Serum levels of soluble suppression of tumorigenicity 2 (sST2) and heart-type fatty acid binding protein (H-FABP) independently predict in-hospital mortality in geriatric patients with COVID-19

Elevation of cardiac damage biomarkers is associated with adverse clinical outcomes and increased mortality in COVID-19 patients. This study assessed the association of admission serum levels of sST2 and H-FABP with in-hospital mortality in 191 geriatric patients (median age 86 yrs., IQR 82 – 91 yrs.) with COVID-19 and available measures of hs-cTnT and NT-proBNP at admission. Cox proportional hazards models were utilized to predict in-hospital mortality, considering clinical/biochemical confounders as covariates. A composite cardiac score was calculated to improve predictive accuracy. Patients deceased during their hospital stay (26%) exhibited higher levels of all biomarkers, which demonstrated good discrimination for in-hospital mortality. Addition of sST2 and H-FABP significantly improved the discriminatory power of hs-cTnT and NT-proBNP. The composite cardiac score (AUC = 0.866) further enhanced the predictive accuracy. Crude and adjusted Cox regressions models revealed that both sST2 and H-FABP were independently associated with in-hospital mortality (HR for sST2 ≥ 129 ng/mL, 4.32 [1.48 – 12.59]; HR for H-FABP ≥ 18 ng/mL, 7.70 [2.12 – 28.01]). The composite cardiac score also independently correlated with in-hospital mortality (HR for 1-unit increase, 1.47 [1.14 – 1.90]). In older patients with COVID-19, sST2 and H-FABP demonstrated prognostic value, improving the predictive accuracy of the routinely assessed biomarkers hs-cTnT and NT-proBNP.


Introduction
The COVID-19 pandemic has presented an unprecedented global healthcare challenge, with millions of individuals affected and significant strain placed on healthcare systems worldwide (Haldane et al., 2021).While respiratory symptoms and complications have garnered significant attention, a substantial body of evidence suggests that cardiac involvement plays a crucial role in the morbidity and mortality associated with COVID-19 (Driggin et al., 2020).Geriatric patients with COVID-19 often present with complex clinical features and comorbidities, including cardiovascular diseases, which contribute to poorer outcomes (Bonafe et al., 2020).
Cardiac damage in COVID-19 is multifactorial and can manifest through various mechanisms.Direct viral invasion, immune-mediated injury, hypoxemia, hypercoagulability, and systemic inflammation are among the contributors to cardiac injury in this context (Tsai et al., 2023).Importantly, cardiac damage biomarkers, such as troponins (cTn), natriuretic peptides, and other non-routine biomarkers, such as GDF-15 (Rochette et al., 2021) and sST2 (Sanchez-Marteles et al., 2021), have emerged as valuable tools for assessing myocardial injury and predicting adverse outcomes in COVID-19 patients (Henry et al., 2020).
Routinely assessed cardiac troponins and natriuretic peptides (BNP and NT-proBNP) have shown a strong and independent association with in-hospital mortality in COVID-19 (Manocha et al., 2021;Menditto et al., 2023;Olivieri et al., 2022;Zaninotto et al., 2020).However, a significant decrease in their specificity for the diagnosis of acute myocardial injury and heart failure is observed in geriatric patients due to non-cardiac causes of elevation, such as age and comorbidities (Von Jeinsen et al., 2020).Noteworthy, in geriatric patients, cTn and natriuretic peptides demonstrated a prognostic role in different clinical scenarios, includingbut not limited to -CV disease (Gravning et al., 2014;Sabbatinelli et al., 2022a).
In addition to cTn and NT-proBNP, emerging evidence suggests the potential prognostic value of other cardiac damage biomarkers, including soluble suppression of tumorigenicity 2 (sST2) and heart-type fatty acid-binding protein (H-FABP), in COVID-19 patients (Omland et al., 2021;Yin et al., 2020).sST2, a member of the interleukin 1 receptor family mainly produced in extracardiac tissues in response to proinflammatory and profibrotic stimuli, acts as a circulating decoy for interleukin-33 (IL-33), thus preventing the cardioprotective effects of the ST2/IL-33 signaling and promoting maladaptive myocardial hypertrophy and cardiomyocyte apoptosis (Homsak and Gruson, 2020).H-FABP is a small cytosolic protein that functions as the principal transporter of long-chain fatty acids in cardiomyocytes and endothelial cells (Sarzani et al., 1988;Ye et al., 2018).H-FABP gained considerable attention as an early marker of myocardial infarction (MI), as it is rapidly released into the circulation in response to myocardial injury (Ye et al., 2018).Elevated levels of sST2 and H-FABP have been associated with adverse outcomes in various CV conditions (Bivona et al., 2018;Sabbatinelli et al., 2022a).
In this framework, the identification of reliable prognostic markers is essential to improve risk stratification and guide clinical decisionmaking, particularly in the in-hospital setting (Sabbatinelli et al., 2022b).To date, studies focusing on the synergic predictive value of these cardiac damage biomarkers specifically in geriatric patients with COVID-19 are limited.A comprehensive analysis that integrates the available evidence is necessary to establish their utility as predictors of in-hospital mortality in this vulnerable population.Here, we aimed to assess the association of serum high-sensitivity troponin T (hs-cTnT), NT-proBNP, sST2, and H-FABPassessed at admission -with in-hospital mortality in an Italian sample of COVID-19 geriatric patients.

