Pneumocystis Pneumonia With Every Standard Alternative Ruled Out Before It's Even Considered
A single patient, whose two standard fallback regimens both failed for reasons that turned out to share the same underlying mechanism. The disagreement is which of the two options left standing is actually the safer bet.
N.A., a 61-year-old woman, is six months into maintenance rituximab for follicular lymphoma, a regimen that has kept her disease controlled but has also quietly stripped away most of her B-cell defenses one infusion at a time. She retired from a long career as a school librarian last year, planning to spend her newfound time gardening, and has instead spent much of the past six months managing the fatigue that comes with ongoing chemoimmunotherapy. Ten days ago a dry cough started, mild at first, then steadily worsening into real dyspnea on exertion that brought her in today with a resting oxygen saturation of 91% on room air — low enough to be unmistakably abnormal, but not the sub-90 hypoxemia that would place her past the severity line without further calculation, which is why the A-a gradient was worked out rather than assumed from the number on the pulse oximeter. CT showed diffuse ground-glass opacities, and bronchoalveolar lavage PCR confirmed Pneumocystis jirovecii pneumonia — an infection her rituximab-depleted B-cell compartment left her genuinely vulnerable to, despite never carrying an HIV diagnosis.
What should be a straightforward treatment decision collapsed fast once her history was reviewed in full. TMP-SMX, first-line for PCP treatment at any severity, is not an option — she carries a documented history of a Stevens-Johnson-like reaction to sulfa in her twenties, severe enough that no amount of severity justifies rechallenge. The next standard fallback for treatment-dose PCP, dapsone combined with trimethoprim, was ruled out for an entirely different reason discovered on pretreatment labs: incidental G6PD deficiency, which makes dapsone a real hemolysis risk. That same enzyme deficiency turns out to rule out primaquine-clindamycin as well, since primaquine carries its own well-described G6PD-hemolysis risk — a connection easy to miss if each drug is evaluated against her allergy history alone rather than against the shared oxidative mechanism underneath both. What remains standing are two options with genuinely different evidence behind them. Atovaquone's place rests on Hughes's 1993 comparison against trimethoprim-sulfamethoxazole and, more directly relevant here, Dohn's comparison of oral atovaquone against intravenous pentamidine — which found the two equivalent in mild-to-moderate disease, with less treatment-limiting toxicity from atovaquone. Both trials enrolled patients with AIDS, and her calculated gradient places her inside their severity band but her immunosuppression outside their population: non-HIV Pneumocystis carries higher mortality on a smaller organism burden, driven by the inflammatory response rather than the fungal load those trials were measuring. So the evidence that makes atovaquone reasonable for her is evidence about a milder illness in a different kind of host, and the question is not which drug her numbers select but how much weight her numbers can carry.
In the pulmonary consult room, watching two fallback options both close
Her marrow reserve is already fragile from ongoing rituximab, and treatment failure in a host this immunosuppressed is the outcome I'm most worried about. I'd rather accept IV pentamidine's toxicity than gamble on an option with a thinner evidence base for a disease at her level of vulnerability.
Her hypoxia is real, but by the actual A-a gradient criteria, not just clinical impression, she classifies as moderate rather than severe disease — and atovaquone's evidence base for moderate PCP, while thinner than TMP-SMX's, is real, not absent. I'd rather not introduce a second, harder-to-monitor toxicity profile in a patient this vulnerable when a less toxic option is actually supported by her own numbers.
"Thinner evidence" isn't the same as "no evidence" — the atovaquone trials specifically targeted this severity band, which is exactly where her calculated gradient places her.
Before we finalize either option, it's worth naming why two of the standard fallbacks are already gone, because it's not the same reason twice. Dapsone-trimethoprim is out for G6PD-mediated hemolysis risk — but primaquine-clindamycin, which some teams might reach for next as "the other alternative," carries that identical G6PD-hemolysis mechanism through primaquine specifically, and is equally unusable here. That leaves atovaquone and pentamidine as the only two real options left standing, which is exactly why getting her severity classification right matters as much as it does.
Agreed: atovaquone started given her severity classification as moderate by calculated A-a gradient, with daily clinical reassessment and an explicit, named threshold to escalate to IV pentamidine if she hasn't improved within 5-7 days.
Not agreed: whether that 5-7 day window is too permissive given how little room she has for a second failed attempt before crossing definitively into severe disease. The hematologist's preference for pentamidine from the outset is documented as overruled by the severity classification today, not resolved as a shared view — she asked explicitly to be part of the reassessment conversation at day 3 rather than waiting for the full window to close.