Biological sex is associated with heterogeneous responses to IL-6 receptor inhibitor treatment in COVID-19—A retrospective cohort study

COVID-19 is associated with higher inflammatory markers, illness severity and mortality in males compared to females. Differences in immune responses to COVID-19 may underpin sex- specific outcome differences. We hypothesised that anti-IL-6 receptor monoclonal antibodies are associated with heterogenous treatment effects between male and female patients. We conducted a retrospective cohort study assessing the interaction between biological sex and anti-IL-6 receptor antibody treatment with respect to hospital mortality or progression of respiratory failure. We used a Cox proportional hazards regression model to adjust for age, ethnicity, steroid use, baseline C-reactive protein, and COVID-19 variant. We included 1274 patients, of which 58% were male and 15% received anti-IL-6 receptor antibodies. There was a significant interaction between sex and anti-IL-6 receptor antibody use on progression to respiratory failure or death (p = 0.05). For patients who did not receive anti-IL-6 receptor antibodies, the risk of death was slightly higher in males (HR = 1.13 (0.72–1.79)), whereas in patients who did receive anti-IL-6 receptor antibodies, the risk was lower in males (HR = 0.65 (0.32–1.33)). There was a heterogenous treatment effect with anti-IL-6 receptor antibodies between males and females; with anti-IL-6 receptor antibody use having a greater benefit in preventing progression to respiratory failure or death in males (p = 0.05).

Inclusion and exclusion criteria.We included patients who were COVID PCR positive on throat swab with a positive test within 5 days of admission.Only patients admitted to hospital were included for analysis.Patients from the first wave (defined as until 11/12/2020, when the Alpha variant became dominant according to COG-UK data) 10 were included if they had required supplemental oxygen during their admission.Following the first wave, patients were admitted for reasons unrelated to COVID-19 but were PCR positive for COVID on routine screening (incidental COVID-19).To ensure patients with incidental COVID-19 were not included, patients from subsequent waves were included only if they required supplemental oxygen therapy within 48h of admission and received treatment for COVID, with either steroids or another recognized COVID therapy.Patients who were transferred to UCLH from another hospital within 48 h of their initial admission with clinically diagnosed COVID were included.
Routine use of anti-IL-6 receptor antibodies (either tocilizumab and sarilumab) for the management of acute COVID-19 at our centre followed the initial announcement of the REMAP-CAP CAP study data 11 .Our criteria for use of anti-IL-6 receptor antibodies included patients admitted with COVID-19 and CRP > 75 mg/L; or patients admitted to ICU requiring respiratory support (non-invasive positive pressure ventilation, high flow oxygen therapy, or mechanical ventilation) within 24 h of admission.We avoided the use of anti-IL-6 receptor antibodies in patients with known or suspicion of bacterial infection, raised liver function tests (ALT or AST > 5 × upper limit), platelet count < 50 × 10 9 , or neutrophil count < 2 × 10 9 .
Primary and secondary outcomes.The primary outcome was hospital mortality.The secondary outcome was a composite of mortality and requirement for increased respiratory support.As the treating clinician may have deemed some patients unsuitable for escalation to invasive mechanical ventilation, a composite of mortality or requirement for increased respiratory support (respiratory deterioration) was taken as a secondary outcome.We previously described a strong correlation between serum IL-6 and CRP; IL-6 being a key regulator of C-reactive protein (CRP) production 9 .We therefore also investigated the association between use of anti-IL-6 receptor monoclonal antibodies and trajectory of CRP.Patients who died within 48 h of hospital admission were excluded.

Data validation.
Data was obtained via an automatic data pull from the hospital electronic health record database, and subsequently validated against independently collected datasets within the hospital.Two months of positive COVID-19 PCR tests were manually validated.Data were also cross validated against an independent virology dataset, manually collected by that department, with this dataset within 3% of the manually validated set.

Statistical analysis.
Analysis was performed using pseudo-anonymised data, and all analysis was performed within the hospital information technology systems and servers.Data analysis was performed in R 4.0.0.Groups were compared at baseline using chi-square, Mann-Whitney U and ANOVA tests (Table 1).Kaplan-Meier plots were produced to visualise survival trajectories, and unadjusted survival was compared between groups using a log-rank test.We used a Cox proportional hazards regression model to adjust for confounders including age, ethnicity, baseline illness severity (using CRP as a surrogate), and variant of COVID-19.
To investigate the association between CRP trajectory and other factors such as such as sex, ethnicity and variant, we used a mixed effects model with repeated measurements of CRP over the first 14 days of admission used.Inter-patient variability was modelled using random intercepts and slopes whilst fixed effects were used for age, sex and anti-IL-6 receptor monoclonal antibody use.Interaction terms between the variables of interest and day of admission were used to investigate whether CRP trajectory over time differs depending on these variables.CRP trajectory was modelled as linear, and plots were produced to validate this.Statistical significance was set at 0.05 for all analysis.
Approximately one quarter (23.3%) of patients received Remdesivir, and only two patients received a JAK inhibitor.All patients apart from one (188/189, 99.5%) receiving anti-IL-6 receptor monoclonal antibodies were also prescribed steroids.On hospital admission, most patients (69%) required supplemental oxygen alone, with only 5.6% of patients requiring mechanical ventilation.Overall hospital mortality was 22.9%.
After adjustment for age and sex, anti-IL-6 receptor monoclonal antibody use was associated with a greater rate of fall in CRP change over time compared to patients not treated with anti-IL-6 receptor monoclonal antibodies (p < 0.001) (Fig. 1).However, there was no significant interaction between the rate of fall of CRP and either sex alone or sex and anti-IL-6 receptor monoclonal antibody use (Supplementary Table 1).

