Three Years of Unreviewed Prophylaxis in Chronic Graft-Versus-Host Disease
Three years into chronic graft-versus-host disease, a trimodal prophylaxis regimen started without an end date meets its first real reassessment against the cumulative cost of staying on it.
Teresa V. is 61, and spent decades tending an elaborate backyard garden before chronic lung graft-versus-host disease made even moderate outdoor exertion leave her short of breath, and over the past year her daughter has gotten her into watercolor painting instead — something she can do sitting at her kitchen table on the days her energy allows. She is three years out from an allogeneic stem cell transplant for acute myeloid leukemia, complicated by chronic graft-versus-host disease affecting her skin and lungs, the latter manifesting as bronchiolitis obliterans that has never fully resolved. She remains on prednisone and ruxolitinib for GVHD control, and for the same three years has been maintained on trimethoprim-sulfamethoxazole, voriconazole, and acyclovir — prophylaxis against Pneumocystis, invasive mold infection, and herpesvirus reactivation respectively, none of which has ever been formally reassessed against a stopping question, because her GVHD, and the immunosuppression treating it, has simply never gone away long enough to raise one.
Her disease course has, in fact, been slowly improving — her prednisone dose is roughly half what it was eighteen months ago, and her most recent pulmonary function tests show her first genuine plateau rather than continued decline. Improving is not the same as resolved, and none of her three prophylactic agents was ever started with a fixed calendar duration attached the way post-transplant PCP or CMV prophylaxis typically is; each was tied instead to 'the duration of significant immunosuppression,' a standard that was easy to apply honestly at eighteen months and has become genuinely harder to apply at three years, both because her own trajectory has shifted and because the cumulative cost of three years on these specific drugs is no longer hypothetical.
Voriconazole in particular carries a dose- and duration-dependent association with phototoxicity and non-melanoma skin cancer that is documented in its own FDA labeling, which carries an explicit squamous-cell-carcinoma warning for long-term use and advises avoiding sunlight — a real concern for a fair-skinned patient three years into daily dosing, distinct from and in addition to its usual hepatotoxicity and QT-prolongation monitoring burden.
Prophylaxis that was never given an end date, three years in
My instinct is to keep all three agents unchanged. She is still on prednisone and ruxolitinib, which means she's still significantly immunosuppressed by any definition we'd use to justify starting this prophylaxis in the first place — the standard was always tied to ongoing immunosuppression, not to a calendar date, and nothing about her current regimen has actually stopped.
You're right that the original standard was tied to immunosuppression duration, not a fixed date — I'd apply that same standard forward, though, not just backward.
'Significant immunosuppression continues' described her accurately at eighteen months; at three years, with her prednisone dose halved and her pulmonary function finally plateaued rather than declining, I think the honest read of her current net immunosuppression is lower than it was, even though it hasn't reached zero. That doesn't mean stop everything — it means this is exactly the point a genuine reassessment was always supposed to happen, rather than defaulting to 'unchanged' because nothing has forced the question until now.
For the antifungal specifically, I'd frame this as a switch, not a stop-or-continue binary. Voriconazole's phototoxicity and squamous-cell-carcinoma association is on its own label, not inferred from case reports, and it is dose- and duration-dependent; she already has new actinic keratoses after three years of daily dosing in someone with fair skin — that risk doesn't reverse just because her GVHD is improving, and it keeps accumulating for as long as she stays on this specific drug.
Posaconazole covers largely the same mold-prophylaxis indication without that particular toxicity signature. If mold prophylaxis is still warranted at all, which I'd defer to infectious disease on, I don't think it needs to keep being voriconazole specifically.
Same logic applies to the TMP-SMX, for a different reason. She's already mildly cytopenic, most likely from the ruxolitinib itself, and TMP-SMX's own myelosuppression risk compounds that rather than acting independently of it. If PCP prophylaxis is still needed — and at her current immunosuppression level, I think it still is — atovaquone gets her the same coverage without adding to a cytopenia she's already managing.
Acyclovir I'd leave alone; it's the one agent in this regimen without a comparable toxicity argument against it, and herpesvirus reactivation risk doesn't track her improving GVHD trajectory the same protective way the other two questions do.
Agreed: acyclovir continues unchanged. Voriconazole switched to posaconazole given her accumulating phototoxicity and skin cancer risk, independent of whether mold prophylaxis itself continues. TMP-SMX switched to atovaquone given her existing cytopenias. Nothing was stopped outright today — every change was a substitution addressing a specific accumulated cost, not a declaration that her immunosuppression has resolved enough to need no prophylaxis at all.
Genuinely unresolved: when, if ever, this gets revisited as a stopping question rather than a switching one. The BMT physician's read is that as long as she's on any dose of ruxolitinib, the honest answer to 'is she still significantly immunosuppressed' stays yes; the ID physician's read is that a meaningfully lower dose, sustained over time, should eventually earn its own reassessment separate from whether the drug has been fully discontinued. Neither position was adopted as the team's official stance going forward — the switches were agreed, the underlying disagreement about when 'improving' becomes 'enough' was not.