High-Dimensional Immunophenotyping of Post-COVID-19 and Post-Influenza Patients Reveals Persistent and Specific Immune Signatures After Acute Respiratory Infection

Abstract

Long-term consequences of SARS-CoV-2 infection are unknown since recovered individuals can experience symptoms and latent viral reactivation for months. Indeed, acute post-infection sequelae have also been observed in other respiratory viral infections, including influenza. To characterize post-COVID-19 and post-influenza induced alterations to the cellular immunome, peripheral blood mononuclear cells (PBMCs) were obtained from patients 3 months after recovery from COVID-19 (n = 93) or influenza (n = 25), and from pre-pandemic healthy controls (n = 25). PBMCs were characterized using a 40-plex mass cytometry panel. Principal component analysis (PCA), classification models, and K-means clustering were subsequently applied. PCA identified distinct immune profiles between cohorts, with both post-COVID and post-flu patients displaying an altered chemokine receptor expression compared to pre-pandemic healthy controls. These alterations were more prominent in post-COVID patients since they exhibited highly increased expression of chemokine receptors CXCR3 and CCR6 by various lymphoid populations, while post-influenza patients mainly showed a decrease in CCR4 expression by naïve T cells, monocytes, and conventional dendritic cells. Classification models using immunophenotyping data confirm the three groups, while K-means clustering revealed two subgroups among post-COVID patients, with younger patients showing more pronounced immune alterations in the chemokine receptor profile, independently of long COVID symptoms. In conclusion, post-COVID and post-influenza patients exhibit distinct and unique persistent immune alterations. Understanding these altered immune profiles can guide targeted therapies for post-COVID syndrome and highlight differences in immune recovery from various respiratory infections.

Keywords: immune signature; immunome; post‐COVID‐19; post‐influenza.

Figure 1

Identification of immune cell subsets by mass cytometry. Representative mass‐cytometry plots showing the hierarchical gating strategy used for the identification of immune cell populations and their subsets.

Figure 2

Post‐COVID and post‐flu immune fingerprint. (A) Principal component analysis (PCA) of post‐COVID (orange), post‐flu (green), and healthy donors (HD, blue) based on mass cytometry data. (B) Biplot of the top 10 features that contributed the most to PCA. The colors in the plot represent the contribution (in percentage) for each feature, varying from blue (lower contribution) to red (higher contribution). Box plots of the 10 features contributing to both component 1 of PCA (C) and component 2 (D) are shown. Mann–Whitney test was applied. p‐values < 0.05 were considered as statistically significant (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Figure 3

Patient classification. Receiver operating characteristic (ROC) curves of logistic regression (orange) and random forest (turquoise) models for each of the predicted classes using (A) the training and (B) the testing set. The area under the curve (AUC) value of each model is annotated in each graph. The dashed line indicates a random prediction.

Figure 4

Unsupervised clustering reveals two immunological groups of post‐COVID patients. The heatmap is a representation of selected features across all individuals. The colors in the heatmap represent the Z‐score for each feature, varying from blue (lower expression) to red (higher expression). The features used for the analysis are shown in rows and split into modules according to the dendrogram on the left, which represents the hierarchical similarity between features (euclidean distance, complete method). In columns, each individual is shown together with it diagnose (healthy donors in blue, post‐COVID in orange and post‐flu in green) and split according to cluster data (cluster 1 in red, cluster 2 in navy blue, cluster 3 in purple and cluster 4 in cyan) above the heatmap (euclidean distance, complete method).

Figure 5

Clinical parameters within the two immunological groups of COVID‐19 patients. Post‐COVID patients within clusters 2 and 4 of K‐means were compared based on their (A) age at COVID‐19 diagnose and (B) subsequent hospitalization days, (C) gender, (D) oxygen need during hospitalization, (E) treatment, or (F) subsequent development of Long‐COVID. Mann–Whitney test was applied, and p < 0.05 was considered statistically significant (**p < 0.01).