation, and an induced sputum sample confirmed Pneumocystis jirovecii by PCR. His HIV test, sent as part of the same workup, was his first ever and came back positive. He reports no other medical history and no medications before this admission.
His CD4 count is 68 cells/µL, low enough to have caused the pneumonia but well above the profound suppression seen with cryptococcal disease, and his LDH — a nonspecific but useful marker of the burden of alveolar injury in PCP — is elevated at 410 U/L, consistent with the moderate-to-severe disease his oxygen saturation already suggested. He started trimethoprim-sulfamethoxazole and adjunctive prednisone yesterday, given his room-air saturation under 90%, and his breathing has already begun to ease. The team's question mirrors exactly the one asked for cryptococcal meningitis — when to add ART — but the evidence here answers it the opposite way. ACTG A5164 randomized early ART, started within 14 days of beginning OI treatment, against deferred ART in a population that was mostly PCP, and the early arm had fewer deaths and AIDS progressions with no significant increase in adverse events. Where COAT found early ART kills patients with cryptococcal meningitis, A5164 found it protects patients with PCP — the same clock, read in opposite directions by the organism it's timed against, and a deliberate pairing worth holding next to each other rather than treating as two unrelated timing rules to memorize separately.
Medicine ward, day two of PCP treatment
This is almost the mirror image of the cryptococcal meningitis conversation down the hall, and it resolves the opposite way. A5164 randomized ART within 14 days of starting OI treatment against deferred ART, in a population that was predominantly PCP, and found fewer deaths and AIDS-defining events in the early arm with no meaningful safety cost. He's also already showing us his trajectory is favorable — 88% on room air to 94% on 2 liters in under a day — which is the kind of respiratory improvement that makes a moderate IRIS reaction, if it happens at all, something he has reserve to tolerate. That's a different starting position than the cavitary TB case, where the concern was IRIS landing on lungs that hadn't shown any improvement yet. Start ART within the next two weeks.
My instinct was to finish his three-week PCP course first — that's what I was trained to do, and it's still what I see written on this floor.
But that instinct predates the trial you're citing, and A5164 tested exactly this diagnosis and found the opposite of what the 'finish the OI course first' reflex assumes. I'll defer to the data over the habit — I don't have a case against starting early here, I just wanted to name where my hesitation was actually coming from before dropping it. One thing I will flag: he's on the prednisone taper for his oxygen saturation, and I want it on record that if his breathing worsens once ART starts, we're deciding in real time whether that's PCP-IRIS or the taper itself stepping down too fast, not assuming it's automatically one or the other.
Agreed on the direction, one calibration point: A5164's early arm had a median start around day 12, not day 0 or day 2. If the instinct now swings to 'start ART today, PCP treatment barely begun,' that's a step past what the trial actually tested, not a straightforward reading of it. I'd target roughly day 10 to 12 — early by the trial's own definition, not earlier than its own definition.
Agreed cleanly, without a lasting disagreement to carry forward: ART targeted to start around day 10-12 of PCP treatment, matching A5164's own early-arm timing rather than either the older deferred-until-course-completion teaching or an unstudied same-day start. His respiratory status will be reassessed before ART begins to confirm he's stable enough to tolerate any early IRIS-related inflammatory response.