Fungal Lung Infection Study Finds Diagnostic Criteria May Miss Real-World Cases

Invasive pulmonary aspergillosis, a devastating fungal infection of the lungs, has long been considered a disease of carefully defined risk groups, chiefly patients with hematologic malignancies whose immune systems have been battered by chemotherapy and prolonged neutropenia. But a new retrospective cohort study from Istanbul, published in BMC Infectious Diseases, suggests that the real world is far messier than the diagnostic frameworks clinicians rely on. In a heterogeneous group of 160 patients treated between January 2018 and June 2020, researchers led by Eda Alp of Başakşehir Çam and Sakura City Hospital, together with colleagues from Istanbul University’s Istanbul Faculty of Medicine and other Turkish institutions, found that the widely used EORTC/MSG classification criteria captured only a fraction of genuinely suspected cases, and that overall mortality in the cohort reached a striking 66.9 percent.

The European Organisation for Research and Treatment of Cancer and the Mycoses Study Group, whose joint criteria have anchored invasive fungal disease research for decades, sort suspected cases into three tiers: possible, probable, and proven invasive aspergillosis. Host factors such as neutropenia, steroid exposure, and prior stem cell transplantation are weighed alongside mycological evidence, typically galactomannan antigen detection or culture, and clinical imaging features like the halo sign or air-crescent sign on computed tomography. In the Istanbul cohort, only a single patient, 0.6 percent, met the strictest proven category. Fifty-one patients, 31.9 percent, were classified as probable, while 87 patients, 54.4 percent, fell into the possible category. Critically, 21 patients, or 13.1 percent, could not be classified at all despite clinical suspicion strong enough that the investigators judged no alternative etiology could explain their findings.

That unclassified group is the study’s most provocative finding. These were patients whom experienced infectious diseases physicians believed had invasive pulmonary aspergillosis, yet they failed to satisfy the formal criteria, often because they lacked the canonical host factors or the characteristic radiological patterns. The authors argue that such cases indicate the EORTC/MSG criteria should be reviewed, particularly for patients outside the classical hematologic risk groups, because rigid application risks delaying clinical decision-making. In a disease where every day of delayed antifungal therapy can be fatal, a diagnostic framework that leaves more than one in ten clinically suspected cases in limbo is more than an academic inconvenience; it is a potential barrier to timely treatment.

The cohort itself reflected the changing face of the disease. Of the 160 patients, 97 were male and the mean age was 51.39 years, with a standard deviation of 14.27. Three quarters of the patients, 120 in total, had an underlying hematologic diagnosis, and within this group 38 had received allogeneic stem cell transplants while 25 had undergone autologous transplantation. Yet the remaining 40 patients had no hematologic condition at all, representing the non-classical populations, including critically ill and otherwise immunocompromised individuals, in whom aspergillosis is increasingly recognized. Graft-versus-host disease, the immunological complication of allogeneic transplantation in which donor immune cells attack recipient tissues, was significantly more frequent among patients classified as proven or probable, a statistically meaningful association with a p value of 0.042 that fits the biology, since graft-versus-host disease and its steroid treatment compound immunosuppression.

Imaging told its own nuanced story. Computed tomography findings consistent with the EORTC/MSG clinical criteria were present in 144 patients, 90 percent of the cohort, which on the surface suggests the radiological arm of the framework performs well. But the details mattered. Cavity formation, a late and destructive pattern of fungal invasion, was significantly more frequent among patients with mycological evidence of infection, with a p value of 0.037. More troubling was the finding that prolonged corticosteroid use was associated with non-criteria CT patterns, with a p value of 0.048. In other words, patients on long-term steroids, a growing population spanning transplant recipients, oncology patients, and those with autoimmune disease, often did not display the textbook halo signs and nodules the criteria expect. Their infections presented atypically, slipping through the radiological net.

Treatment patterns in the cohort also tracked with the diagnostic categories. Voriconazole, the triazole antifungal that guidelines favor as first-line therapy for invasive aspergillosis, was more often chosen as initial treatment in patients with clinical imaging evidence, a statistically significant association with a p value of 0.034. This suggests that clinicians anchor their therapeutic decisions on visible pulmonary lesions, which may explain why patients lacking criteria-consistent imaging face diagnostic and therapeutic delays. The authors also highlight a practical constraint: invasive diagnostic procedures such as bronchoscopy with bronchoalveolar lavage are often difficult or impossible in the sickest patients, underscoring the need for reliable non-invasive diagnostic methods that could confirm infection without subjecting fragile patients to risky procedures.

