Laboratory markers associated with COVID‐19 progression in patients with or without comorbidity: A retrospective study

Abstract Objectives To investigate laboratory markers for COVID‐19 progression in patients with different medical conditions. Methods We performed a multicenter retrospective study of 836 cases in Hubei. To avoid the collinearity among the indicators, principal component analysis (PCA) followed by partial least squares discriminant analysis (PLS‐DA) was performed to obtain an overview of laboratory assessments. Multivariable logistic regression analysis and multivariable Cox proportional hazards regression analysis were respectively used to explore risk factors associated with disease severity and mortality. Survival analysis was performed in patients with the most common comorbidities. Results Lactate dehydrogenase (LDH) and prealbumin were associated with disease severity in patients with or without comorbidities, indicated by both PCA/PLS‐DA and multivariable logistic regression analysis. The mortality risk was associated with age, LDH, C‐reactive protein (CRP), D‐dimer, and lymphopenia in patients with comorbidities. CRP was a risk factor associated with short‐term mortality in patients with hypertension, but not liver diseases; additionally, D‐dimer was a risk factor for death in patients with liver diseases. Conclusions Lactate dehydrogenase was a reliable predictor associated with COVID‐19 severity and mortality in patients with different medical conditions. Laboratory biomarkers for mortality risk were not identical in patients with comorbidities, suggesting multiple pathophysiological mechanisms following COVID‐19 infection.


| INTRODUC TI ON
The COVID-19 pandemic has spread rapidly across the world. By July 13, more than 13 million people were infected with COVID-19 across the world, and over half a million people have died, according to the data provided by Johns Hopkins University. The risk classification based on the laboratory markers can be helpful to provide early intervention and reduce mortality of COVID-19 patients.
However, the reported laboratory biomarkers associated with disease progression were inconsistent. [1][2][3][4] The type and distribution of comorbidities in the studied cohorts varied, 5 which may affect the biomarker analysis as the comorbidities are the leading cause for the conversion to severe and critical cases in COVID-19 patients. The features of laboratory indicators in specific subgroups remain ambiguous, although clinical characteristics of COVID-19 patients have been documented. [6][7][8] We performed a retrospective and multisite study of COVID-19 cases from designated tertiary hospitals in Hubei province to investigate the association between laboratory indicators and disease progression (severity and mortality) in patients with different medical conditions. In order to avoid the collinearity among the indicators, principal component analysis (PCA) followed by partial least squares discriminant analysis (PLS-DA) was respectively performed in patients with or without comorbidity. Multivariate logistic regression model was respectively used to explore the risk factors associated with disease severity using the variables extracted from the PCA/ PLS-DA analysis. The association between laboratory makers at admission and short-term survival rate was analyzed in patients with the most common comorbidities including hypertension, diabetes, and liver diseases. The data may yield a novel insight for the progression predictors of COVID-19, informing the therapy management and the pathophysiological mechanism involved in COVID-19 infection.

| Study design
This was a retrospective study that included patients (age 18

| Data collection
We collected the information including demographic, epidemiological, clinical, laboratory, treatment, and outcome data. A standardized case report form was used to extract medical records. The epidemiological characteristics (including recent exposure history), clinical symptoms and signs, and laboratory findings at the time point for the first hospital admission were extracted from electronic medical records. Laboratory assessments consisted of complete blood count, blood chemistry, coagulation test, liver and renal function, electrolytes, C-reactive protein (CRP), lactate dehydrogenase (LDH), creatine kinase, and chest computed tomographic (CT).

| Study definition and primary outcome
The categories of COVID-19 cases were defined according to the diagnosis and treatment protocol for novel coronavirus pneu-

monia (Trial Version 5) by National Health Commission & State
Administration of Traditional Chinese Medicine. 10 Mild case refers to the clinical symptoms were mild, and there was no sign of pneumonia on imaging. Moderate case refers to showing fever and respiratory symptoms with radiological findings of pneumonia. Severe case meets any of the following criteria: (a) respiratory distress (≥30 breaths/min); (b) oxygen saturation ≤ 93% at rest; (c) arterial partial pressure of oxygen to fraction of inspired oxygen ratio < 300.
Critical case meets any of the following criteria: (a) respiratory failure and requiring mechanical ventilation; (b) shock; (c) with other organ failure that requires ICU care.
To assess risk factors associated with disease severity of COVID-19, the cases were classified according to the diagnostic criteria (mild/moderate cases versus severe/critical cases). For the evaluation of risk factors associated with disease mortality, patients were followed up for a maximum of 30 days after the diagnosis of COVID-19. Survival time was defined as the time from hospitalization until the occurrence of death.

| Statistical analysis
Continuous variables were expressed as medians and interquartile ranges (IQR). Categorical variables were summarized as counts and percentages. One-way ANOVA followed by the least significant difference (LSD) post hoc test was used to test the differences among the groups for continuous variables when the data were normally distributed; otherwise, the Mann-Whitney test was used. For categorical variables, χ2 test was used to test the differences among the groups.  hazard ratio (HR) along with the 95% confidence interval (CI) were respectively reported. Survival in patients with comorbidities was estimated by the Kaplan-Meier method, and any differences in 30-day survival rate were evaluated with a stratified log-rank test. A twosided α value less than 0.05 was considered statistically significant.

