Low-density lipoprotein cholesterol levels are associated with poor clinical outcomes in COVID-19

2.1 Study design and population

We conducted this ambispective single-center study at a Spanish tertiary hospital (Hospital Clínico Universitario de Valladolid, Spain, Valladolid). Between March 1^st, 2020, and May 15^th, 2020, data from a total of 654 consecutive patients with a definitive diagnosis of SARS-CoV-2 infection confirmed through positive reverse transcriptase-polymerase chain reaction (RT-PCR) were prospectively collected. Exclusion criteria were age < 18 years old, pregnant women, and death in the first 24 h after admission. We retrospectively collected the baseline clinical features, radiologic procedures, and laboratory tests. We also recorded ongoing treatment and clinical outcomes from the electronic medical records. Special attention was given to the accurate recording of the prescribed therapy before admission and during hospital stay according to our institution’s protocols and discretion of the medical team. Follow-up continued through until July 25^th, 2020. The local ethics committees approved the study protocol. Informed consent was waived, given the ambispective and observational nature of the study. The work was carried out by following the guidelines of the Declaration of Helsinki of the World Medical Association.

2.2 Outcome measure

The primary endpoint was to investigate any relationship between serum levels of lipid markers in the first hours of admission and all-cause mortality within 30-days. Secondary endpoints were to determine (1) the correlation between lipid markers at admission with other inflammatory markers and (2) to monitor the changes in lipid mediator’s levels during hospitalization as well as their correlation with the prognosis.

Following the institution’s protocols, a blood sample with a metabolic and lipid profile was drawn on the first 24 hours and the 7^th day during admission according to medical criteria. We assessed fasting total cholesterol (TC), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), and triglycerides (TG). We also recorded lipid profiles from the last eighteen months before the index event and during follow-up at the discretion of the general practitioner if available to evaluate lipid metabolism further.

2.3 Clinical laboratory tests

We carried out all tests at our certified clinical laboratory (ISO 9001:2015). White blood cells, lymphocyte, and monocyte counts were performed on Sysmex XN-1000® analyzer using the manufacture’s reagents (Sysmex Corporation, Japan). Alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase, lactate dehydrogenase (LDH), LDL-c, HDL-c, TC, and TG were tested on Roche Cobas 8100 sampling system analyzer (Module Cobas® c 702, Roche Diagnostics, Switzerland) using manufacture’s reagents. The methodology for direct LDL-c, HDL-c, and TC methods are based on a standard homogeneous enzymatic colorimetric assay. We tested interleukin 6 (IL6) on IMMULITE® 2000 immunoassay system using manufacture’s luminescent immunoassay reagents (IMMULITE® 2000 IL6, Siemens Healthcare Diagnostic, Germany). Procalcitonin (PCT) measurement in plasma was performed by electrochemiluminescence immunoassay on a chemistry analyzer (Cobas 6000, Roche Diagnostics) limit of detection, as below 0.02 ng/ml. We measured serum C-reactive protein (CRP) by particle enhanced immunoturbidimetric (e501 Module Analyser, Roche Diagnostics). The limit of detection was set below 0.3 mg/L.

2.4 Statistical analysis

We reported categorical variables as absolute values and percentages. Continuous variables were reported as median (interquartile range [IQR]). The normal distribution of continuous variables was verified with the Kolmogorov-Smirnov test and q-q plot. Categorical variables were compared with the chi-square test and the Fisher exact test when necessary. We compared continuous variables with the Mann-Whitney U test. A Spearman test was performed to analyze the correlation between lipid variables with the rest of the serum markers. We assessed the accuracy of analyzed variables to identify non-survivors patients by using the area under the receiver operating characteristic curve analysis (AUROC). We determined the optimal operating point (OOP) in the AUROC as the one that equaled sensitivity and specificity regarding mortality, and we used it as the cut-off point in the lipid profiles. We analyzed time to 30-day mortality by Kaplan–Meier survival curves and compared using the log-rank test. In a further step, we performed a multivariable Cox-regression analysis to determine the predictors of 30-day mortality in the global study population. Proportional hazard assumptions were verified by Shoenfeld residual test and check with using the log(-log(survival)) plots. We performed the statistical analyses with the use of R software, version 3.6.1 (R Project for Statistical Computing) and IBM SPSS Statistics, Version 25.0. Armonk, NY: IBM Corp. Differences were statistically significant when the p-value was < 0.05.

3.1 Baseline and clinical characteristics

Main baseline and clinical characteristics at admission are summarized in Table 1. A total of 654 patients were admitted due to COVID-19 with an estimated 30-day fatality rate of 22.8%. Non-survivors were older (82 vs. 66 years old; p<0.001), had lower baseline oxygen saturation levels at admission (91 vs 95; p<0.001), developed a significant higher degree of respiratory failure (92.5% vs. 43.3%; p<0.001) and required ICU admission more frequently (20.3% vs. 8.1%; p<0.001) than survivors. Non-survivors showed a greater prevalence of hypertension (74.5% vs. 46.7%; p<0.001), diabetes mellitus (31.5% vs 17%; p<0.001), dyslypemia (52.3% vs. 34.3%; p<0.001) as well as other comorbidities such as chronic kidney disease (19.5% vs. 5.7%; p<0.001) or ischemic heart disease (14.8% vs. 8.1%; p= 0.016). In accordance, non-survivors were more commonly treated prior to admission with antihypertensive drugs, oral antidiabetics, antiplatelets and statins. Time from symptom onset to admission was shorter (4 vs. 7 days; p<0.001) in the non-surviving cohort. They also displayed greater levels parameters of organ damage and inflammation such as creatinine (1.16 vs. 0.81 mg/dL; p<0.001), D-dimer (1,394 vs. 664 ng/ml; p <0.001), C-reactive protein (128.4 vs. 54.5 mg/dL; p<0.001), procalcitonin (0.33 vs. 0.08 ng/mL; p< 0.001), LDH (357 vs. 265 UI/L; p<0.001) and increased IL6 levels (52.1 vs. 18.4 pg/mL; p<0.001). In contrast, non-survivors had lower blood count of haemoglobin [12.4 vs. 13.3 g/dL; p<0.001) lymphocytes [805 vs. 1,130 cells/mm³); p<0.001) and platelets [183 vs. 218 cells/mm³×10³; p <0.001).

