Multiple indicators and biomarkers have been proposed, alone or in combination, to identify the most serious COVID-19 cases, but none proved to be entirely effective (1–3).

Pro-adrenomedullin is a multipotent regulatory peptide expressed in different tissues and organs and upregulated by inflammation, hypoxia, bacterial products, and shear stress. Mid-regional pro-ADM (MR-proADM), its stable precursor, is currently considered an effective biomarker of endothelial damage as its increase in plasma seems to correlate with disease severity (4). In fact, it plays a role in vascular permeability, inflammatory cascade, endothelial barrier regulation, and microcirculation performance, as well as essential in maintaining endothelial stability. The increase of MR-proADM has been demonstrated as an indicator of organ dysfunction and failure, and its predictive value has been highlighted in the context of respiratory infections, sepsis, and septic shock (5, 6).

Regarding COVID-19-related severe acute respiratory syndrome (SARS-CoV2), an association between MR-proADM levels and virus-induced endothelial damage is assumed, as endotheliitis has emerged as a prominent feature of the severe COVID-19 disease (7). MR-proADM levels seem to reflect disease progression, allowing the identification of patients who are at the most risk of developing a more severe form of the disease (8). It could even be able to predict SARS-CoV2-induced mortality, although the pathological mechanism underlying this correlation has not been fully clarified. In fact, most studies had limited dimensions and were designed in the context of a pandemic emergency, with heterogenicity of objectives and contexts (2, 9).

In an intensive care unit (ICU) setting, the evidence seems to be particularly limited. A recent systematic review and meta-analysis designed to clarify the use of MR-proADM in severe COVID-19 disease included 21 studies, published between 2020 and 2022 from European countries, addressing the use of pro-adrenomedullin in COVID-19 (9). The analysis included data from 252 patients, only in the ICU setting. At ICU admission, the average MR-proADM level was 1.01 vs. 1.64 in surviving (n = 182) and non-surviving (n = 70) patients, respectively, with the mean differences of MR-proADM values in survivors vs. non-survivors being −0.96 (95% CI: −1.26 to −0.65). Although MR-proADM levels at admission seem to predict mortality in the critical COVID-19 population, a cutoff value able to provide adequate guidance for the use of MR-proADM as an adequate prognostic index is still missing.

Moreover, there are no prospective observational studies exploring the relationship between the host immune response status of SARS-CoV-2-infected patients, both antibody and cellular immunity, and outcomes. It is known that the decreases in the number and function of some lymphocyte populations suggest close monitoring of patient immunological status, as lymphopenia, which is inversely proportional to the severity of the disease, is often reported in severe COVID-19 cases, while little is known about immunoglobulin changes and specific subtypes (2, 10, 11). To date, few studies have comprehensively assessed the dynamic changes in antibody levels and lymphocyte subpopulations in patients with COVID-19, and the results obtained so far are inconsistent and scarcely conclusive (12–14).

A previous preliminary analysis conducted on 57 patients by this group showed that MR-proADM values are higher in non-surviving ICU-COVID-19 patients, its predictive ability compared with other inflammatory biomarkers, and how its changes over time tend to be different in surviving and non-surviving patients (15). In this study, we analyze the evidence obtained by extending the data collection and adding the evaluation of the possible correlation between lymphocyte subpopulations and immunoglobulins.

This prospective, observational study, on a cohort of over 200 critically ill ICU-COVID-19 patients, confirms the validity of the biomarker MR-proADM in predicting mortality (9, 15). Its values, measured within the first 48 h of ICU admission, proved to be significantly higher in patients with a fatal outcome and revealed an ability to discriminate between surviving and non-surviving patients better than other biomarkers commonly used in ICU, such as PCT, C-RP, LDH, and D-dimer. Moreover, patients who had a predictive value of MR-proADM >1.5 nmol/L showed higher mortality, confirming this value as a possible cutoff value in this population [OR of 1.9 (95% CI: 1.220–3.060)] (9).