Study design and participants
The present study was a subgroup analysis of a retrospective cohort involving geriatric patients hospitalized for COVID-19 at the Italian National Center on Aging (IRCCS INRCA) between March 2020 and June 2021.The main inclusion criteria were i) age ≥ 65 yrs., ii) confirmed COVID-19 diagnosis through a positive RT-PCR assay of nasopharyngeal swabs, iii) availability of hs-cTnT and NT-proBNP measures at admission.The only exclusion criterion was the lack of informed consent.
Ethics approval was obtained from the Ethics Committee of the IRCCS-INRCA, Ancona, Italy (reference number CE-INRCA-20008).The protocol has been registered under the ClinicalTrial.govdatabase (reference number NCT04348396).All procedures followed the tenets of the Declaration of Helsinki.

Biomarkers
Hs-cTnT and NT-proBNP were assessed on serum samples using the electrochemiluminescence Elecsys® Troponin T high sensitivity STAT and proBNP II assays, respectively, within six hours from hospital admission on a Roche Cobas 6000 automated analyzer (Roche Diagnostic, Meylan, France).Samples were stored at − 80 • C for subsequent analyses.
sST2 serum levels were assessed on an Atellica® CH 930 Analyzer (Siemens Healthineers, Germany) using the latex bead turbidimetric immunoassay Sequent-IA™ ST2 Assay (Critical Diagnostics, United States), currently available in Europe under the CE-IVD mark.The linearity range is 15-300 ng/mL, the intra-and inter-assay CVs are 6.5% and 8.6%, respectively, at a sST2 concentration of 26 ng/mL.At a sST2 concentration of 75 ng/mL, the CV values are 2.8% and 3.8% respectively.
Serum H-FABP was assessed on a Snibe Maglumi® 2000 analyzer (Snibe Co., Shenzhen, China) using a commercial sandwich chemiluminescence immunoassay provided by the manufacturer.The linearity range is 0.450 -360 ng/mL.The intra-and inter-assay CVs are 6.0% and 4.3%, respectively, at an H-FABP concentration of 6.57 ng/ mL, and 2.3% and 1.1%, respectively, at an H-FABP concentration of 103 ng/mL.

Outcomes and covariates
The primary outcome was in-hospital mortality, defined as death during the incident hospitalization.Information on the following comorbidities was collected: hypertension (HT), coronary artery disease (CAD), heart failure (HF), atrial fibrillation (AF), diabetes mellitus (DM), chronic obstructive pulmonary disease (COPD), stroke history, cancer history, chronic kidney disease (CKD).Information on medication use was obtained and recorded using Anatomical Therapeutic Chemical (ATC) codes.At the admission entry, vital signs including arterial blood pressure, heart rate, peripheral oxygen saturation, and body temperature have been recorded.The blood sample was analyzed for creatinine, C-reactive protein (CRP), pro-calcitonin (PCT), interleukin-6 (IL-6), ferritin, and D-dimer using automated methods.The definition of CKD was based on the estimated glomerular filtration rate (eGFR) < 60 mL/ min obtained by CKD-EPI formula.

