Transplant, CMV & Herpesviruses

Educational study digest pitched at a new ID fellow. Clinical recommendations trace to the source teaching cases and the cited guidelines and trials; the enrichment box is reviewer synthesis of additional evidence, each claim cited. Not medical advice, and not a substitute for the primary guidelines. Invasive moulds in the immunocompromised host live in the companion Mycology digest; they are cross-referenced here, not covered.

◆What does the post-transplant calendar predict?

The single most useful orienting tool after transplant is the timeline: time from transplant narrows the differential faster than any one symptom, because it tracks the net state of immunosuppression. Roughly three windows are worth memorising. In the first month the problems are those of any post-surgical inpatient — nosocomial and hospital-derived infections, surgical-site and technical/line complications, and organisms carried in by the recipient’s own colonisation or the donor graft; the classic opportunists are still uncommon this early because cumulative immunosuppression has not yet done its work. The one-to-six-month window is the opportunistic window, and what you see depends on prophylaxis: on prophylaxis the field shifts toward C. difficile, viral reactivations, and the occasional cryptococcus, whereas without it the textbook reactivation opportunists surface — PCP, CMV and the other herpesviruses, reactivated (previously suppressed) hepatitis, and Listeria. Beyond six months most patients look like the general community — community-acquired infections — with a tail of late or atypical bacterial and fungal disease, late viral infections, and post-transplant lymphoproliferative disorder (PTLD). When a focal problem such as a pulmonary nodule appears around the six-month mark, work the source deliberately: donor-derived, recipient reactivation, or de novo acquisition.

◆How does CMV differ between the organ-transplant host and the HIV host?

CMV behaves like two different diseases depending on the immune defect, and the discriminator is which pattern predominates. In the solid-organ transplant population, CMV most often presents as CMV syndrome — fever, malaise, and bone-marrow suppression (leukopenia, thrombocytopenia) — and when it becomes tissue-invasive the gastrointestinal tract is the most common site, while pneumonitis and CNS disease are the most severe; retinitis is rare in this group. The dominant risk variable here is donor/recipient serostatus: a seropositive donor into a seronegative recipient (D+/R−) is the highest-risk mismatch, with up to 70% developing infection, and the highest-risk transplanted organs are lung, pancreas, and intestine — renal transplants sit at lower risk. In the HIV host the picture flips: end-organ CMV is predominantly retinitis (then colitis and esophagitis, with pneumonitis described), and it clusters at CD4 <50 — the lower the count, the greater the tendency to cross the blood–brain barrier into ocular and neurologic tissue.

◆The CMV viral load won’t fall — when is it resistance, and to what?

Picture an allogeneic HCT recipient on tacrolimus and mycophenolate who completed a full therapeutic course of valganciclovir, saw an initial decline, and now has persistent low-level viremia with a rising load. That trajectory after a completed full-dose course points to resistance, and genotypic testing settles it. The most frequent culprit is a mutation in the UL97 kinase gene (classically M460, A594, L595, or C603), which confers resistance to ganciclovir and valganciclovir by preventing the phosphorylation that activates the drug. The board-defining point is that UL97 mutations usually retain susceptibility to foscarnet and cidofovir, because those agents do not depend on UL97 for activation — so the next step is to switch to foscarnet (cidofovir is the other UL97-independent option). Simply increasing the dose is only reasonable for low-level (<5-fold) UL97 resistance, not for a rising load after full therapy. The mutation you do not want to see is in UL54, the viral DNA polymerase: UL54 mutations cause cross-resistance across ganciclovir, foscarnet, and cidofovir, closing off the usual escape routes.

◆Letermovir: a prophylaxis drug with three catches

Letermovir is a CMV terminase-complex inhibitor (it acts on the pUL56 terminase, a mechanism distinct from the polymerase targeted by ganciclovir and foscarnet), used for CMV prophylaxis in seropositive transplant recipients. Three properties define how a fellow uses it and where it fails. First, it has no cross-coverage of the other herpesviruses — it does nothing for HSV or VZV, so those agents (e.g., acyclovir) must continue alongside it. Second, it carries meaningful drug–drug interactions through CYP3A, which matters in a patient already juggling calcineurin inhibitors and azoles. Third, it has a low barrier to resistance, which is precisely why it is positioned as a prevention tool rather than as treatment for established, actively replicating disease. Hold those three catches together and letermovir’s niche — and its failure modes — become predictable. The prophylaxis evidence is expanded in the enrichment box below.