The increase in MR-proADM, due to a dose-response mechanism induced by the host-pathogen interaction, seems to occur in the initial stage of pathogen recognition, i.e., at the time of hospital admission or even earlier. However, our findings confirm that the increase persists in the following days in accordance with the evolution of the disease, as shown by the trend of the values (48 h, day 3, and day 7). We consider this temporal trend analysis particularly interesting as it could help to overcome some limits of the single-value evaluation.

Other indicators, less studied in this patient setting, such as lymphocyte subpopulations or the level of serum immunoglobulins (subgroups IgA, IgM, and IgG), were not able to predict disease severity and mortality. This is in line with recent findings in the literature (13, 14), which seem to suggest that no defined prognostic value can be uniquely attributed to immunologic biomarkers or cytokines.

Our cohort is relatively young, with a predominance of the male gender (76.1%), without impact on mortality, unlike the results reported by the ISARIC clinical characterization group (27–29). Patients presented with at least one comorbidity in most cases (89%), and three or more in 46.4% of cases, in line with literature evidence (30). Among comorbidities, only cardiovascular diseases (p = 0.0083), already known risk factors (31, 32), and chronic pulmonary diseases (p = 0.0128) were demonstrated to play a significant role in the univariate analysis.

All clinical severity scores (SOFA, median value 7; MuLBSTA, median value 11; and SAPS II, median value 52) highlighted how our population has greater severity compared to previous studies (33–35). This is also confirmed by the high percentage of patients requiring invasive mechanical ventilation (89.9%), vasopressors (71.7%), renal replacement therapy (RRT) (10.1%), and developing septic shock (29.2%). As a result, the length of ICU and hospital stay (13 and 23 days, respectively), and ICU and hospital mortality rates (55.5 and 61%, respectively) were higher than those observed in other studies (36). In our population, co-infections and superinfections are relatively common complications of severe COVID-19; in particular, in our analysis, the presence of superinfections was associated with increased odds of death, in line with other studies showing a positive association between co-infection or superinfection and an increased risk of death, especially in ICU (24).

The biomarkers we analyzed in this study were those already in use or deemed potentially useful at the beginning of the pandemic and during subsequent waves. Many of these were elevated on admission (LDH, D-dimer, NT-proBNP, PCT, myoglobin, CK, ferritin, troponin-I (hs), and IL6). Among these, the mean values of LDH (p < 0.0001), D-dimer (p-value 0.0019), NT-proBNP (p-value < 0.0001), PCT (p-value 0.0251), myoglobin (p-value 0.0356), troponin-I hs (p-value < 0.0001), and IL-6 (p-value 0.0093) seem to be able to identify patients subsequently burdened by major mortality. C-RP, PCT, ferritin, IL-6, and LDH predictive values were found to be altered in most of our critical population, in line with previous literature (2).

The role of generic inflammation biomarkers, to which much attention was given at the beginning of the pandemic, has been greatly reduced in light of the most recent findings, as the elevation of inflammatory biomarkers as well as cytokine parameters does not seem to be able to provide a clear prognostic indication (14, 37–39). In our population, the analysis of the AUC for PCT, LDH, D-dimer, and lymphocytes suggested underperformance compared to MR-proADM (Figure 3) and no predictive ability for C-RP, ferritin, copeptin, lymphocytes subpopulation, and immunoglobulin (Table 2). MR-proADM, instead, appears to be a biomarker with a strong prognostic value, as supported by a series of experimental evidence attributing to ADM an important role in the regulation of vascular and endothelial barrier permeability, inflammatory mediators, and microcirculation (40, 41).

