Largest Global Study of Rare Fungal Infection Finds Drug Choice May Matter Less Than Immune Recovery

Chronic disseminated candidiasis, a rare and feared fungal complication of cancer treatment, has long been managed with months of antifungal therapy and repeated imaging, largely on the basis of small, single-centre studies. Now the largest international cohort ever assembled has delivered a finding that could reshape how clinicians approach the disease: the class of antifungal drug chosen first appears to make little difference to survival or treatment response, while recovery of the patient’s own immune system emerges as the dominant force determining outcome.

The study, published in eClinicalMedicine, brought together 111 adults with chronic disseminated candidiasis from 19 countries, diagnosed between January 2007 and December 2025. The condition, also known as hepatosplenic candidiasis, typically strikes patients with blood cancers such as acute myeloid leukaemia as their white blood cell counts recover after prolonged chemotherapy-induced neutropenia. It produces persistent fever, abdominal pain, cholestasis and characteristic abscesses in the liver and spleen, and it can force delays in cancer treatment that may ultimately matter more than the infection itself.

Almost all patients in the cohort had an underlying haematological malignancy. Acute myeloid leukaemia accounted for 48.6 percent of cases and acute lymphoblastic leukaemia for 24.7 percent, with lymphoma making up a further 15.6 percent. Notably, 24.8 percent of patients developed the infection despite receiving systemic antifungal prophylaxis in the days before diagnosis, so-called breakthrough disease, a figure that underscores the limits of current preventive strategies in the era of widespread mould-active azole prophylaxis.

The researchers applied a sophisticated statistical approach rarely used in this field: marginal structural models with inverse probability of treatment weighting, designed to account for time-varying confounders such as neutropenia and corticosteroid exposure that shift over the course of illness and can distort simple treatment comparisons. After this adjustment, neither the class of first-line antifungal, whether echinocandins, liposomal amphotericin B or azoles, nor the choice between fluconazole and other azoles for long-term oral step-down therapy was significantly associated with treatment response or overall survival.

This is a striking result given that international guidelines currently recommend echinocandins or liposomal amphotericin B as initial therapy, followed by an oral azole. Echinocandins were indeed the most common first-line choice in the cohort, used in 58.6 percent of patients, reflecting alignment with contemporary guidance. But the new data suggest clinicians may have greater flexibility than guidelines imply, particularly for patients with prior azole exposure or resistant organisms, where alternatives to fluconazole have previously rested on thin evidence.

Perhaps the most practice-changing observation concerns treatment duration. The median total course of targeted antifungal therapy was 107 days, substantially shorter than the four to seven months reported in earlier cohorts, and no relapses were documented after treatment was stopped during a median post-discontinuation follow-up of 92 days. Because conventional CT abnormalities can persist for many months after the infection has been controlled, with resolution rates of only 30 to 34 percent at three months in prior studies, the findings support the idea that extending therapy until complete radiological resolution may be unnecessary in selected patients.

Advanced imaging may hold the key to individualising that decision. Positron emission tomography combined with CT was performed in 26.1 percent of patients, and in 12 of them, regression of PET/CT findings was explicitly cited as the reason for stopping antifungal therapy. These patients had markedly longer treatment courses than those without PET-guided decisions, a median of 219 versus 88 days. The results complement the earlier French CANHPARI pilot study, which suggested that PET/CT normalisation at three months may better reflect resolution of active inflammation than conventional imaging, and they point toward PET/CT-guided assessment as a promising tool for distinguishing residual infection from immune reconstitution inflammatory syndrome, the paradoxical worsening that occurs as neutrophils recover.

The study also documented under-recognised manifestations of the disease. Nine patients showed central nervous system involvement, with multifocal punctate or ring-enhancing lesions on neuroimaging and, in some cases, positive cerebrospinal fluid beta-D-glucan testing. Skin and soft tissue involvement occurred in nine patients, six of whom had Candida tropicalis infection, a species known for cutaneous dissemination. The authors suggest that clinicians should consider CNS imaging in patients with neurological symptoms or refractory disease, since cerebral spread may be more common than previously appreciated.

Microbiological findings carried their own geographic and therapeutic signals. Candida albicans accounted for 32.4 percent of identified isolates and C. tropicalis for 24.3 percent, with the latter clustering markedly in Taiwan and Colombia. Fluconazole resistance was found in 11 isolates and was far more frequent in C. tropicalis than in other species, affecting 29.6 percent versus 3.6 percent, and nearly two-thirds of resistant cases occurred in the context of breakthrough infection. Yet resistance did not translate into worse survival, and beta-D-glucan, a fungal cell-wall marker measured at baseline in 35 patients, was positive in 88.6 percent, suggesting it may serve as a useful diagnostic adjunct.

Ultimately, the numbers tell a story about host biology rather than pharmacology. Only 4.5 percent of deaths were attributable to the fungal infection itself, even though all-cause mortality reached 33.3 percent, and patients with breakthrough disease actually survived more often than those without, likely because their underlying cancers were in a more favourable state. The authors caution that their comparative treatment analyses were exploratory and not powered to exclude clinically meaningful differences, and they call for prospective studies incorporating immune biomarkers such as cytokine profiles and Candida-specific T-cell responses alongside PET/CT. But for a disease that has been managed largely by tradition and caution, the message is clear: immune recovery, not drug selection, may be the true engine of cure, and shorter, individually tailored treatment could spare vulnerable patients months of toxicity, cost and delayed cancer therapy.

Subject of Research: Treatment strategies and outcomes of chronic disseminated candidiasis in patients with haematological malignancies in the era of antifungal prophylaxis

Article Title: Treatment and outcome of chronic disseminated candidiasis in the era of antifungal prophylaxis: a retrospective multicentre cohort of 111 patients

Article References: Reinhold, I., Ramirez-Sanchez, I. C., Chen, P.-Y., Chen, Y.-C., Dhariwal, A., Lambourne, J., Agrawal, S., Bastida, M., Gudiol, C., Alfandari, S., Di Pilla, A., Khostelidi, S., Aguilar-Zapata, D., Barac, A., Cipriano, A., Ehrlich, S., Grothe, J., Illarramendi, J., Jennrich, S., … Zager, L. (2026). Treatment and outcome of chronic disseminated candidiasis in the era of antifungal prophylaxis: a retrospective multicentre cohort of 111 patients. eClinicalMedicine, 100, Article 104194. https://doi.org/10.1016/j.eclinm.2026.104194

Image Credits: AI Generated

DOI: 10.1016/j.eclinm.2026.104194

Keywords: chronic disseminated candidiasis, hepatosplenic candidiasis, haematological malignancies, antifungal therapy, echinocandins, azoles, Candida tropicalis, fluconazole resistance, PET/CT imaging, immune reconstitution, neutropenia, beta-D-glucan