Serum glucose, lactate dehydrogenase and hypertension are mediators of the effect of body mass index on severity of COVID‐19

Abstract Background COVID‐19 has a broad clinical spectrum. We investigated the role of serum markers measured on admission on severity as assessed at discharge and investigated those which relate to the effect of BMI on severity. Methods Clinical and laboratory data from 610 COVID‐19 cases hospitalized in the province of Zheijang, China were investigated as risk factors for severe COVID‐19 (assessed by respiratory distress) compared to mild or common forms using logistic regression methods. Biochemical markers were correlated with severity using spearman correlations, and a ROC analysis was used to determine the individual contribution of each of the biochemical markers on severity. We carried out formal mediation analyses to investigate the extent of the effect of body mass index (BMI) on COVID‐19 severity mediated by hypertension, glycemia, Lactose Dehydrogenase (LDH) at the time of hospitalization and C‐Reactive Protein levels (CRP), in units of standard deviations. Results The individual markers measured on admission contributing most strongly to prediction of COVID‐19 severity as assessed at discharge were LDH, CRP and glucose. The proportion of the effect of BMI on severity of COVID‐19 mediated by CRP, glycemia or hypertension, we find that glucose mediated 79% (p < .0001), LDH mediated 78% (p < .0001), hypertension mediated 66% (p < .0001); however, only 44% (p < .005) was mediated by systemic inflammation (CRP). Conclusion Our data indicate that a larger proportion of the effect of BMI on severity of COVID‐19 is mediated by glycemia and LDH levels whereas less than half of it is mediated by systemic inflammation.


| INTRODUC TI ON
The clinical spectrum of SARS-CoV-2 infection appears to be wide, encompassing asymptomatic infection, mild upper respiratory tract illness with cough, fever and fatigue to more severe viral pneumonia with a risk of consequent respiratory failure and death. Risk factors associated with symptomatic disease requiring hospitalization are uncertain and factors that influence the risk of developing severe disease remain incompletely understood. Age has been consistently associated with increased mortality 1 ; type 2 diabetes and hypertension, two most prevalent comorbidities in hospitalized patients have been also associated with severe manifestations. [2][3][4] Recent data coming from the US and Western Europe show that being overweight or obese is a significant risk factor associated with hospitalization, 5 requirement to mechanical ventilation 6,7 and in-hospital mortality from COVID-19. 1 A number of mechanisms may underlie association of body mass index (BMI) and pneumonia including increased airway resistance and reduced respiratory muscle strength, lung volumes, and gas exchange. 8 However, it has emerged that patients with high BMI are at greater risk of death from COVID-19 than patients with asthma. 1 This suggests that the mechanisms by which SARS-Cov2 develops into a severe form of COVID-19 may be directly related to metabolic pathways that are specifically dysregulated in obese or overweight individuals. Obesity is also a key risk factor for type 2 diabetes, hypertension and cardiovascular disease all of which have been associated with COVID-19. Furthermore, metabolic impairment characterized by insulin resistance, hyperglycaemia and dyslipidaemia may remain unrecognized in substantial proportion of the population. 9 Evaluating of these parameters is important to risk stratify patients with COVID-19. 10 The objective of this study was to identify risk factors for symptomatic COVID-19 and the development of its severe/critical form using a multicentre retrospective cohort of 610 patients with COVID-19 from Zhejiang province, China. In particular, we have investigated to what extent the effects of high BMI (BMI of ≥ 23 kg/m 2 ). 11 of COVID-19 severity in the current cohort are mediated by systemic inflammation and how much by LDH, glycemia or hypertension.

| PATIENTS AND ME THODS
The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki. The local ethics committees of all hospitals approved the retrospective study of cohorts COVID-19. The requirement for written consent was waived due to the retrospective and anonymous nature of this study.

| Clinical categories
The severity of the disease was classified according to the

Guidelines for the Diagnosis and Treatment of New Coronavirus
Pneumonia (fifth edition) 12 and categorized into 4 groups as described previously. 13 In brief, patients were classified into the fol-

| Analyses of severity
Individual demographic and clinical features were tested for association with the 4-point scale of severity described above using a Spearman's correlation. Those variables that were significantly associated with the categorical measure of severity were then included in a LASSO logistic regression to select specific features to test as predictors of severe outcome (grade 3 or 4 of the categorical scale) vs less severe outcome (grade 1 or 2).