ROC curve analysis of MR-proADM showed that this biomarker has a significantly greater predictive capacity than other biomarkers. The increase in MR-proADM, resulting from a dose-response mechanism induced by the host-pathogen interaction, appears to occur in the early stages of pathogen recognition, at the time of hospital admission or even before. However, it is interesting to note that the increase persists in the following days in accordance with the evolution of the disease, underlying the additional value of this biomarker in predicting the patient's outcome. In line with the evidence of other recent studies, which also proposed the analysis of the trend over time of this marker in the course of sepsis (42), we highlighted how MR-proADM could play a role in monitoring the progression of the disease (8), as evidenced by the results obtained on days 3 and 7. This analysis appeared to be a peculiarity of great interest in this work since it could allow the overcoming of some limitations of the punctual evaluations. In fact, this new biological marker of endothelial damage and vascular permeability could have an important clinical impact, particularly in the ICU setting. As it has been reported that MR-proADM can contribute to the correct triage of patients with COVID-19 in the emergency department (6, 42, 43), its role in guiding new early diagnostic interventions and making possible the anticipation of more intensive treatment, regardless of the causative pathogen, such as bacteria, fungi, or viruses, could be crucial in the ICU setting. MR-proADM could also be used, together with the traditional severity score on admission (SAPS, APACHE, etc.), to predict the outcome but also to allow better monitoring of the patient's course, to evaluate the effectiveness of treatments, and to anticipate an early identification of any possible worsening, especially if collected repeatedly over time. Finally, the MR-proADM trend might safely guide transfer from ICU to general wards without increasing the number of re-admissions and/or mortality. Further studies are needed to confirm these hypotheses.

A further innovative aspect of our study is represented by the analysis of the so-called immunologic biomarkers, namely, lymphocyte subpopulations and circulating levels of immunoglobulins Ig (A, M, and G) (13, 14). It has in fact been hypothesized that the pulmonary involvement, typical of severe forms of COVID-19, may be due to a dysregulated systemic inflammatory response induced by a macrophage activation syndrome that mimics acquired hemophagocytic lymphocytic histiocytosis (43–45). This might be reflected in severe lymphopenia with an inflammatory response and release of a cytokine cascade, previously considered markers of disease severity (46, 47). In fact, it is known that T cells, including CD4+ and CD8+, have an important antiviral role in balancing the response against pathogens and the risk of developing autoimmunity or excessive inflammation. The reduction of CD8+ T cells and B cells and the consequent increase in the CD4/CD8 ratio has been reported in numerous cases of critically ill patients with COVID-19. Furthermore, the reduction in lymphocyte counts correlates with the severity of SARS-CoV-2 disease and represents an important risk factor for a poor prognosis (48, 49).

Lymphocyte subpopulations were previously evaluated by Zhang et al. in a study that found a statistically significant difference in terms of prognosis and in-hospital length of stay in patients with mild, severe, and critical COVID-19 disease (50). While those data confirmed a clear depletion of lymphocyte subpopulations in patients suffering from severe vs. mild disease, the study population was more heterogeneous than that included in ours. It is precisely the homogeneity of the population we enrolled that we believe might explain the absence of statistical significance observed in our study, similar to the results reported by Pan et al. previously (51). Indeed, the pattern of lymphocyte subpopulations in patients with COVID-19 has been described in conflicting literature that, however, focused more on the severity of the presentation of the disease than on mortality. Additionally, the possible impact of corticosteroids on lymphocyte numbers and subpopulations deserves a note. However, the duration of our study covered a rather large period, during which the indications in the literature with respect to steroid therapy changed in terms of indication, duration, and dosage. Due to this heterogeneity, it was not possible to conduct a focused analysis differentiated by steroid type and dosages.

Immunoglobulins M (IgM) are known to represent the first line of defense during viral infections, prior to the generation of the high-affinity immunoglobulin G (IgG) adaptive immune response, which is important for long-term immunity and immunological memory. The specific antibody response against SARS-CoV-2 is related to the severity of the disease and the prognosis of patients with COVID-19. In the previous phases of the pandemic, the impaired immunological response was hypothesized as a potential target for the treatment of the COVID-19 disease, so much so that treatment with intravenous immunoglobulin IgG (IVIG) was proposed with the aim of mitigating the immunosuppression caused by the virus and guaranteeing broad-spectrum immune protection, not without adverse events (52). Treatment with IVIG has neither been standardized in COVID-19 nor has the assessment of the pre-treatment IgG level (53).

