Ferroptosis and pyroptosis signatures in critical COVID-19 patients

Critical COVID-19 patients admitted to the intensive care unit (ICU) frequently suffer from severe multiple organ dysfunction with underlying widespread cell death. Ferroptosis and pyroptosis are two detrimental forms of regulated cell death that could constitute new therapeutic targets. We enrolled 120 critical COVID-19 patients in a two-center prospective cohort study to monitor systemic markers of ferroptosis, iron dyshomeostasis, pyroptosis, pneumocyte cell death and cell damage on the first three consecutive days after ICU admission. Plasma of 20 post-operative ICU patients (PO) and 39 healthy controls (HC) without organ failure served as controls. Subsets of COVID-19 patients displayed increases in individual biomarkers compared to controls. Unsupervised clustering was used to discern latent clusters of COVID-19 patients based on biomarker profiles. Pyroptosis-related interleukin-18 accompanied by high pneumocyte cell death was independently associated with higher odds at mechanical ventilation, while the subgroup with high interleuking-1 beta (but limited pneumocyte cell death) displayed reduced odds at mechanical ventilation and lower mortality hazard. Meanwhile, iron dyshomeostasis with a tendency towards higher ferroptosis marker malondialdehyde had no association with outcome, except for the small subset of patients with very high catalytic iron independently associated with reduced survival. Forty percent of patients did not have a clear signature of the cell death mechanisms studied in this cohort. Moreover, repeated moderate levels of soluble receptor of advanced glycation end products and growth differentiation factor 15 during the first three days after ICU admission are independently associated with adverse clinical outcome compared to sustained lower levels. Altogether, the data point towards distinct subgroups in this cohort of critical COVID-19 patients with different systemic signatures of pyroptosis, iron dyshomeostasis, ferroptosis or pneumocyte cell death markers that have different outcomes in ICU. The distinct groups may allow ‘personalized’ treatment allocation in critical COVID-19 based on systemic biomarker profiles.

number of clusters and the elbow method confirmed that separation into 5 clusters provide acceptable separation between clusters [4].Model-based clustering is frequently used to discern subgroups with distinct profiles of independent variables within heterogenous patient cohorts [5][6][7].

Unsupervised clustering of critical COVID-19 patients by trajectories/kinetics of individual biomarkers using longitudinal k-means clustering
In this study, levels of each biomarker were measured on the first three consecutive days after ICU admission in every critical COVID-19 patient.We aimed to assess the possible relation of kinetics in biomarker levels and clinical disease score in the same time interval, as well as clinical outcome in ICU.
Therefore, we constructed trajectories of each biomarker level per individual patient and separated patients in clusters with similar trajectories by longitudinal k-means clustering using the Latrend and Kml packages in R version 4.1.1[8,9].The optimal number of clusters for each biomarker was determined by applying the elbow method to find the lowest BIC, lowest Mean absolute error weighted by cluster-assignment probability (WMAE) and highest Model log-likelihood (logLik) in function of the number of clusters.Similar methods were used elsewhere to study the impact of changes in independent parameters on clinical outcomes [10].
All R coding for the unsupervised clustering can be found at: https://github.com/cedricpeleman.
scatterplot smoothing (LOESS) regressions on scatterplots at the intersections of rows and columns (presented in lower left-hand corner of the figure).T-tests were performed to assess significance of correlation coefficients; * p<0.05, ** p<0.01, *** p<0.001.Supplementary Figure S4.Unsupervised clustering of critical COVID-19 patients based on biomarker profiles by Gaussian Mixture models.(A) Values of Bayesian Information Criterion (BIC) were plotted for a given number of clusters by each model of Gaussian Mixture modelling.The EEI model with 5 clusters resulted in the lowest BIC value.(B) Likewise, values of the Integrated Completedata Likelihood (ICL) were plotted per number of clusters by each fitted Gaussian Mixture model.The EEI model with 5 clusters resulted in one of the lowest ICL values.(C) Values of the Davies-Bouldin's index were plotted for any given number of clusters defined by Gaussian Mixture modelling.Lower levels of this index indicate a better separation between the clusters.Increasing the number of clusters to a total of 5 enhanced the separation.(D) Principle component analysis (PCA) plot was used to visualize the 5 clusters of critical COVID-19 patients constructed by the EEI model of Gaussian mixture modelling.COVID-19 patients into 3 clusters was most appropriate.(B) Graphs represent the summarizing trajectory of each cluster of Fec-based trajectories (in color), while the black lines visualize the kinetic of this biomarker per individual patient that pertains to this cluster.(C) The change in biomarker Fec in each of the three constructed clusters is plotted over the first three days after ICU admission.Clusters A and C both start at moderate levels of biomarker Fec on the first day of measurement.While the former maintains stable moderate levels, the latter represents patients that develop an increase in Fec levels on the next 2 measurements.Cluster B comprises patients that start at low levels of biomarker Fec and maintain low levels thereof.(D-K) Critical COVID-19 patients were separated into clusters with distinct trajectories/kinetics by longitudinal k-means clustering for each individual biomarker.Graphs represent the summarizing trajectory of each biomarker in function of time for the following individual biomarkers: MDA, ferritin, lactoferrin and myoglobin, IL-1β, IL-18, sRAGE and GDF15.The optimal number of clusters for each biomarker was determined using the elbow method on BIC, WMAE and logLik parameters (not shown for each individual biomarker).Supplementary Figure S7.Relation of the number of ventilation-free days (VFDs) in function of myoglobin-trajectory based clusters of critical COVID-19 patients.(A) Boxplot represents the mean, interquartile range and min-max values of the number of VFDs in different clusters of critical COVID-19 patients, as defined by unsupervised clustering based on their trajectories of the biomarker myoglobin.(B) Odds ratios (OR), 95% confidence intervals of OR and p-values were generated when comparing the possible association of different myoglobin-trajectory based clusters with the number of VFDs using multiple linear regression.The effect of different clusters was assessed through generation of dummy variables which compared each of the other myoglobin-trajectory-based clusters with the reference cluster F. Regression was performed with adjustment for age, gender, body mass index and interleukin-6 levels.* p<0.05, ** p<0.01.