Premorbid IL-6 levels may predict mortality from COVID-19

Purpose: There is evidence that interleukin-6 (IL-6) levels are elevated in cases of complicated COVID-19, but it is also possible that this cytokine has a far more important role in the pathogenesis of viral infection. Methods: Data on COVID-19 mortality from Italy and the UK were compared with previously published results of mean IL-6 levels from these countries as well as from the USA. Results: There was a highly significant correlation (r = 0.9883;p = 0.00025) between age-stratified mortality rates and IL-6 levels from similar healthy individuals. Levels of IL-6 were proportionately higher in males, the elderly, individuals of black ethnicity and obese individuals, with similar findings in relation to COVID-19 mortality in these groups. Conclusions: IL-6 levels prior to infection may predict mortality, given that this cytokine directly facilitates viral cell entry and replication. This provides a rationale for prophylactic and therapeutic measures directed at lowering IL-6, including Vitamin D prescription.


Introduction
Previous studies have suggested that interleukin-6 (IL-6) levels are elevated in cases of complicated COVID-19 1 , and that this cytokine may be a potential biomarker for disease progression 2 . While increased IL-6 production may well be a consequence of severe infection, could it also be a predisposing factor? COVID-19 requires cathepsin L to infect pulmonary epithelial cells 3 , with cathepsin L production upregulated by IL-6 4 . This interleukin also likely upregulates expression of the ACE2 receptor which the virus utilizes for cellular entry 5 . If IL-6 levels prior to infection were found to correlate with disease severity, this might explain the documented poorer outcomes in males, with increasing age and in obese individuals 6 , given that levels of this cytokine are higher in males 7 , increase with advancing age 8 and in obesity 9 . This might also explain why Vitamin D appears to be protective against severe infection 10 , given that IL-6 production is down regulated by this vitamin 11 as well as the worse outcomes experienced by patients of African-American descent 12 , given that these individuals have higher IL-6 levels than other population groups 13 . I sought to determine whether IL-6 levels in healthy individuals from a similar demographic to patients infected with COVID-19 correlate with severity of clinical manifestations.

Methods
Given the range of severity of clinical manifestations of COVID-19, I limited my analysis to assessment of mortality. To limit any potential confounding factors, I searched the literature for results of IL-6 determinations in healthy individuals with included age and sex data from speci c geographical regions for which published mortality rates from COVID-19 were available. The largest study, in terms of numbers of subjects, by Stowe and colleagues 13 did not include detailed demographic information broken down into age groups, so I performed my analysis on data from an Italian study that examined IL-6 from the Tuscany region 14 . Data on COVID-19 mortality from Italy was sourced from statista.com 15 . I also examined published data on normal levels of IL-6 broken down by ethnic minorities 13 and Body Mass Index(BMI) 9 and compared these with a recently published study describing mortality by ethnic group and BMI 6 . Statistical analysis on age data was carried out using the Pearson Correlation Coe cient Calculator. Data on sex, ethnic minority status and BMI were insu cient for statistical analysis, and these were assessed qualitatively. Age data were strati ed slightly differently in each study, with the majority quoting in increments of 10 years from 20 years of age upwards. Where included studies strati ed age groups in different formats, ages were rounded to nearest multiple of 10. For example, recently published UK data included a category of 18 to 40 years of age 6 ; this category was rounded up to 20 to 40 in my analysis. Other studies, including that by Ferruci and colleagues 14 included categories in the format "20 to 39" years of age, which was rounded up to 20 to 40.

Results
The data on Italian mortality 15 provided raw numerical deaths by population groups but did not provide information on strati ed mortality rates by age group. To assess the validity of using this data, I compared these results with strati ed mortality rate data in a large UK study 6 (see Table 1). There was an extremely strong correlation between age strati ed Italian raw numerical deaths and UK mortality rates (r(6) = 0.9837; 95% CI: 0.85 -1.0; p = 0.00396, see Fig. 1). Given the likely validity of using the Italian data, I compared this with IL-6 levels derived from a healthy Italian population 14 . There was an extremely strong correlation between age strati ed Italian raw numerical deaths and IL-6 levels (r(6) = 0.9883; 95% CI: 0.89 -1.0; p = 0.00025, see Table 2 and Fig. 2). As a broad test to con rm this, I also compared UK death rates with historic data published by Wei and colleagues 16 on IL-6 levels in a similar population. Mean death rates in patients over 60 were more than 10 times higher than in patients under 50; mean levels of IL-6 were higher in older patients, although only by 26% (Table 3). A similar discord was present in a comparison between UK death rates and a diagnosis of obesity, using IL-6 levels from a US study 9 . Mean death rates in non-obese patients were 48% higher than in non-obese patients; this compared with IL-6 levels more than 300 times greater in obese patients (Table 4). However, sex and ethnicity data demonstrated a much closer relationship between death rates and mean IL-6 levels. Male death rates were 72% higher than amongst females, compared to mean IL-6 levels for males being 112% greater than mean female levels (Table 5). Similarly death rates amongst UK patients of black ethnicity were 40% higher than amongst white patients, compared with 59% greater IL-6 levels amongst black patients derived from a US study 13 (Table 6).