◆Reading the infected retina — and why CMV blood tests disappoint

Vision loss in the immunosuppressed patient is a pattern-recognition exam, and several retinal findings are close to pathognomonic. “Frosted branch angiitis” — thick white sheathing of the retinal vessels — signals a vasculitis and is associated with HIV and HSV. “Pizza-pie” retinopathy — confluent yellow-white retinal necrosis (the cheese) mixed with dense hemorrhage (the red sauce) — is the signature of CMV, and it characteristically shows less vitritis than HSV or VZV. “Headlight in the fog” — a focal white/yellow necrotising retinochoroiditis glimpsed through an overlying haze of vitreous inflammation — is toxoplasmosis. “Snow-banking” — snow-white fluffy exudates layered in the vitreous — points instead to intermediate uveitis (pars planitis). The reason these ophthalmologic patterns carry so much weight is that CMV blood diagnostics are unhelpful: serology mainly helps when the IgG is negative (making disease unlikely), viral blood culture is insensitive and slow, and blood PCR is neither sensitive nor specific for end-organ disease (in AIDS it merely tracks the CD4). Cytology is suggestive but not specific. CMV retinitis is therefore an eye diagnosis, treated with valganciclovir, with a response expected in 7–14 days; failure signals resistance (more likely with ongoing immunosuppression and viral replication), and salvage options are ganciclovir, foscarnet, cidofovir, and intraocular (intravitreal) injections.

◆Start voriconazole, and the tacrolimus climbs — which dose moves?

When a stem-cell recipient on tacrolimus is started on voriconazole for a mould, the predictable event — if nothing is adjusted — is supratherapeutic tacrolimus with azotemia, headaches, or seizures. The mechanism is clean: voriconazole is a potent CYP3A4 inhibitor, and tacrolimus (a calcineurin inhibitor) is a CYP3A4 substrate, so blocking the enzyme raises the CNI level. The move is to reduce the calcineurin-inhibitor dose — roughly a 60% cut in tacrolimus — when voriconazole starts, then follow serum levels closely, since absorption and other interactions can shift the kinetics. Note the direction carefully: the azole does not lower tacrolimus, and tacrolimus does not lower voriconazole. It is worth owning the tacrolimus toxicity profile the exam leans on — renal (azotemia) and neurologic (headache, tremor, seizures, mutism) — and the fact that acyclovir and valacyclovir are not hepatically metabolised, so they are safe bystanders here with no comparable interaction. The principle generalises to the whole azole class: azoles raise calcineurin-inhibitor levels, so reduce the CNI, don’t chase the azole.

◆When is zoster “disseminated,” and how do you isolate it?

Two things about VZV get tested at the bedside: what counts as dissemination, and how to isolate. Disseminated VZV is defined as involvement of >2 contiguous dermatomes, >20 vesicles outside the initial dermatome, or systemic (visceral) involvement. The infection-control consequence is the high-yield point: for suspected or confirmed disseminated disease, use Airborne and Contact precautions — the same combination applied to primary varicella — rather than the contact-only approach that suffices for localised zoster in an immunocompetent host. In the transplant or haematology patient the threshold to escalate is low, because these are exactly the hosts in whom localised zoster is most likely to disseminate.

◆Tender plaques and fever in the recovering AML patient — infection, or not?

A neutrophilic-dermatosis mimic rounds out the immunocompromised-host pattern set. In a patient recovering from AML induction — febrile, recently neutropenic, on prophylaxis and G-CSF — the appearance of multiple painful, tender red papules and plaques on the arms, dorsal hands, and trunk is classic for Sweet’s syndrome, an acute neutrophilic dermatosis tied to AML, MDS, and other malignancies. The tell in timing is that the lesions erupt as the neutrophils return (often during G-CSF), and the skin biopsy shows a dense neutrophilic infiltrate without organisms — had this been fungal, the fungal stains would show the organism and culture would turn positive (Candida, or occasionally moulds such as Fusarium, at 48 hours). The treatment is not another antimicrobial but a short, tapering course of prednisone, which produces rapid improvement (relapse can occur). Recognising it spares the patient an escalation of empiric antifungals for what is an inflammatory, steroid-responsive process.

Sources: ID board-review and case-conference teaching (source-of-record for the body). Ganciclovir-resistant CMV: Saullo & Wolfe, Annu Rev Med 2023;74:89–105. Drug interactions with immunosuppression: Sparkes & Manson, Clin Transplant 2019; doi:10.1111/ctr.13510. Enrichment — ECIL Cytomegalovirus (CMV) in Hematology Patients (2024), local guideline archive; letermovir prophylaxis pivotal trial: Marty FM et al., N Engl J Med 2017;377(25):2433–2444; doi:10.1056/NEJMoa1706640; PMID 29211658 (retrieved via PubMed).