Anticipating the evolution of septic patients with community-acquired pneumonia (CAP) is challenging for front-line physicians in the Emergency Department (ED). Prognosis depends mainly on early identification, antibiotics, organ support, but also immune status. The objective of this proof-of-concept study was to perform a cluster analysis to investigate whether specific phenotypes, including cellular immunology parameters, are associated with the prognosis in patients with CAP presenting to the ED. We performed an exploratory study in the ED of Limoges university hospital (France) on patients with a confirmed CAP. Deterioration was defined by a composite criterion monitored during 7 days following admission: (i) acute respiratory failure with a high flow oxygen requirement, (ii) subsequent ICU admission, (iii) shock, (iv) worsening of organ dysfunction, and (v) in-hospital mortality. Multicolor Flow Cytometry (MFC) was performed within 12 h after ED admission. Monocyte HLA-DR (mHLA-DR) panels consisting of 11 colors for neutrophils and eight colors for lymphocytes were utilized. Phenotypes were defined using non-supervised hierarchical clustering, including 65 clinical, biological, and immunological variables. During 5 months, 63 patients were prospectively studied (age = 66 ± 19 years
38 men [60%]
SOFA score = 2.6 ± 1.5
Sepsis = 71%
in-hospital mortality = 5%) of whom 11 patients (17%) were assigned to the deterioration group. Upon admission, we observed no differences in any markers or in the demographic or clinical characteristics of the patients. In contrast, by performing hierarchical clustering, we identified three groups: Cluster #1 corresponded to a population with a low deterioration (5%) compared with Clusters #2 (23%) and #3 (31%). Markers from the myeloid lineage, including mHLA-DR, immature neutrophils, and CD64+ neutrophils, were among the parameters discriminating for cluster construction. Cluster #3 displayed the most severe profile, characterized by elevated markers such as CRP, PCT, and immature granulocytes, along with reduced mHLA-DR levels. A clustering strategy, based on myeloid markers obtained through flow cytometry, provided prognostic insights by identifying three phenotypes with distinct outcomes, while none of the individual markers studied (n = 65, both clinical and biological) showed similar prognostic value. A panel of myeloid markers, alongside clinical management, could optimize patient triage and resource allocation upon ED admission.