Three biologically distinct pulmonary endotypes are present among mechanically ventilated patients with suspected severe pneumonia, despite broadly similar severity of respiratory failure, according to a recent study.

Severe pneumonia is a major cause of critical illness and a common trigger for acute respiratory distress syndrome (ARDS), but its underlying biology remains unclear.

A phenotype groups patients according to shared clinical or biological characteristics, while an endotype is defined by a distinct underlying biological mechanism. Patients with similar clinical features, such as comparable severity of respiratory failure, may therefore have different endotypes reflecting different inflammatory processes.

Inflammatory phenotypes have previously been identified in peripheral blood in pneumonia-associated sepsis and ARDS, but comparable approaches have not been applied directly in the lungs, despite evidence that inflammatory responses can differ between the lungs and systemic circulation.

Commenting on the unmet need in this area, corresponding author Dr Andrew Conway Morris, associate professor of anaesthesia at the University of Cambridge and honorary consultant in intensive care medicine at Cambridge University Hospitals NHS Foundation Trust, said: ‘The current approach of classifying patients by their clinical syndromes – sepsis, ARDS and so on – without looking at the underlying biology risks missing what’s key. Instead of asking “Does this patient have pneumonia?”, we should be asking “What’s the inflammatory pattern in this patient’s lungs?”.’

His research team therefore profiled bronchoalveolar lavage (BAL) and blood samples from mechanically ventilated patients with clinically suspected pneumonia. They examined gene expression alongside inflammatory protein concentrations, cellular composition and respiratory pathogen testing.

Published in the journal Nature Communications, the study, for which Dr Conway Morris is corresponding author, recruited 95 patients admitted to a mixed medical-surgical intensive care unit, of whom 80 had BAL samples with RNA suitable for sequencing.

Clustering of alveolar gene expression identified three biologically distinct pulmonary endotypes, termed Pneumotypes (Pn) 1, 2 and 3. The three groups contained patients with and without confirmed pneumonia and had similar severity of respiratory failure, with ARDS present in around 58% of each group.

Inflammatory mechanisms across Pneumotypes

Pn1, the largest group, was characterised by an expanded macrophage population alongside evidence of epithelial injury and suppression of epithelial repair pathways. The authors proposed that this reflected lung injury occurring without prominent neutrophilic inflammation.

Pn3 showed a different pattern, with increased neutrophils and monocytes, higher concentrations of multiple inflammatory proteins in BAL and evidence of inflammasome activation.

Genes and proteins associated with interleukin-6-related signalling and emergency granulopoiesis were enriched, alongside signatures consistent with immature neutrophil infiltration. Genes involved in alveolar fluid clearance and surfactant function were also downregulated.

Pn2 showed intermediate pulmonary inflammatory protein levels and gene-expression patterns associated with epithelial and endothelial barrier repair. The researchers described this as the most adaptive of the three Pneumotypes, combining a balanced immune response with pro-resolution epithelial and endothelial responses.

Clinical outcomes also differed between the Pneumotypes. Pn2 had the fastest resolution of respiratory failure, with significantly shorter time to extubation than Pn3. Relative to Pn2, the hazard ratio for successful extubation was 0.31 for Pn3 (95% CI 0.14–0.71; P=0.006). Differences in one-year mortality between Pneumotypes were not statistically significant.

Pulmonary endotypes poorly reflected in blood

The researchers found that the pulmonary endotypes were largely compartmentalised to the lungs. Among 48 inflammatory proteins measured, 35 differed significantly between Pneumotypes in BAL, whereas plasma inflammatory proteins were largely comparable. Peripheral blood gene-expression patterns also did not reliably distinguish the three pulmonary endotypes.

The researchers identified similar biological patterns in three external pulmonary datasets, but cautioned that further validation using comparable patient populations, sampling and analytical methods is required.

They also noted limitations including the fact that the study was conducted at a single centre, serial sampling was limited and that additional pulmonary endotypes could emerge in larger or clinically different cohorts.

The researchers concluded that patients with severe pneumonia and acute lung injury can have pulmonary endotypes with distinct underlying mechanisms and outcomes. The findings also suggested that relying on blood phenotyping alone may not adequately characterise inflammatory processes occurring within the lung.

Study author Dr Vilas Navapurkar, consultant anaesthetist at Cambridge University Hospitals NHS Foundation Trust, said: ‘If we know which subtype of pneumonia an individual has, we can potentially tailor their treatment more precisely, boosting the immune response in some, while calming harmful inflammation in others. This has the potential to help critically ill patients, reduce deaths from pneumonia, shorten ICU stays and cut unnecessary antibiotic use.’

Reference
Jeffrey M et al. Pulmonary inflammation in severe pneumonia is characterised by compartmentalised and mechanistically distinct sub-phenotypes. Nat Commun 2026;17:5312.