CT Clues Distinguish Pseudomonas From Aspergillosis

Key Takeaways

• At a single tertiary center in Seoul, blinded chest CT review in immunocompromised patients found that approximately one-third of Pseudomonas aeruginosa pneumonia cases were read as invasive pulmonary aspergillosis.
• Pseudomonas aeruginosa pneumonia often shared invasive pulmonary aspergillosis-like CT appearances, including macronodules, mass-shaped consolidation, and cavitary lesions.
• Ill-defined centrilobular nodules and peribronchial consolidation were more common in Pseudomonas aeruginosa pneumonia than in invasive pulmonary aspergillosis, although the authors described these as supplementary CT clues with high interobserver variability.
• Solid organ transplantation, peribronchial consolidation, and ill-defined centrilobular nodules remained independent predictors of Pseudomonas aeruginosa pneumonia in this cohort, although the authors cautioned that the CT signs had high interobserver variability and should be treated as supplementary rather than definitive, and that probable invasive pulmonary aspergillosis may have been underrepresented in solid organ transplant recipients because galactomannan-based criteria are less sensitive in that population.

Nodular, cavitary, or mass-like pneumonia on chest computed tomography (CT) can blur the visual boundary between bacterial and fungal infection in immunocompromised patients, especially transplant recipients and patients with hematologic malignancies. When Pseudomonas aeruginosa (PA) pneumonia enters the same radiologic territory as invasive pulmonary aspergillosis (IPA), early classification may depend on pattern recognition before microbiology is complete. Investigators at Asan Medical Center in Seoul, South Korea, examined how often those appearances converged and which findings helped separate the 2 diagnoses.

In a retrospective single-center J Korean Medical Science analysis of PA pneumonia mimicking IPA on CT, investigators identified immunocompromised patients with confirmed or probable PA pneumonia from 2014 through 2023 and compared them with patients who had proven or probable IPA by 2020 European Organisation for Research and Treatment of Cancer/Mycoses Study Group Education and Research Consortium criteria, including compatible imaging plus mycologic evidence such as serum or bronchoalveolar lavage (BAL) galactomannan. Final cohorts included 33 PA cases and 75 IPA cases. A 2-part CT evaluation used 33 PA and 25 randomly selected IPA cases for blinded review, then compared CT features in 33 PA and 73 IPA cases after 2 IPA scans with inadequate image quality were excluded.

On blinded review, 36.3% (12/33) of PA cases were unanimously misclassified as IPA. Reader sensitivities for PA were 60.6% and 33.3%, with fair agreement at kappa 0.377. Sensitivity for identifying IPA was higher for both thoracic radiologists, underscoring how readily bacterial infection assumed a fungal-looking CT pattern in this cohort.

PA frequently showed IPA-like macronodules, mass-shaped consolidation, and cavitary lesions, but a bronchocentric pattern more often separated the groups. Ill-defined centrilobular nodules were seen in 66.7% of PA cases versus 20.5% of IPA cases, and peribronchial consolidation in 57.6% versus 20.5%. Halo sign, reverse halo sign, and internal low attenuation were reported more often in IPA, while multivariate analysis retained solid organ transplantation, peribronchial consolidation, and ill-defined centrilobular nodules as independent predictors of PA pneumonia in this cohort; the authors cautioned that the CT signs had high interobserver variability and were supplementary rather than definitive, and that IPA may have been underrepresented in solid organ transplant recipients because galactomannan-based criteria are less sensitive in that population.

The authors noted that CT-only separation remained imperfect because probable IPA classification incorporated radiologic criteria suggestive of IPA, creating potential verification bias. They also described the comparison as retrospective and single center, with possible selection bias, no IPA cases after lung transplantation in this cohort, and no stratified analysis across underlying immunocompromised conditions because the diagnostic groups were imbalanced. The PA-associated bronchocentric findings were not specific, appearing in a subset of airway-invasive IPA cases. Because the data came from a tertiary center in Seoul, the findings are best framed within the immunocompromised populations studied rather than assumed to translate directly to North American practice.

According to the investigators, PA pneumonia can closely resemble IPA on CT in immunocompromised patients, although this comparison excluded cases with PA-Aspergillus and other listed co-infections, so applicability to mixed infections is uncertain. They identified transplant history together with peribronchial consolidation and ill-defined centrilobular nodules as features associated with PA rather than IPA in this cohort, although the authors noted that probable IPA may have been underrepresented in solid organ transplant recipients because galactomannan testing is less sensitive in that population, while emphasizing that radiologic overlap between the 2 infections remained substantial.

Clinician Questions

Which immunocompromised patients were represented in this PA versus IPA CT comparison?

The cohort came from a single tertiary center in Seoul and focused on immunocompromised patients, particularly solid organ transplant recipients and patients with hematologic malignancies. The comparison was limited to confirmed or probable PA pneumonia versus proven or probable IPA, so applicability is best kept to similar immunocompromised populations rather than assumed broadly.

How was probable PA pneumonia distinguished from probable IPA in this cohort?

Probable PA pneumonia required a positive culture from a nonsterile site together with clinical improvement without antifungal therapy, whereas probable IPA followed 2020 European Organisation for Research and Treatment of Cancer/Mycoses Study Group Education and Research Consortium criteria requiring compatible radiology plus positive serum or BAL galactomannan. Those case definitions shaped how PA and IPA were compared throughout the study.

What CT features favored IPA rather than PA in this comparison?

Halo sign, reverse halo sign, and internal low attenuation were reported more often in IPA than in PA, providing a contrasting pattern to the bronchocentric findings linked more often to PA pneumonia.

Why did the authors say CT findings alone had limits for separating PA pneumonia from IPA?

CT findings alone had limits because radiologists had difficulty distinguishing PA pneumonia from IPA on blinded review, probable IPA classification incorporated radiologic features and may have introduced verification bias, and the bronchocentric findings associated with PA also appeared in some airway-invasive IPA cases. Those factors left the distinction between the 2 infections imperfect even within this immunocompromised cohort.