The immune checkpoints storm in COVID‐19: Role as severity markers at emergency department admission

Dear Editor, Despite the improvement in prophylaxis by vaccination, severe acute respiratory syndrome coronavirus 2 (SARSCoV-2) infection and its associated pathology, COVID19, still constitute a challenge. Clinicians lack effective treatments and robust predictors to identify patients who will develop severe disease during their stay. The role of cytokines (e.g., IL-6) as biomarkers and therapeutic targets was evaluated at the beginning of the pandemic. However, their relevance are still being discussed.1 Patients with severe COVID-19 show an impairment of the immune system allowing secondary infections,2–5 and numerous studies have found how T cell polyfunctionality decreases.5,6 In this regard, immune checkpoints (ICs), a family of molecules known for their ability to modulate immune response and induce T cell exhaustion and apoptosis, become interesting not only as potential early biomarkers of patients’ evolution but also as possible pharmaceutical targets.7 Thus, we recruited 69 patients and 15 healthy volunteers (HVs) from the emergency department (ED) of La PazUniversity Hospital. We analysed nine ICs on admission, previous to any treatment, in plasma from patients classified according to their outcome: mild (outpatients + hospitalized with no O2 requirement, n = 29), severe (hospitalized with O2 requirement, n= 26) and deceased (exitus, n= 14; 28-daymortality accordingWHOU07.1 code) (Figure 1A,B and Supplemental Figure S1 and Table S1). We found sCD25, sTim-3, Galectin-9, sPD-L1 and sCD86 exhibited differences between groups (Figure 1C–L). sCD25, sTim-3 and Galectin-9 showed to be increased in patients versus HV despite the fact that they developed mild disease (Figure 1C–E). In addition, sCD25, sTim-3, Galectin-9, sPDL1 and sPD-1 showed increased levels in severe patients and/or exitus compared with mild ones (Figure 1C–G). In contrast, the co-stimulatory molecule CD86 (sCD86) exhibited lower levels in mild and exitus patients compared with severe ones (Figure 1H). Levels of other ICs


