In a recent study posted to the Research Square* preprint server, researchers assessed the role of renalase (RNLS) as an independent predictor of coronavirus disease 2019 (COVID-19)-related mortality.
Study: Renalase Identified by Machine Learning Methods As A Novel Independent Predictor Of Mortality In Hospitalized Patients With COVID-19. Image Credit: crystal light/Shutterstock
*Important notice: Research Square publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.
Background
In the past few months, a number of biomarker studies have shed light on specific elements of the immune response against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and prompted vital new questions. Individuals with fatal symptoms have been reported to exhibit pathophysiology distinguished from those having milder symptoms by an aggressively disturbed inflammatory response. Several changes in indicators, including renal damage, appear to be temporary.
Little is known about the depleted markers in SARS-CoV-2 infection or the dynamic behavior of these biomarkers. Renalase, a peptide produced endogenously by the heart, kidney, and endothelium, has prosurvival qualities, such as inhibiting cytokine release in viral infections, such as COVID-19, in mice models.
About the study
In the present study, researchers determined whether renalase independently predicted COVID-19-related mortality.
The team investigated COVID-19-positive adult patients hospitalized in an urban academic center from 1 March and 30 June 2020. Reverse transcription-polymerase chain reaction (RT-PCR) of nasopharyngeal swab samples from all patients revealed severe SARS-CoV-2 infection. The collection of all specimens and imaging was part of standard medical care.
The team utilized the Department of Medicine COVID Explorer (DOM-CovX), a group of hospitalized COVID-19 patients, to obtain clinical information from the electronic medical record system, including sociodemographics, vital signs, comorbidities, laboratory measurements, disposition, and procedures throughout the entire hospital stay.
The admission date, symptoms with onset dates, smoking history, immunocompromised state, cardiopulmonary resuscitation, as well as dates of death, intubation, and last completed follow-up, were extracted using manual chart reviews. The cohort's serum or plasma specimens were analyzed as follows by assessing inflammatory markers, including interferon (IFN)-ɑ, IFN-β, IFN-λ, IFN-Ɣ, interleukin (IL)-1β, IL-6, and tumor necrosis factor (TNF), along with kidney injury molecule (KIM-1).
Mortality was defined as death occurring within 180 days of the index visit. Traditional and contemporary machine learning techniques were utilized. For conventional models, the team employed logistic regression, and for more recent approaches, XGBoost.
Results
From March 2020 to June 2020, a total of 3,450 COVID-19 patients were hospitalized. The study group comprised 473 patients who volunteered for research, were older than 18 years, and submitted sufficient blood samples. Compared to hospitalized patients who were not included, the age and gender distribution of the cohort population were comparable. The whole cohort included 366 surviving patients and 71 fatalities.
The average length of stay for patients who passed away was 17 days, and they were older and more frequently men. In addition, they reported more comorbidities, higher β-natriuretic peptide (BNP), creatinine, troponins, ferritin, d-dimer, and procalcitonin levels, and lower platelets than individuals who survived.
Those who died showed lower average renalase levels and a trend toward higher levels of IL-1, IFNs, and KIM-1 than individuals who lived. The team identified age, patient gender, and mean renalase as significant predictors of mortality using the standard logistic regression model. In addition to renalase, the team identified clinical factors and various conventional laboratory markers as predictors of death.
According to the XGBoost model, a high BNP marker of heart strain was the most significant predictor of mortality. Patients with the lowest RNLS and the highest BNP quartiles showed significantly greater mortality rates than those with the highest RNLS and the lowest BNP quartiles. At the conclusion of their stay, individuals with high BNP and low renalase reported a higher mortality rate than those with low BNP and high renalase.
Renalase exhibited a comparable association with high-sensitivity troponin, a biomarker of cardiac damage, as seen by the correlation between the two. In conventional models, troponins were observed to predict death, despite their lack of significance in the XGBoost model.
When renalase levels were compared with inflammatory markers like IFN, IL-1, and IL-6, the association was less significant. Mortality was significantly different between individuals with low RNLS and high IL6 and those with high RNLS and low IL6. However, comparing similar groups for IFN and IL-1 revealed no significant differences.
In the XGBoost model, the platelet count was also identified as a significant predictor of death. Patients with low renalase levels and low platelet counts revealed the highest mortality rates compared to those with high renalase levels and high platelet counts.
Conclusion
The study findings showed that machine learning techniques could efficiently supplement traditional statistical methods when identifying COVID-19 mortality predictors. Renalase is a reliable and independent predictor of death in hospitalized COVID-19 patients.
Prospective studies should investigate the progression of renalase, particularly in combination with other markers of endothelial and cardiac dysfunction, as well as the possible therapeutic use of exogenous renalase.
*Important notice: Research Square publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.
Did California's pediatric COVID-19 vaccination program reduce reported cases and hospitalizations?