Statistical analysis
Continuous variables were reported as either mean and standard deviation or median and interquartile range based on their distribution (assessed using the Shapiro-Wilk test).Comparison of variables between groups was performed by unpaired Student's t-test or Mann-Whitney U test as appropriate.Tertiles of laboratory parameters were calculated and, as other categorical variables, were expressed as absolute numbers and percentages and analyzed by Chi-square test.The association of cardiac biomarker concentrations with in-hospital mortality was explored by Kaplan-Meyer curves and assessed by the log-rank test of equality.The ability of the biomarkers to discriminate in-hospital mortality was compared using area under the receiver operating characteristic curve (AUC) and net reclassification improvement (NRI).The continuous, category-free, net reclassification improvement (NRI>0) was used to assess reclassification (Pencina et al., 2011).A bootstrap procedure with 1000 replications was used to estimate 95% Confidence Interval (CI) for AUCs and NRIs.Multivariable Cox proportional hazard models were computed to obtain adjusted estimates of the association between exposure variables and the primary outcome.A two-tailed P value < 0.05 was considered significant.Data were analyzed using STATA version15.1 Statistical Software Package for Windows (Stata Corp, College Station, TX).

Results
A total of 191 consecutive geriatric patients (median age 86 yrs., IQR 82-91 yrs.) who were hospitalized at the INRCA hospital (Ancona, Italy) due to COVID-19 were included in the analysis.26% (50/191) of the enrolled patients were deceased during the in-hospital stay.The mean number of days from hospital admission to discharge for the recovered patients was 18.8 ± 10.5, and that for the deceased patients was 16.2 ± 11.3.The minimum number of days for which patients remained hospitalized was 1 day, while the maximum number was 49 days for survived patients and 51 days for deceased patients.
With respect for CV medications, deceased patients had a higher prevalence of diuretics (p = 0.006) and beta-blocker (p = 0.039) consumption and a lower prevalence of drugs acting on the reninangiotensin system (p = 0.018).No significant differences between deceased and survivor patients were observed for lipid-lowering agents, platelet inhibitors and oral anticoagulants.Regarding comorbidities, AF (46% vs. 27%, p = 0.013) and CKD (46% vs. 24%, p = 0.004) were more prevalent among deceased patients.On the contrary, no significant differences were observed for other comorbidities, including diabetes, HF, and history of myocardial infarction.
Biochemical and clinical characteristics of study participants are reported in Table 1.Significant differences were observed for a number of biochemical variables, i.e. creatinine, CRP, procalcitonin, IL-6, ferritin, and D-dimer, which were discussed in our previous reports on the whole cohort (Olivieri et al., 2022).
Receiver Operating Characteristic (AuROC) curves were computed to assess of cardiac biomarkers to predict in-hospital mortality.Fig. 2A and B shows the ROC curves for variables categorized into tertiles or considered as continuous respectively.All the biomarkers showed an acceptable discrimination (AUC >0.70), with the greatest performance reached by sST2 and H-FABP considered as continuous (AUC >0.80).
Then, we tested the reliability of the predictions using a bootstrap procedure with 1000 replications, to obtain 95% CI for the AUCs of the categorized biomarker, compare ROC curves, and assess the additive value of the biomarkers in terms of net reclassification improvement (NRI).The results, reported in Table 2, show that the progressive inclusion of each biomarker significantly and progressively improves the AUC of the mortality classifications, with an AUC of 0.848 (0.788-0.909) when considering the four biomarkers, which is comparable with that of the cardiac score.The additive value of the biomarkers was confirmed also by the significant NRIs compared to hs-cTnT.
We applied Kaplan-Meier survival analysis to estimate survival time distributions among tertiles of each biomarker and the score obtained as abovementioned.The KM survival curves, reported in Fig. 2, showed that each categorized predictor achieved a significant stratification of patients according to the risk of in-hospital mortality (log-rank test, p < 0.001).Specifically, a significant step up in mortality was observed for patients with serum levels of sST2 (Fig. 2A) and H-FABP (Fig. 2B) in the highest tertiles.. Univariable and multivariable Cox regression models were computed to assess the ability of each categorized CV biomarker and for the cardiac score as a continuous variable to predict COVID-19 in-hospital mortality.Models were adjusted for age, gender, comorbidities relevant to CV health, and biochemical predictors that were different among groups.At the univariable analysis, all the biomarkers were significantly associated with in-hospital mortality, with subjects in the intermediate and highest strata of hs-cTnT and NT-proBNP having a higher mortality risk compared to subjects with the lowest concentrations.On the other hand, only subjects in the highest tertiles of sST2 and H-FABP had a higher mortality risk compared to subjects in the lowest tertiles.All the biomarkers retained their predictive value after adjustment for age, gender, and relevant comorbidities, while only sST2 and H-FABP were independently associated with in-hospital mortality after  when considering also the laboratory parameters as confounders.Specifically, the HR with 95% CIs for patients with sST2 and H-FABP in the highest tertiles were 4.32 (1.48 -12.59) and 7.70 (2.12 -28.01) for H-FABP, respectively.Interestingly, the cardiac score independently predicted in-hospital mortality after adjustment for all the clinical and laboratory confounders, with a 1.47-fold (95% CI 1.14-1.90)increased risk for each unit of increase..