The immune checkpoints storm in COVID-19: Role as severity markers at emergency department admission
Dear Editor, Despite the improvement in prophylaxis by vaccination, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and its associated pathology, COVID-19, still constitute a challenge. Clinicians lack effective treatments and robust predictors to identify patients who will develop severe disease during their stay. The role of cytokines (e.g., IL-6) as biomarkers and therapeutic targets was evaluated at the beginning of the pandemic. However, their relevance are still being discussed. 1 Patients with severe COVID-19 show an impairment of the immune system allowing secondary infections, [2][3][4][5] and numerous studies have found how T cell polyfunctionality decreases. 5,6 In this regard, immune checkpoints (ICs), a family of molecules known for their ability to modulate immune response and induce T cell exhaustion and apoptosis, become interesting not only as potential early biomarkers of patients' evolution but also as possible pharmaceutical targets. 7 Thus, we recruited 69 patients and 15 healthy volunteers (HVs) from the emergency department (ED) of La Paz University Hospital. We analysed nine ICs on admission, previous to any treatment, in plasma from patients classified according to their outcome: mild (outpatients + hospitalized with no O 2 requirement, n = 29), severe (hospitalized with O 2 requirement, n = 26) and deceased (exitus, n = 14; 28-day mortality according WHO U07.1 code) ( Figure 1A,B and Supplemental Figure S1 and Table S1). We found sCD25, sTim-3, Galectin-9, sPD-L1 and sCD86 exhibited differences between groups ( Figure 1C-L). sCD25, sTim-3 and Galectin-9 showed to be increased in patients versus HV despite the fact that they developed mild disease (Figure 1C-E). In addition, sCD25, sTim-3, Galectin-9, sPD-L1 and sPD-1 showed increased levels in severe patients and/or exitus compared with mild ones ( Figure 1C-G). In contrast, the co-stimulatory molecule CD86 (sCD86) exhibited lower levels in mild and exitus patients compared with severe ones ( Figure 1H Figure 1I-L. Note that similar results were obtained when patients were classified according to a multi-organ failure score (qSOFA) (Supplemental Figure S2). In line with this, IC expression on T lymphocytes was also evaluated (Supplemental Figures S3 and S4).
Next, we focused on the five ICs with significant differences between groups. sCD25, sTim-3, Galectin-9, and sPD-L1 showed a positive correlation with severity (Supplemental Figure S5). sCD86 did not show a patent trend (Supplemental Figure S5G). Severity in COVID-19 patients has been associated with differences in sex, age and comorbidities. 8-10 Nonetheless, we did not find significant differences in IC levels between genders (Supplemental Figure S6) and only slight differences within comorbidities (Supplemental Figure S7). However, levels of both sCD25 and sTim-3 on admission not only correlated with a poor prognosis but also with age (Supplemental Figure  S8). Other correlations are shown in Supplemental Figure  S9.
We moved to study the changes in ICs throughout the hospital stay until discharge or exitus. Despite initial similarities between severe and exitus groups, their evolutions were disparate ( Figure 2). Levels of sCD25, sTim-3 and Galectin-9 for mild and severe patients showed a trend toward the levels reported for HVs, while exitus exhibited a progressive elevation ( Figure 2, right panels). Levels of sCD86 increased or levelled off in patients in the mild and severe groups, in contrast to the decrease observed in exitus ( Figure 2E). Similar results were obtained when the analysis was performed using days from onset of symptoms (Supplemental Figure S10).
To explore the possible role of ICs in both the T cell exhaustion and lymphocytopenia reported in patients with COVID-19, 4,5 we calculated the correlation between absolute lymphocyte count (ALC) and levels of circulating ICs on admission. Levels of sCD25, sTim-3, Galectin-9 and sPD-L1, but not sCD86, showed a negative correlation with ALC (Supplemental Figures S11A and S12). Moreover, negative correlation between T cell proliferation and circulating levels of sCD25, sTim-3 and Galectin-9 was also observed (Supplemental Figure S11B-C). Along these lines, a significant increase in the response and a reduction of apoptosis of CD4+ and CD8+ T cells from COVID-19 patients was experienced when anti-Tim-3 and anti-Galectin-9 blocking antibodies were used ex vivo ( Figure  S11D-G).
We moved to analyse whether ICs are stronger severity biomarkers than cytokines in COVID-19. We analysed the levels of 12 cytokines/chemokines in our cohort. Only CCL-2, CXCL10, IL-6 and IL-8 showed significant differences between groups (Supplemental Figure S13). Next, using an area under the ROC curve (Area Under the Curve/Receiver Operating Characteristic [AUC/ROC]) analysis, we compared ability as predictor of the ICs, chemokines and cytokines. We found sCD25 and sTim-3 exhibited the highest AUCs for both mortality and ICU requirement prediction (Supplemental Figures S14 and S15) of all the molecules studied.
Beyond ICs and cytokines, we studied the predictive potential of 53 variables including epidemiological, physiological and other relevant parameters by AUC/ROC and univariate regression (Table 1). We performed a binary logistic regression model, including the 14 variables with an AUC/ROC > 0.7 and statistical significance in univariate analysis. After 12 steps (Supplemental Table S2), the final model included age, SpO 2 /FiO 2 , D-Dimer, sCD25 and sCD86 as predictors of mortality ( Figure 3A). The score obtained from this model showed an AUC/ROC of 0.9753 for mortality prediction ( Figure 3B). The optimal cut-off calculated by the Youden index showed a sensitivity of 0.93 and a specificity of 1 ( Figure 3C). Note this predictive capacity is remarkably higher compared to the performance of all 53 variables alone (none of them had an AUC/ROC > 0.84; Table 1).
To corroborate the relevance of this score, we estimated it in a prospective validation independent cohort of COVID-19 patients (n = 166, Supplemental Table S3). We classified those patients into two groups according to the cut-off estimated in the discovery cohort ( Figure 3D). The chi-squared test showed statistically significant differences in mortality frequency between the groups (χ 2 = 102.8; z = 10.14; Figure 3D), and the score performed the discrimination of survivors and exitus with a specificity of 0.93 and a sensitivity of 0.96 ( Figure 3E).
In summary, our data demonstrated IC levels on admission are better mortality predictors than other biomarkers such as cytokines in COVID-19. Combination of ICs and other easily measurable parameters identifies at ED admission those patients with the worst outcome. In addition, our ex vivo assays suggested the potential of ICs as pharmaceutical targets.

A C K N O W L E D G E M E N T S
This study was supported by Fundación Familia Alonso, Santander Bank, Reale Seguros, Fundación Mutua Madrileña, Fundación Uria, Fundación Caixa and Ayuntamiento de Madrid. In memoriam of Norma López-Collazo.