Discussion
This is the rst study to have demonstrated a signi cant correlation between IL-6 levels in matched healthy individuals and age-strati ed mortality from COVID-19. Despite the format of data available for other variables deemed to be risk factors for increased mortality including sex, obesity and ethnicity 6 , not lending itself to statistical analysis, premorbid IL-6 in matched populations had similar higher levels when compared with mortality rates based on these risk factors. This suggests that IL-6 may have a far more important role in the pathogenesis of viral infection than previously thought. IL-6 upregulates cathepsin L production 4 , which then cleaves the virus, enabling it to infect pulmonary epithelial cells 3 . IL-6 also upregulates the Angiotensin II AT1 receptor 17 , and likely upregulates pulmonary epithelial cell ACE2 receptors, with a positive correlation between IL-6 and ACE2 levels in bronchioalveolar lavage (BAL) in patients with severe acute pulmonary in ammation 5 . Viral reproduction within pulmonary epithelial cells then recruits in ammatory cells which, in turn produce cytokines (as part of the "cytokine storm" phenomenon 18 ) including IL-6. This could be expected to set up a noxious cycle between immune cells and pulmonary epithelium which the virus utilizes to enter cells and replicate. The principal predisposing factor could well be IL-6, with premorbid levels correlating with strati ed mortality rates. There are other manifestations of this infection that could readily be attributed to IL-6. Levels of this interleukin are generally low in children, but there is a short spurt in early childhood 19 , correlating with a recently described severe COVID-19 manifestation in young children, a Kawasaki-like condition 20 . IL-6 is also thrombogenic, which may well account for this signi cant cause of death from the virus 21 . Intense exercise also produces large quantities of IL-6 22 , which may account for the more severe clinical manifestations of infection amongst athletes 23 .
How can this be utilized to reduce mortality? The role of Vitamin D de ciency as a predictor for worse clinical outcomes remains the focus of investigation, although early results appear to support this contention 10 . What is unclear is whether Vitamin D exerts a general immune-enhancing effect, or whether it may have a speci c role in prevention and treatment of COVID-19. The results herein reported point strongly to the latter, given that Vitamin D lowers immune cell production of IL-6 11 . In addition to Vitamin D prescription, both prophylactic, and, potentially as a speci c therapy 18 , there may well be other therapeutic strategies which could be deployed, including nutritional supplementation, such as ingesting Brazil nuts 24 , and administration of drugs which lower IL-6 production such as doxycycline 25 and budesonide/formoterol 26 .
While the results of this work offer a rationale for the diverse clinical manifestations of COVID-19 and a range of therapeutic options for trial, this must be tempered by the limitations of my analysis. The data analysed were not strictly comparable, with premorbid data on IL-6 levels derived from different populations to those with mortality data. Further, only the age-strati ed results are supported by statistical analysis as the other data on other factors including sex, ethnicity and obesity could not be assessed directly for statistical signi cance. While substantial premorbid data on IL-6 levels may never be obtainable in patients who die from COVID-19, prospective trials of therapeutic measures can certainly be undertaken, but, given that Vitamin D causes little harm if administered under clinical supervision, this study certainly supports its widespread prescription.
In conclusion, this study has demonstrated a highly signi cant correlation between premorbid IL-6 levels and mortality rates from COVID-19, accounting for a range of previously described clinical predictors and, potentially directing future therapeutic strategies.  Legend: Mean IL-6 levels and death rate both higher in older individuals Legend: Mean IL-6 levels and death rate both higher in obese populations Legend: Mean IL-6 levels and death rate both higher in males Figure 1 Age and death data from Italy and UK populations.Legend: Age-strati ed deaths in Italy correlate strongly with UK death rates