Chest CT patterns in Pneumocystis pneumonia vary with immune status and co-infection

Diffuse ground-glass opacities remain the dominant chest CT finding in critically ill patients with Pneumocystis jirovecii pneumonia (PJP), but the imaging picture is far from uniform, and the pattern that predominates on a scan appears to depend on why a patient is immunosuppressed and whether another infection is present in the lungs. Those are the central conclusions of a retrospective, single-center study published in the journal Critical Care: Sepsis and Severe Infection, which systematically reviewed chest CT examinations of 106 adults admitted to an intensive care unit with confirmed PJP over a 15-year period.

The research team, led by Virginie Lemiale and colleagues, analyzed consecutive adult patients admitted to the ICU with confirmed Pneumocystis jirovecii pneumonia between 2006 and 2020. To be included, patients had to have undergone a chest CT within seven days surrounding ICU admission; in practice, the scans were performed very early, with a median delay of zero days from ICU admission. Diagnosis of PJP was based on host factors, clinical symptoms, radiological signs, and microbiological confirmation by immunofluorescence or PCR. The study was approved by the ethics committee of the Société de Réanimation de Langue Française, and the need for specific informed consent was waived in accordance with French law given the retrospective, observational design.

The cohort was divided into three groups according to the underlying cause of immunosuppression: 39 patients (37%) were living with HIV, 34 patients (32%) had hematological malignancies including allogeneic stem cell transplant recipients, and 33 patients (31%) had other immunosuppressive conditions. Among this last group, 18 patients had received organ transplants, 12 were receiving immunosuppressive treatment, mostly steroids, for autoimmune disease, and 3 had metastatic cancer. The median age of the cohort was 56 years, with an interquartile range of 48 to 64, and 33 patients (31%) were women. Severity of illness at inclusion was substantial: the median PaO2/FiO2 ratio was 167, and 47 patients (44%) required invasive mechanical ventilation during the ICU stay. Overall ICU mortality was 10.4%, with 11 deaths.

Across the entire cohort, diffuse ground-glass opacities were by far the most frequent CT pattern, observed in 97 patients (91%). Alveolar consolidations were present in 70 patients (66%), the crazy-paving pattern in 32 (30%), nodular lesions in 19 (18%), and cystic lung lesions in 15 (14%). Additional findings included bronchiectasis in 38 patients (36%), pleural effusions in 15 (14%), and pneumothorax in 4 (4%). When the reviewers determined which single pattern predominated on each scan, ground-glass opacities were predominant in 72 patients (68%), followed by consolidation in 12 (11%), crazy-paving in 9 (8.5%), nodules in 8 (7.5%), and reticular thickening in 4 (3.8%).

A striking minority of patients showed none of the classical imaging features at all. Six patients in the cohort had neither ground-glass opacity nor crazy-paving on their CT scan; all six had bilateral nodular lesions instead. Four of them had chronic lymphoid leukemia, and the remaining two were receiving immunosuppressive treatment with steroids combined with ciclosporin or rituximab. Three of these six patients had a documented co-infection at the time of PJP diagnosis. This observation underpins one of the study’s key messages: PJP should not be excluded solely on the basis of atypical imaging features, because radiological phenotypes may differ depending on host-related factors.

To assess how reliably the patterns could be identified, all CT examinations were reviewed independently by two experienced physicians, a senior thoracic radiologist and an intensivist with expertise in acute respiratory failure and opportunistic pulmonary infections, using a standardized reading grid of predefined features. Neither reader was blinded to the PJP diagnosis, and disagreements were resolved by consensus. Interobserver agreement was excellent for ground-glass opacities, with a kappa coefficient of 0.91, and substantial for nodular lesions and alveolar consolidations, with kappa values of 0.69 and 0.76 respectively. Agreement was moderate for the crazy-paving pattern, at 0.40. Given the radiologist’s higher level of expertise in thoracic imaging, only that reader’s assessments were retained for the subsequent analyses. The authors suggest that this level of agreement indicates intensivists could rely on their own assessment of the most common PJP patterns, potentially shortening the time to diagnosis and treatment.

When CT patterns were compared across the three immunosuppression groups, reticular thickening was significantly more frequent in the group with other immunosuppressive conditions, while no other individual CT feature differed significantly between groups. However, when the predominant pattern on each scan was considered, significant differences emerged between the three groups (p = 0.03). The study also found that the duration of respiratory symptoms before PJP diagnosis differed markedly between groups, with a median of 21 days in the HIV group compared with 14 days in the hematological malignancy group and 10 days in the other immunosuppression group (p = 0.001). Notably, the severity of acute respiratory failure was similar across groups, with comparable respiratory rates, PaO2/FiO2 ratios, and ARDS rates. Steroid treatment at the time of diagnosis, which 39 patients (36.7%) were receiving, made no significant difference to the predominant pattern or to the presence of ground-glass opacities or alveolar consolidations.