Discussion
COVID-19 is characterized by high rates of hospitalization and mortality, especially in the oldest patients.Following respiratory complications, cardiac complications are the most prevalent finding.Intensive efforts were undertaken to predict the most severe COVID-19 outcomes, ideally at the point of hospital admission.
In this study, we evaluated the prognostic value of two biomarkers of cardiac damage, assessed using fully automated immunoassays, in a sample of geriatric patients with COVID-19.To evaluate their additive value over the routinely assessed cardiac markers, we included in the study only those patients that were tested for hs-cTnT and NT-proBNP at admission.
We showed that the serum levels of the assessed biomarkers were consistently higher in patients that were deceased during the in-hospital stay.Notably, a great proportion of subjects had serum levels of the investigated biomarkers that were above the decision limits applied in the clinical practice.This finding was quite expected for hs-cTnT and NT-proBNP, for which aging determines a consistent elevation in their levels that is further exacerbated by comorbidities with CV relevance (Franzini et al., 2015;Rorth et al., 2020).Notably, about 75% of subjects had sST2 levels above the 35 ng/mL cutoff, which is associated with a poorer prognosis in patients with HF (Aimo et al., 2019).Similarly, almost 95% of patients had H-FABP levels that were above the 99th percentile identified by the manufacturer in apparently healthy individuals, i.e., 6 ng/mL.Overall, these results confirm the involvement of the CV system in COVID-19, to an extent that is significantly more consistent than that observed in patients of similar age with chronic CV comorbidities (Sabbatinelli et al., 2022a).Our findings remark on the need for the identification of specific reference intervals and decision limits for all the acute disease states with a proven impact on the CV system, as prompted by previous reports in the context of COVID-19 (Qin et al., 2020).
To our knowledge, this is the first study assessing the combination of biomarkers capable of describing different aspects of cardiac damage, including direct myocardial damage (hs-cTnT and H-FABP) and heart failure, intended both as the response of cardiac myocytes (NT-proBNP) and the negative influence of extracardiac tissues on myocardial adaptation (sST2).In our previous reports on the same cohort of geriatric patients, we focused on the role of routinely assessed biomarkers and showed that both NT-proBNP and hs-cTnT were associated with inhospital mortality (Menditto et al., 2023;Olivieri et al., 2022).Here, we demonstrated that the progressive addition of NT-proBNP, sST2, and H-FABP to hs-cTnT significantly improves the ability to discriminate between survivors and deceased patients, increasing the AUC from 0.673 to 0.848.When considering the survival analysis, sST2 and H-FABP showed a strong ability to predict in-hospital mortality, that persisted after adjustment for the existing comorbidities and biochemical variables reflecting renal function and systemic inflammation.A recent report showed a correlation between H-FABP and COVID-19 clinical and radiological severity (Mohamed et al., 2023).The authors proposed an H-FABP threshold of 19.5 ng/mL for the identification of high-risk patients, which, although derived from a different immunoassay, is comparable to the 18 ng/mL cutoff that is associated with a 7.7-fold increased risk of death in our cohort.In a smaller case-control study, admission levels of H-FABP > 7 ng/mL were associated with  more severe presentations of COVID-19, although mortality and other outcomes were not investigated (Yin et al., 2020).Despite the conflicting reports on the diagnostic role of H-FABP in acute HF and myocardial infarction that discouraged extensive investigation over the last years (Haltern et al., 2010;Hoffmann et al., 2015;Liou et al., 2015), the inclusion of H-FABP could improve the risk assessment in patients with COVID-19, especially in the oldest patients where the interpretation of cTn and natriuretic peptides is significantly affected by age and comorbidities.In this regard, data on the prognostic role of H-FABP is limited and confined to patients with established CV disease (Bajaj et al., 2015;Zhang et al., 2020).