The mortality data are sobering. Six-week all-cause mortality stood at 21.2 percent and 12-week mortality at 31.9 percent, figures broadly consistent with published hematologic-malignancy cohorts, which typically report mortality in the range of 25 to 35 percent. But over the full follow-up period, 107 of the 160 patients died, a 66.9 percent overall mortality rate, with a median survival of 240 days. The authors note that this figure is higher than in hematologic cohorts but comparable to intensive care unit and non-hematologic populations, reinforcing the point that aspergillosis arising in atypical hosts carries a prognosis at least as grim as the classical one. The comparison also implies that deaths in this heterogeneous group were driven not only by the fungus itself but by the burden of underlying disease, a distinction with real consequences for how trials and outcome measures are designed.

Two predictors of death emerged from the analysis. Patients who died were significantly older than survivors, with a p value of 0.029, consistent with the accumulated vulnerabilities of advanced age. More intriguingly, pleural effusion, the accumulation of fluid in the space surrounding the lung, visible on chest CT, was associated with death, with an odds ratio of 2.326 and a 95 percent confidence interval of 1.156 to 4.681, yielding a p value of 0.018. However, the association lost significance when the analysis was restricted to the 6-week and 12-week timepoints. The authors interpret this pattern carefully: pleural effusion appears to signal a comorbidity-driven rather than IPA-specific prognostic risk, marking patients whose overall physiological reserve is compromised rather than those in whom the fungal infection itself is uniquely aggressive. It is a subtle but important distinction for clinicians weighing prognosis at the bedside.

The study’s design carries the usual caveats of retrospective cohort research. Classification depended on records assembled for clinical rather than research purposes, and the single-center design limits generalizability, even if the center is a major referral hospital. The authors received no external funding and declared no competing interests, and the work was approved by the Clinical Research Ethics Committee of Istanbul University Istanbul Faculty of Medicine. Still, the scale of the cohort, 160 patients with findings unexplained by any other etiology, gives the findings weight, and the message is consistent with a growing international conversation about whether diagnostic criteria built around neutropenic hematology patients can serve an era in which aspergillosis strikes influenza patients, COVID-19 survivors, steroid-treated transplant recipients, and the critically ill.

What emerges from Istanbul is a portrait of a disease outpacing its diagnostic tools. The EORTC/MSG criteria remain indispensable for standardizing clinical trials and enabling comparison across studies, but this real-world cohort shows their limits: most patients cluster in the weakest possible category, a substantial minority escape classification entirely, and the radiological signatures the criteria demand may be absent precisely in the growing population of corticosteroid-treated patients. With overall mortality approaching 67 percent and median survival measured in months, the stakes of closing that gap are high. The authors’ call is twofold: maintain vigilance and early diagnosis even when criteria are not met, and invest in non-invasive diagnostics that can confirm infection quickly in patients who cannot undergo invasive sampling. Until then, clinicians facing a suspicious lung lesion in an atypical host may need to trust their clinical judgment as much as the checklist.

Subject of Research: Real-world performance of EORTC/MSG diagnostic criteria and mortality predictors in invasive pulmonary aspergillosis

Article Title: Invasive pulmonary aspergillosis in a heterogeneous patient cohort: real-world performance of EORTC/MSG criteria and predictors of mortality

Article References: Alp, E., Bali, E. A., Evlice, O., Çağatay, A., Yavuz, S. Ş., & Eraksoy, H. (2026). Invasive pulmonary aspergillosis in a heterogeneous patient cohort: real-world performance of EORTC/MSG criteria and predictors of mortality. BMC Infectious Diseases. https://doi.org/10.1186/s12879-026-14567-x

Image Credits: AI Generated

DOI: 10.1186/s12879-026-14567-x

Keywords: invasive pulmonary aspergillosis, EORTC/MSG criteria, fungal infection, diagnostic criteria, mortality, pleural effusion, corticosteroids, computed tomography, voriconazole, stem cell transplantation, hematologic malignancy, survival analysis