As previously said, however, studies in the literature are inconclusive. A German study conducted on 62 patients found that patients with lower IgG levels are characterized by more severe forms of the disease, an earlier need for ICU admission, a lower P/F ratio, a higher SOFA, a higher incidence of AKI, and lower lymphocyte levels. In these patients, the clinical course is characterized by a higher mortality rate (46.2 vs. 14.3%; p = 0.012), a longer ICU stay [28 (6-48) vs. 12 (3-18) days; p = 0.012], and hospital length of stay [30 (22-50) vs. 18 (9-24) days (p = 0.004)] (54). Another study conducted on 707 patients looked at both subpopulations and IgM and IgG levels in patients with COVID-19, finding low lymphocyte levels in more severe patients and lower IgM and IgG levels in the most severe cases (55). Nevertheless, no statistically significant correlation was found between the total number of Ly T, CD4+, and CD8+ cells and those of Ig. It was then observed that the total number of T cells, CD4+, and CD8+ gradually recovered in critically ill patients who had a favorable course while remaining low in those with moderate forms. The production of IgM and IgG was delayed in the critically ill group.

In this situation of uncertainty on the real impact of immunoglobulins and lymphocyte subpopulations in the critically ill ICU patients context, we do believe that our data, which show no statistically significant correlation between IgG, IgM, and IgA levels and mortality, need to be confirmed in larger studies with a less severe comparison population and no potentially confounding factors (in particular, comorbidities and superinfections).

Finally, the analysis of the vv-ECMO cohort (47 patients, 22.5% of the total) certainly deserves a comment (56–58). In this subgroup, no differences in MR-proADM values were evidenced between survivors and non-survivors probably due to the confounding factor represented by the intrinsic endotheliitis linked to ECMO support and the high mortality observed in this cohort of patients (82.9%) (59). However, it should be noted that, even in this subgroup, the MR-proADM values measured on days 3 and 7 were higher in non-survivors without reaching statistical significance (p = 0.08). Further studies with a larger sample size are needed to better understand the variables that influence MR-proADM trends in the subpopulation of critically ill patients with COVID-19 undergoing vv-ECMO.

This study has limitations: (1) it is a monocentric experience (although in different ICUs); (2) it refers to a highly complex university center, receiving critical patients as secondary hospitalization and vv-ECMO support; (3) the high mortality of patients and their turn-over in a pandemic period may have created a selection bias. Furthermore, due to the pandemic context, there is a lack of a comparison population, represented by patients admitted to ordinary or semi-intensive hospital wards. Finally, the possible impact of confounding factors, such as bacterial superinfections, which are widely represented in our population (60–62), and cardiovascular and renal dysfunction cannot be clearly defined.

In the near future, the use of biomarkers, especially MR-proADM, would be focused to obtain possible early indications about specific treatment (e.g., pharmacological treatment) and patients/resource allocation (ICU vs. ordinary ward), including a comparison population with different severity and different stage of organ dysfunction (e.g., ARDS, hemodynamic disorders, and superinfections).

MR-proADM can effectively predict the risk of death in severe COVID-19 patients. Higher MR-proADM values at ICU admission can identify patients with worse outcomes. In addition, measuring the temporal evolution of MR-proADM values with repeated monitoring could help in assessing clinical progression.

The values of IgA, IgM, and IgG, as well as lymphocyte subpopulations (except for natural killers), do not appear to be related to outcome, even if the peculiarities of the enrolled cohort may have influenced this analysis, which should be confirmed by larger studies that can better assess the role of possible confounding factors.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee: Comitato Etico Interaziendale A.O.U. Città della Salute e della Scienza di Torino—A.O. Ordine Mauriziano—A.S.L. Città di Torino; ethics approval number 0121515. The patients/participants provided their written informed consent to participate in this study.

GMo, GS, and LB: conceptualization. GMo, FR, GMe, and CF: methodology. GMo, EB, and CF: formal analysis. GMo, GS, EB, DL, AG, GC, GD'A, US, and CB: data curation. GMo and EB: writing—original draft preparation. GS, VF, US, CB, MZ, SC, and LB: writing—review and editing. LB and GMe: supervision. All authors have read and agreed to the published version of the manuscript.