| RE SULTS
Laboratory markers at admission associated with disease sever-  Compared with mild/moderate cases, the levels of LDH were 1.5fold (P < .001) higher in severe/critical cases; additionally, severe/ critical cases had lower levels of prealbumin (P < .001) ( Figure 1C,D).
Among the cases, 3 patients had a progressive disease during the hospitalization. They were classified as a moderate case at admission and progressed to severe disease. After the treatment, they were converted from severe to moderate. Dynamic curves of LDH and prealbumin in these inpatients indicated that the increase or decrease in LDH and prealbumin levels preceded the deterioration or alleviation of respiratory indexes ( Figure 1E). Although LDH levels in these patients were within the reference range at admission, the levels were increased before they progressed from moderate to severe cases.
Since the older age has been reported as a risk factor of COVID-19 severity and mortality, [11][12][13] we chose age and the top 5 variables extracted from PCA/PLS-DA analysis to perform multivariable logistic regression analysis ( The data were plotted individually in Figure 2. The observations out of the 0.95 Hotelling ellipse (1 cases) were distinguished and excluded from the final model (Figure 2A). The PLS-DA model to distinguish the severe/critical from mild/moderate cases was fitted generating cumulative R2Y (0.25) and Q2 (0.24). In terms of contribution to the separation of severe/critical from mild/moderate cases, the loadings for PC1 illustrated that the significance rank was CRP > D-dimer > LDH > prealbumin > lymphopenia ( Figure 2B).
CRP was the largest positive loading responsible for the separation of observations, while prealbumin was the largest negative loading responsible for the separation.
Compared with mild/moderate cases, the levels of CRP, D-dimer, and LDH were significantly higher in severe/critical cases; additionally, severe/critical cases had reduced prealbumin and lymphocytes levels ( Figure 2C-g).
We chose age and the top 5 variables extracted from PCA/PLS-DA analysis including CRP, D-dimer, LDH, prealbumin, lymphopenia for multivariable logistic regression model. The data showed that the increased odds of disease severity were associated with age, elevated LDH, and reduced prealbumin in patients with comorbidities; however, CRP, D-dimer, and lymphocytes were not associated with disease severity (Table 3).
Laboratory markers at admission associated with mortality risk in COVID-19 patients with comorbidities.
The proportion of death was 8.98% (n = 57) in patients with comorbidities, while only 5 of deaths were identified in patients with no comorbidity. Thus, we performed multivariable Cox proportional hazards regression analysis to evaluate the association of laboratory makers at admission with short-term mortality risk in patients with comorbidities.
The data from Cox proportional hazards regression analysis revealed that the mortality risk was associated with age, LDH, CRP, D-dimer, and lymphopenia in cases with comorbidities (Table 4). No significant correlation was observed between prealbumin levels and mortality risk. however, no significant change in survival rate was observed in patients with elevated CRP or reduced lymphocytes.

| D ISCUSS I ON
This study demonstrated the laboratory markers at admission associated with disease severity and mortality of COVID-19 patients. , lymphocytes (C), and D-dimer (D). A, In the total cases with comorbidities or in the subgroup of hypertension, patients with high C-reactive protein (CRP) levels were at higher mortality risk than those patients with normal levels of CRP. B, In the subgroups of hypertension, diabetes, and liver diseases, patients with high lactate dehydrogenase (LDH) levels were at higher mortality risk than those patients with normal levels of LDH. C, In the subgroups of hypertension and diabetes, patients with reduced lymphocytes were at higher mortality risk than those patients with normal levels of lymphocytes. D, In the subgroups of hypertension and liver diseases, patients with high D-dimer levels were at higher mortality risk than those patients with normal levels of D-dimer The predictors for the mortality of COVID-19 were different among the patients with or without comorbidities. We showed that LDH was a reliable predictor associated with the severity and mortality of COVID-19 patients with different medical conditions. The inflammation marker CRP was a risk factor associated with short-term mortality in patients with hypertension, but not liver diseases. The coagulation system marker D-dimer was an independent risk factor for death in patients with liver diseases. These results suggest that there are multiple pathophysiological mechanisms following COVID- 19 infection, and that personalized treatment (eg, anti-inflammatory and thromboembolic prophylaxis) based on individual medical condition is required to reduce the mortality risk.
The PCA/PLS-DA models generated from patients with or without comorbidities indicated that both LDH and prealbumin were im-  20,[25][26][27][28] The data that elevated D-dimer levels were accompanied with a parallel rise in markers of inflammation (eg, CRP) 29 supported the hypothesis that a severe inflammatory response subsequently activated coagulation and thrombin generation. Notably, the survival analysis based on our dataset showed that the reduced survival rate in patients with liver diseases was associated with the increased D-dimer, but not the elevated CRP. The thrombotic microangiopathy and microcirculatory impairment due to the endothelial cell injury induced by COVID-19 could be aggravated by liver dysfunction as most coagulation factors, anticoagulants, and fibrinolytic proteins are synthesized in the liver. The coagulation disorders could be an important mechanism in COVID-19 patients with liver disease rather than cytokine storm.
This study has a number of limitations. For the evaluation of disease severity, we did not have information of qPCR Ct value to assess the viral loads. The relationship between the indicators and viral loads needs to be elucidated in future studies to gain a better understanding.
In conclusion, LDH and prealbumin were relatively reliable markers for the assessment of disease severity in adult patients regardless of the comorbidities. CRP was an independent risk factor associated with short-term mortality in patients with hypertension, but not liver diseases. The laboratory biomarkers for disease prognosis of COVID-19 in patients with different comorbidities need to be further refined. Although the laboratory indicators are non-specific, the features of the biomarkers could be helpful to understand the pathological mechanism of COVID-19 infection and for therapy management.

ACK N OWLED G EM ENTS
We