In our study, we showed that patients with the highest sST2 levels (≥129 ng/mL) had a higher risk of mortality.Serum levels of sST2 above 70 ng/mL were previously related to higher risk of the composite outcome of intensive care unit admission or death in COVID-19 (Sanchez-Marteles et al., 2021).The authors reported that sST2 levels peaked at 48-72 h after admission and declined at discharge, suggesting that this marker could track the disease course and the onset of complications also during the hospital stay.Another report evaluating a large number of immune response biomarkers showed that sST2 was able to stratify COVID-19 patients according to disease severity and survival status (Abers et al., 2021).Notably, its levels were directly related to other markers of systemic inflammation and endothelial dysfunction, but not its ligand IL-33, confirming that the detrimental effects of sST2 on cardiac adaptation are not compensated by an increase in IL-33 in these patients.From a pathophysiological perspective, the sustained release of this mediator by a great variety of cell types, including endothelial and myeloid cells (Mildner et al., 2010), could be framed in the context of extreme systemic inflammation in COVID-19, and its ability to interfere with positive adaptive myocardial responses, may explain its association with adverse respiratory outcomes, as previously reported by other groups (Omland et al., 2021).On the other hand, NT-proBNP and hs-cTnT, despite being significantly higher in deceased patients, did not achieve statistical significance in the fully adjusted survival model, probably due to the limited sample size of the present study compared to our previous reports (Menditto et al., 2023;Olivieri et al., 2022).However, their prognostic role was confirmed by the independent ability of the score derived by summing up the points corresponding to each cardiac biomarker strata to predict in-hospital mortality.Indeed, we showed that each 1-point increase in the score, which corresponds to an increase of the level of either biomarker to the higher tertile is associated with a 50% increased risk of mortality.The additive value of hs-cTnT, NT-proBNP, and sST2, which was previously remarked in heart failure (Emdin et al., 2018) and type 2 diabetes (Sabbatinelli et al., 2022a), may reflect the pathophysiology underlying each biomarker.Notably, a previous study in a cohort of adults hospitalized due to COVID-19 showed that patients with a high inflammatory burden, assessed in terms of IL-6, had the highest levels of hs-cTnT and NT-proBNP (Ruscica et al., 2021), supporting the contribution of several mechanisms, including inflammation, endothelial dysfunction, and prothrombotic alterations, to the myocardial injury observed in COVID-19 (Connors and Ridker, 2022).
The present study has limitations that need to be addressed.First, only the admission levels of the cardiac biomarkers were considered for the present study.Therefore, we were unable to verify whether shortterm variations of the biomarkers could be associated with poorer outcomes.Second, we were unable to discriminate between the contribution of COVID-19 and preexisting comorbidities in determining the high concentrations of the biomarkers.However, the presence of comorbidities associated with increased CV risk was accounted for in the multivariable survival models.Third, due to the high proportion of missing data, we did not include the lipid profile in our analysis.Previous investigations showed that the functionality of HDL was reduced in patients with COVID-19 (Papotti et al., 2021), suggesting that part of the observed CV damage could be mediated by reduced anti-inflammatory properties of specific lipoprotein subsets.In this regard, we did not observe any significant difference between survived and deceased patients in terms of prevalence of secondary prevention drugs, including statins.
In conclusion, the combined assessment of multiple biomarkers of cardiac damage, in the absence of disease-specific cutoff values, may provide useful information on the extent of COVID-19-associated damage and a more informative stratification of patients according to their mortality risk, allowing to tailor treatment strategies aimed at reducing the risk of COVID-19-related acute CV events.Our data emphasize the significance of evaluating cardiac damage markers to predict clinical outcomes not only in COVID-19 but also in other acute conditions that necessitate hospitalization and are known to affect the cardiovascular system.This approach enables healthcare professionals to establish precise and objective criteria for identifying geriatric patients with the highest risk of in-hospital mortality.

Ethical approval
This study was approved by the Ethics Committee of the IRCCS-INRCA, Ancona, Italy (CE-INRCA-20008).

Informed consent
Informed consent was obtained from all individuals included in this study, or their legal guardians or wards.

Fig.
Fig. Area under the Receiver Operating Characteristic (AuROC) curve comparison for the ability of (A) continuous or (B) categorical hs-cnT, NT-proBNP, sST2, H-FABP, and the derived cardiac score to discriminate between deceased and survivor COVID-19 patients.

Table 1
Demographic, clinical, and biochemical characteristics of study subjects.
(continued on next page) J.Sabbatinelli et al.

Table 2
Summary of AUCs with 95% confidence intervals and Net Reclassification Improvement (NRI) estimated by bootstrap procedure with 1000 replications.Categorized biomarkers were used for the analysis.