Effect of IL-6 inhibitors on mortality
The unadjusted hospital mortality was not significantly different between males and females who did (p = 0.12) and did not (p = 0.17) receive an anti-IL-6 receptor monoclonal antibody (Fig. 2a), although the Kaplan-Meier plots do suggest a trend towards difference.Following adjustment for age, baseline CRP, ethnicity, steroid use, and COVID-19 variant, there was evidence that the effect of treatment with anti-IL-6 receptor monoclonal antibodies differs between males and females (p = 0.14), although this was not statistically significant.Among patients who did not receive anti-IL-6 receptor monoclonal antibodies, we observed increased mortality for male sex, although this was not statistically significant (HR = 1.56;CI (0.89-2.71)).By contrast, among patients who did receive treatment with anti-IL-6 receptor monoclonal antibodies, we observed decreased mortality for male sex (HR = 0.89; CI (0.36-2.24)).There was no effect of sex on response to treatment with steroids (Supplementary Figure 1).

Effect of IL-6 inhibitors on composite of progression of respiratory failure and mortality
There is a significant difference in baseline respiratory support between patients who were treated with and without anti-IL-6 receptor monoclonal antibodies (p < 0.001) (Table 1).Fifty-three percent of males who were administered anti-IL-6 receptor monoclonal antibodies required more than supplemental O 2 support on hospital admission, compared to 30% males who did not receive anti-IL-6 receptor monoclonal antibodies.Similarly, 49% of females who were administered IL-6 inhibitor required more than supplemental O 2 support, compared to 23% females who did not receive IL-6 inhibitors.Among patients who did not receive anti-IL-6 receptor monoclonal antibodies, there was an increased unadjusted risk of progression of respiratory failure or death among males compared to female patients; although this did not quite reach statistical significance (p = 0.058).The reverse was true among patients who received anti-IL-6 receptor monoclonal antibodies.There was an increased unadjusted risk of progression of respiratory failure or death among females compared to male patients; although this did not reach statistical significance (p = 0.066) (Fig. 2b).
Following adjustment for age, baseline CRP, ethnicity, and COVID-19 variant, there was a significant difference in progression to respiratory failure or death in response to treatment between males and females (p = 0.05).Among patients who did not receive IL-6 inhibitors, males were at increased risk of progression to respiratory failure or death compared to females (HR = 1.13;CI (0.72-1.79)).In contrast, among patients who did receive IL-6 inhibitors, males were at lower risk of progression to respiratory failure or death compared to females (HR = 0.65; CI (0.32-1.33)).This differential effect of treatment on the composite of progression of respiratory failure or death between males and females was not seen with treatment with steroids (p = 0.82) (Supplementary Figure 1).

Discussion
We demonstrate heterogeneity in response to treatment with anti-IL-6 receptor monoclonal antibodies between males and females with COVID-19.Sex had a significant interaction with anti-IL-6 receptor monoclonal antibody use on progression to respiratory failure or death; with male patients having a greater benefit associated with anti-IL-6 receptor monoclonal antibody use.Although the association between greater severity of illness and mortality in men compared to women in COVID-19 is well-described, there is paucity of data on the impact of COVID-19 treatments on clinical responses between male and female patients.
The immune response between males and females is fundamentally different.As a large number of genes related to immune functions are located on the X chromosome, X-linked mosaicism confers a highly polymorphic gene expression program that allows women to respond with a more expanded immune repertoire as www.nature.com/scientificreports/compared with men 5 .Differences in immune response between males and females extend from responses to bacterial infections to viral vaccines 12,13 .Despite a similar incidence of COVID-19 diagnoses in males and females in the community, the case fatality rate among males is significantly higher 14,15 .Biological mechanisms underpinning these observations have been investigated in an attempt to better understand the pathophysiology of COVID-19 15 .Several differences in the immune response between males and females have been described including higher pro-inflammatory innate immunity chemokines and cytokines in male patients 6,16 .Greater expression of virus entry factors (Angiotensinconverting enzyme 2 (ACE2) and accessory proteases (TMPRSS2 and CTSL) in airway secretory cells and alveolar type 2 cells in males may explain the greater cytokine levels in male patients 17 .In addition to differences in cytokine levels between sexes, a poor T cell response is associated with worse disease outcome in male patients, but not in female patients 16 .Despite stark differences in the proportions of males compared to females affected by COVID-19 in early reports 2 , few clinical trials in COVID-19 reported outcomes by sex.Where reported, differences in responses to treatment with immunosuppression have been described in some instances 18 .The National Institutes of Figure 2. Kaplan-Meier plots for (a) hospital mortality and (b) composite of progression of respiratory support or death in male and female patients who did or did not receive IL-6 inhibitors.The unadjusted hospital mortality was not significantly different between males and females who (ai) did not (p = 0.17) or (aii) did (p = 0.12) receive an anti-IL-6 receptor monoclonal antibody.(bi) Among patients who did not receive anti-IL-6 receptor monoclonal antibodies, there was an increased unadjusted risk of progression of respiratory failure or death among males compared to female patients (p = 0.058), although not statistically significant.(bii).Among patients who received anti-IL-6 receptor monoclonal antibodies, there was an increased unadjusted risk of progression of respiratory failure or death among females compared to male patients (p = 0.066) although not statistically significant.All p values were calculated using log-rank test.

Figure 1 .
Figure 1.Box and whisker plots of C-reactive protein (CRP) by day of admission in male and female patients who did or did not receive IL-6 inhibitors.Note logarithmic y-axis.