| Laboratory methods
Clinical laboratory test results, including SARS-CoV-2 RNA de-

| Statistical methods
Categorical variables were expressed as frequency and percentages. Continuous variables were expressed as mean and standard deviations. Logistic regressions were carried out either unadjusted or adjusting for covariates as indicated in the table legends. p < .05 was considered statistically significant. To formally assess the contribution of CRP, LDH, glycemia and hypertension to the association between BMI and disease severity, we conducted mediation analysis using the 'mediation' package in R version 3.6.1 where we considered the various clinical traits as mediator for the causal effect of BMI on severity.
Analyses were carried out using R version 3.6.1. p < .05 was considered statistically significant.

| RE SULTS
The selection of the study population was carried based on a previously published protocol. 13 We proceeded to investigate the biochemical and clinical factors that correlate with the disease severity. The demographic, clinical presentation features and biochemical markers in each of the four severity categories are shown in Table 1. We found that BMI greater than 23 kg/m 2 was strongly associated with a more severe form of the disease, but BMI below 23 kg/m 2 was not. We also found significant associations with male sex and older age. In terms of the biochemical measures, we found glucose and LDH having the most dramatic effect on severity compared to others. The presence of hypertension was significantly associated with severity, but type 2 diabetes was not.
We further ranked the biochemical markers correlated with severity and put them side by side with some of the demographic characteristics in terms of how much they can explain severity on a ROC analysis including each variable independently ( Figure 2).
We found that levels of LDH, CRP and glucose on admission were each individually able to predict 75% or more of COVID-19 severity.
Furthermore, we found that LDH was a greater predictor of severity we also explored the mediatory effect of hypertension, given its strong links to metabolic syndrome and found that it mediated 66% (p < .001) of the effect of BMI on severity as shown in Table 2. We also carried a regression analysis of severity with glucose and BMI and found significant associations for glucose (beta = 0.25, 0.89-1.53 95% CI, p=<0.001) and for BMI (beta = 0.14, 0.33-1.25 95%CI, p = .001). Furthermore, a regression of glucose on BMI was also found to be significant (beta = 0.14, 0.77-0.27 95% CI, p = .001).
TA B L E 1 Association between demographics, biochemical measures, comorbidities and severity. odds ratios were derived by logistic regression severe (severe + critical) vs not severe (mild + common Finally, although diabetes was not a strong risk factor for severity, it could be that the high levels of glucose seen to be predictive of severity may be only seen in diabetic patients. Therefore, we further investigated its effect in patients with and without type 2 diabetes and in the overweight/obese (BMI > 23 kg/m 2 ) and normal weight (BMI < 23 kg/m 2 ) categories. We found that glucose was  Note: Models were built in to assess the effect of association between overweight/obese (BMI > 23) and severity mediated by clinical traits. The effect observed is described in the 'effect' column where: Null, corresponds to no association observed between high BMI ad severity mediated by the traits of interest. Partial, the association between the high BMI and severity is partially mediated by the clinical trait of interest. Direct, the association between high BMI and severity is fully mediated by the clinical trait of interest.

TA B L E 2 Summary statistics of mediation analysis for the various clinical traits
significantly associated with severity in patients with and without type 2 diabetes (Figure 3). Similarly, the association of glucose with severity was significant in the overweight/obese category compared to the normal weight category. Furthermore, we found that CRP, a common marker of systemic inflammation, was significantly associated with severity in the presence or absence of type 2 diabetes and amongst normal and overweight categories as shown in Figure 3. A similar pattern was seen with regards to LDH levels. We have carried out formal mediation analyses to investigate the extent to which the effect of body mass index (BMI) on COVID-19 severity is mediated by glycaemia, LDH, CRP and hypertension. Both serum glucose and CRP have been functionally linked to higher BMI in the general population. [14][15][16] Obesity-induced meta-inflammation is associated with systemic increases in circulating inflammatory cytokines and acute phase proteins such as CRP and recruitment of leucocytes to inflamed tissues. 17 Obesity has also been shown to impair immune response to influenza virus. 18 When we investigated the proportion of the effect of BMI mediated by these two factors, we found that a higher proportion of the effect of BMI on COVID-19 severity is mediated by glycaemia than by CRP. Our mediation analysis also finds that the effect of BMI is also mediated by hypertension more so than by CRP (Figure 3). Previous meta-regression analysis has shown in African American and white individuals under the age of 50 with a median BMI of 29 found that hypertension and diabetes did not affect the relationship between obesity and severity. 19 However, in our study population, half of the patients were 50 or  and is unlikely to bias our results as there were not any sites with significantly larger numbers of severe cases.

| D ISCUSS I ON
In conclusion, we find that baseline patient-related and laboratory parameters in combination can be used to risk stratify COVID-

CO N FLI C T O F I NTE R E S T
The authors who have taken part in this study declared that they do not have anything to disclose regarding funding or conflict of interest related to this manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author, upon reasonable request.