Our Take
Pneumonia stratification based on lung biology, not clinical appearance, is real and published in Nature Communications—but the path from bronchoalveolar lavage sampling to bedside clinical use remains unbuilt.
Why it matters
Severe pneumonia kills 2.5 million people annually and accounts for 60% of ICU infections. Current one-size-fits-all anti-inflammatory treatments fail or harm patients in trials because they ignore the three mechanically distinct disease subtypes doctors cannot yet distinguish using standard tests.
Do this week
ICU medical directors: flag this Nature Communications paper for your pneumonia/ARDS protocols team now, so you can plan which inflammation markers to track prospectively when simplified typing tools emerge.
Cambridge Researchers Identify Three Distinct Lung Inflammation Patterns in Severe Pneumonia
Researchers at the University of Cambridge analyzed bronchoalveolar lavage fluid—cells and proteins sampled directly from the lungs—in 80 hospitalized patients with suspected severe pneumonia. Using bulk RNA sequencing, immune cell profiling, and cytokine measurement, they identified three biologically distinct subtypes, or pneumotypes, none detectable by standard blood tests.
The first pneumotype (Pn1) accounts for nearly half of cases (49%) and is characterized by immune suppression, epithelial damage, and bleeding in the alveoli despite low inflammation signals. Patients in this group often fail anti-inflammatory therapies because the underlying problem is not excessive inflammation but immune dysfunction and tissue damage.
The second pneumotype (Pn2), found in just under a quarter of cases (23%), shows a balanced immune response and active epithelial-endothelial repair. Patients with this subtype recover fastest and spend the shortest time on mechanical ventilation, even when initially presenting as severely ill as those in other groups.
The third pneumotype (Pn3) accounts for the remainder and resembles "classic" pneumonia, marked by immature neutrophil infiltration and persistent IL-6-STAT3 activation. These patients spend the longest on ventilators and may benefit most from anti-inflammatory drugs. This is the population where immune-modulating therapies show promise, yet current clinical trials mix all three subtypes together, diluting or obscuring benefit signals.
The team noted that these three pneumotypes appeared in patients both with and without confirmed bacterial or viral pneumonia, suggesting that the inflammatory patterns reflect common lung-injury mechanisms regardless of the triggering infection. The researchers published their findings in Nature Communications under the title "Pulmonary inflammation in severe pneumonia is characterised by compartmentalised and mechanistically distinct sub-phenotypes."
Current Pneumonia Trials Fail Because They Ignore Underlying Biology
Severe pneumonia is the leading infectious cause of death worldwide, responsible for an estimated 2.5 million deaths per year (per the researchers' statement). Severe cases require ICU admission and mechanical ventilation; severe pneumonia accounts for six in ten infections managed in intensive care units.
Clinicians have long observed that patients who appear equally ill on presentation—similar fever, oxygen levels, imaging findings, and clinical diagnoses—diverge sharply in their recovery trajectories. Some wean off ventilators within days; others remain critically ill for weeks or die. The mismatch between clinical appearance and outcome has frustrated treatment innovation. Anti-inflammatory therapies have shown mixed or contradictory results in trials, benefiting some cohorts while harming others.
The Cambridge study explains this paradox: pneumonia is not a single disease but three mechanistically distinct conditions that happen to present similarly. A drug that suppresses inflammation works for Pn3 patients but may worsen outcomes in Pn1 patients who are already immunosuppressed. Blood tests fail to capture the compartmentalized inflammation happening inside the lungs, so clinicians cannot currently distinguish which patient needs which strategy.
The Simplification Gap Remains Open
The researchers acknowledge that their typing method—bronchoalveolar lavage with transcriptomics—is too complex and slow for real-time clinical decision-making. Patients cannot wait hours for RNA sequencing results while on mechanical ventilation.
The team is working to develop a simplified tool capable of rapid pneumotype classification at the bedside, but no such tool is yet available. Until then, the clinical application is limited to retrospective analysis, research cohort stratification, and design of future trials that enroll by pneumotype rather than clinical syndrome.
For practitioners, the immediate value is conceptual: it provides a biological rationale for failed trials and a framework for designing future studies that match anti-inflammatory or immune-modulating therapies to the patients most likely to benefit. It also suggests that biomarkers from bronchoalveolar fluid—not just blood—may be necessary to predict treatment response. Larger studies will test whether additional pneumotypes exist and whether simpler surrogate markers (perhaps measurable in blood or exhaled breath) can approximate the lung-fluid signatures that now define the three subtypes.