Plasma Exchange for Diffuse Alveolar Hemorrhage in ANCA-Associated Vasculitis, Pending Dual Serology
Samuel's lungs are hemorrhaging faster than his team can be certain which antibody is driving it. Plasma exchange's value may hinge on a second serology result still a day away.
Samuel O., a 61-year-old man, retired two years ago from thirty years running a small print shop. Until six weeks ago he had no chronic illness requiring regular medication, but he had been feeling generally unwell since then — fatigue, joint aches, a persistent low fever he attributed to a cold that wouldn't quite clear — before he coughed up blood twice in one afternoon and went straight to the emergency department instead of waiting for his already-scheduled primary care visit. He arrived hypoxemic, requiring 6 liters of supplemental oxygen to keep his saturation above 92%, with bilateral ground-glass opacities on chest CT and a hemoglobin that had dropped from 13.8 g/dL two months ago to 8.9 g/dL today — the pattern of diffuse alveolar hemorrhage, not pneumonia. His creatinine, at 2.1 mg/dL, is also new; urinalysis shows red cell casts. Serology drawn on arrival returned PR3-ANCA strongly positive, consistent with granulomatosis with polyangiitis affecting both his lungs and his kidneys at once. A second serology — anti-glomerular basement membrane antibody, sent at the same time because dual-positive disease, while uncommon, behaves more aggressively than either antibody alone — is still pending, expected sometime tomorrow.
Pulse-dose corticosteroids and rituximab are already running; nobody on the team disputes that starting point. The disagreement is whether to add plasma exchange tonight or wait for the anti-GBM result. PEXIVAS (Walsh and colleagues, 2020) found no significant reduction in death or kidney failure when plasma exchange was added to standard immunosuppression, alongside a real increase in serious infection, and it moved much of the field away from routine use. It is often called a renal study that happened to include some hemorrhage patients; that is not quite right. Alveolar hemorrhage was a co-equal entry route — eligibility was an eGFR below 50 and/or diffuse alveolar hemorrhage — and 191 of its 704 participants came in through it. The finer point is the true one: only 61 had hemorrhage PEXIVAS classed as severe, meaning a room-air saturation of 85% or less, or mechanical ventilation. The dedicated analysis of that group (Fussner and colleagues, 2024) found death within a year in 8.4% with plasma exchange against 15.6% without it — a hazard ratio of 0.52 whose confidence interval, from 0.21 to 1.24, comfortably includes both a halving of mortality and no effect at all. Samuel needs six liters to hold a saturation above 92%. Nobody has recorded what he does on room air, which is the number that would place him inside or outside the only subgroup where the question is still genuinely open.
ICU consultation, day one of induction therapy
PEXIVAS found no significant reduction in death or kidney failure when plasma exchange was added to standard immunosuppression, and a real increase in serious infection. He's already on pulse steroids and rituximab. I'd add something that cuts against my own convenience here: 85% of PEXIVAS's patients were induced with cyclophosphamide and only 15% with rituximab, so his regimen is the minority arm of that trial's experience either way. But the direction of the infection signal doesn't depend on which agent you pick, and adding apheresis to a man we are about to deplete of B cells is a real cost for a benefit the largest trial in this disease could not demonstrate.
I'm not disputing the headline, and I'll correct something we both say loosely — hemorrhage wasn't an afterthought in PEXIVAS, it was an entry route, and 191 patients came in through it. The subgroup that matters is the severe one, and there were only 61 of those.
Which is exactly my point. Fussner's analysis of that group puts one-year death at 8.4% with exchange against 15.6% without, hazard ratio 0.52 — and a confidence interval running from 0.21 to 1.24. You are reading that interval as containing no effect. I'm reading it as containing a halving of mortality. Neither of us is misreading it; it's 61 patients and it cannot adjudicate between us. His hemoglobin fell almost five grams in two months and he needs six liters of oxygen. When the evidence genuinely cannot decide, I'd rather act on the mechanism — and circulating antibody is what is driving the bleeding in his lungs tonight.
You're both arguing about which side of a confidence interval to stand on when there are two cheaper facts we don't have yet. The first is the anti-GBM serology, back tomorrow — if he's dual-positive, plasma exchange becomes independently indicated by anti-GBM disease's own paradigm, which doesn't share PEXIVAS's null result at all, and this argument evaporates. The second is smaller and we could have it within the hour: nobody has documented his saturation on room air. PEXIVAS drew its severe line at 85% or mechanical ventilation. Six liters to hold 92% tells us he's hypoxemic; it doesn't tell us whether he's in the 61 patients Fussner analyzed or the 130 who weren't. His oxygenation is supported, his immunosuppression is running, and a day's delay to learn which disease and which subgroup we're treating is cheap against committing to an intervention with a real infection cost.
Agreed: continue pulse steroids and rituximab, continue oxygen support, document a room-air saturation, and defer the plasma exchange decision pending anti-GBM serology due within 24 hours.
Not agreed, and left genuinely open rather than deferred: what the team should do if the anti-GBM returns negative and his room-air saturation lands just above 85%, placing him outside the only subgroup anyone has analyzed. The critical care physician would treat him as though he were inside it, on the grounds that a threshold drawn for trial enrollment is not a biological boundary. The rheumatologist would treat PEXIVAS's overall result as governing wherever a patient falls outside the severe group. Neither could say what evidence would settle it, which both of them noticed.