Specific COVID-19 treatment was more commonly prescribed in survivors with comparable prescription rate in respect to statins (7.4% vs. 7.4%; p= 0.996) and corticosteroids (57.2% vs. 63.1%; p= 0.202). Nosocomial infection (31.5% vs. 14.9%; p<0.001) were more common among non-survivors during hospitalization. Table 1 shows the main clinical outcomes and laboratory findings during hospitalization.

3.2 Lipid profile among COVID-19 patients

Lipid profiles were tracked from previous laboratory reports before SARS-CoV-2 infection (when available), at hospital admission, at 7^th day during hospitalization, and until death or first-time follow-up after discharge (see Figure 1 and Table 2). In the overall population (including survivors and non-survivors), serological levels of all the lipid markers analyzed, except TG, displayed a significant decrease at the time of admission concerning the previous baseline values. Besides, baseline serum TC and LDL-c levels were significantly higher in survivors than non-survivors at any time; whereas, HDL-c was comparable at admission but significantly lower in non-survivor before (47.2 vs. 52.6 ng/dL; p = 0.004) and in the 7^th day (27 vs. 34 mg/dL; p= 0.011) after hospital admission. Survivors had a progressive recovery of TC, LDL-c, and HDL-c levels; however, this was not the case in the non-survivor cohort. Finally, median follow-up among survivors was 73 days with complete recovery to previous basal levels of their lipid profiles (see Figure 1 and Table 2).

• Download figure
• Open in new tab

Figure 1.

Temporal changes in lipid profile levels in COVID-19 patients according to the clinical course of the disease: A) Total cholesterol; B) LDL-c; C) HDL-c.

The horizontal lines represent the median value in each group. HDL-c: High-density lipoprotein cholesterol; LDL-c: Low-density lipoprotein cholesterol; TC: Total cholesterol

* p <0.05

• Download figure
• Open in new tab

Figure 2.

Kaplan-Meier estimates of mortality in the total COVID-19 population according to LDL-c A) at admission and B) at the 7^th day of hospitalization based on their optimal cut-off point.

3.3 Correlation of lipid markers with other inflammatory markers

A Spearman correlation analysis assessed the relationship of lipid parameters with all the gathered analytical parameters. Interestingly, LDL-c and TC levels at admission were inversely correlated with the levels of CRP (r= −0.217 and −0.209, respectively; p<0.001), PCT (r= −0.313 and − 0.229; p <0.001) and IL6 (r=-0.334 and −0.301; p<0.001) with a positive correlation with lymphocytes (r=0.179 and 0.191; p= 0.001), which was maintained throughout hospitalization and until recovery (see supplementary Table 1). LDL-c did show a very strong positive correlation with TC (r= 0.937; p<0.001) as opposed to HDL-c (r= 0.201; p<0.001).

3.4 AUROC curve analysis

We analyzed the diagnostic performance accuracy of lipid profiles to predict 30-day mortality (see supplementary Figure 1). Only LDL-c and TC showed a significant AUROC during admission and 7^th day of hospitalization; in particular, LDL-c showcased better AUROC than TC regardless time course of the disease. Hence, AUROC curve for LDL-c at admission was 0.7 (95% CI 0.64-0.75; p<0.001) as opposed to TC with an AUROC of 0.67 (95% CI 0.63-0.71; p<0.001). On the 7^th day, AUROC for LDL-c and TC were 0,75 (95% CI 0.63-0.82; p<0.001) and 0.7 (95% CI 0.6-0.77; p<0,001), respectively. The estimated threshold values for LDL-c were 69 mg/dl on admission and 75 mg/dl in the 7^th day of hospitalization. On the contrary, TC estimated cut-off values were 132 mg/dL and 147 mg/dL, respectively.

3.5 Association between mortality and lipid profile

Variables associated with 30-day mortality in multivariate analysis are summarized in Table 3. Independent predictors of mortality were estimated through a Cox multivariate regression analysis. The variables included in the model were those with a p-value <0.05 on the univariate analysis (age, hypertension, diabetes, dyslypemia, ischemic heart disease, chronic renal disease, angiotensin receptor antagonist, angiotensin converting enzyme inhibitors, statins, lymphopenia, c-reactive protein, antiviral treatment, anticoagulation, total cholesterol and LDL-c). It revealed that age (hazard ratio [HR] 1.08; 95% confidence interval [CI], 1.05-1.11]; p <0.001), lymphopenia <1,000 cells/ml (HR 2.68; 95% CI, 1.91-3.78; p<0.001), LDL-c <69 mg/dL (HR 1.94; 95% CI, 1.14-3.31; p = 0.014) and CRP >88 mg/dL (HR 2.44; 95% CI, 1.41-4.23; p= 0.001) at admission were independent variables associated with a greater risk of 30-day mortality. Those findings remained consistent on the seventh day of hospitalization with a cut-off of 75 mg/dL and 33 mg/dL for LDL-c and CRP, respectively. Unadjusted survival Kaplan-Meier curves for 30-day global mortality was performed and shown in Figure 3, those patients with LDL-c levels <69 mg/dl at the time of admission and <75 mg/dl on the 7^th day showed a 20% higher 30-day mortality rate than the rest of the patients.