The relationship between imaging and co-infection emerged as one of the most consequential findings. At the time of PJP diagnosis, 29 patients (27%) had a documented pulmonary co-infection, categorized as opportunistic, bacterial, or viral. Among these, 10 had opportunistic infections, including five with cytomegalovirus, two with fungal infections, two with tuberculosis, and one with combined aspergillus and CMV infection; 16 had bacterial co-infections and 3 had viral co-infections. The predominant CT pattern differed significantly according to co-infection status (p = 0.027). Nodular lesions or alveolar consolidations were the predominant pattern in 10 of 29 patients (34%) with co-infection, compared with only 10 of 77 patients (13%) without co-infection. Conversely, ground-glass opacities or crazy-paving were the main radiological presentation in 64 of 77 patients (83%) without co-infection, versus 17 of 29 patients (59%) with co-infection. The presence of co-infection was also associated with a trend toward more frequent alveolar consolidations, observed in 83% of patients with co-infection compared with 60% of those without (p = 0.06). The authors suggest that focal, consolidative, or nodular patterns may reflect superimposed infectious processes rather than PJP alone, and they recommend that a predominantly nodular or alveolar consolidative pattern on CT should prompt a comprehensive microbiological work-up for concomitant bacterial and viral pathogens in addition to specific testing for Pneumocystis jirovecii. No difference in pattern was found among the different types of co-infection.

The study also explored links between imaging and disease severity. There was no significant correlation between the predominant CT pattern and the need for intubation (p = 0.064), although ground-glass opacities were more frequent in non-intubated patients (94.9% versus 80.9%, p = 0.05) and crazy-paving was more frequent in intubated patients (36.5% versus 25.5%, p = 0.007). The authors interpret these associations cautiously, noting that several mechanisms, including the amount of alveolar inflammation, interstitial involvement, and pre-existing pulmonary lesions, may contribute to the differences in CT patterns observed between immunosuppression groups. They also note that cystic lesions were relatively uncommon in their cohort, consistent with previous reports describing cysts as a less frequent manifestation of PJP, particularly in non-HIV immunocompromised patients.

The authors acknowledge several limitations. The retrospective design exposes the analysis to potential selection and information biases, and clinical variables outside those used for PJP diagnosis were not recorded, including data on cardiac function and lymphangitic lesions in patients with cancer, which may have influenced the findings regarding reticular thickening in the other-immunosuppression group. CT scans were not obtained at a standardized stage of disease progression, which may have influenced the observed patterns. Bronchoalveolar lavage was not performed systematically, so the true prevalence of concomitant viral or bacterial infections may have been underestimated. The single-center design may limit generalizability, and the physicians who analyzed the scans were not blinded to the PJP diagnosis, although the authors point to the very good interobserver agreement as a mitigating factor. Within those constraints, the study’s conclusion stands: diffuse ground-glass opacities remain the major CT feature of PJP in critically ill patients, but a broad spectrum of additional imaging patterns varies with the underlying cause of immunosuppression, and a predominance of nodular lesions or alveolar consolidations on CT should raise suspicion for concomitant pulmonary co-infection and prompt an extended microbiological work-up.

Subject of Research: A retrospective ICU study of chest CT patterns in Pneumocystis jirovecii pneumonia and their association with immunosuppression type and pulmonary co-infection.

Article Title: Thoracic imaging in patients with Pneumocystis jirovecii pneumonia: CT-scan patterns correlation with underlying immunosuppression and co-infection

Article References: Virginie, L., de Margerie-Mellon, C., Valade, S., Mourtada, L., Caradec, E., Bonny, V., Yvin, E., Lafarge, A., & Friol, A. (2026). Thoracic imaging in patients with Pneumocystis jirovecii pneumonia: CT-scan patterns correlation with underlying immunosuppression and co-infection. Critical Care: Sepsis and Severe Infection, 1(1), Article 6. https://doi.org/10.1186/s44541-026-00007-5

Image Credits: AI Generated

DOI: 10.1186/s44541-026-00007-5

Keywords: Pneumocystis jirovecii pneumonia, chest CT, ground-glass opacities, immunosuppression, HIV, hematological malignancies, pulmonary co-infection, intensive care, radiology, acute respiratory failure, interobserver agreement, critical care