DISCUSSION AND CONCLUSIONS
Our retrospective cohort study reveals that, during the firsts weeks of the SARS-CoV-2 outbreak in Spain (March 2020), individuals with IMIDs exposed to biologic and synthetic DMARDs presented a lower risk of hsCOVID-19 symptoms than individuals with IMIDs unexposed to these treatments. Previous preliminary observations also support our findings (Haberman et al., 2020; Feldmann et al., 2020; Michelena et al., 2020). These results underline the interest of starting clinical trials with compounds that decrease immunological responses to evaluate their possible efficacy in minimizing the progression of the disease.
The three families of monoclonal antibodies approved to treat rheumatoid arthritis are directed against IL-6, B lymphocyte surface protein CD20 and TNFα, three targets of potential protective compounds for hsCOVID-19 symptoms. TNFα, IL-6 and B lymphocytes have been reported to play a crucial role in the inflammatory cascade taking place days before the manifestation of the most severe forms of SARS-CoV-2 infection (Zhou et al., 2020), and in the physiopathological processes leading to rheumatoid arthritis (Ceribelli et al., 2020). In the present study, TNF-α antagonists were the most abundant biologic DMARDs and showed a 50% reduction in the relative risk of hsCOVID-19 symptoms. Nevertheless, our cohort included a limited number of patients treated with two important groups of immunomodulatory compounds, IL-6 (52 patients) and B lymphocyte antagonists (42 patients). Interestingly, none of these 94 patients showed hsCOVID-19 symptoms, which agrees with the reported efficacy of the IL-6 antagonists tocilizumab (Xu X et al., 2020) and sarilumab (unpublished observations) in COVID-19 treatment, and with the protective effect of IL-6 antagonists shown in a cross-sectional recent study (Michelena et al., 2020).
Regarding synthetic DMARDs, although their global effect was protective (RR=0.62; 95%CI 0.43, 0.91), our results point that not all synthetic DMARDs would play a protective role against the appearance of hsCOVID-19 symptoms. On the one hand, methotrexate was the most abundant synthetic DMARD and presented a 30% reduction in the RR of hsCOVID-19 symptoms. Among individuals exposed to leflunomide (N=111) only 5 presented hsCOVID-19 symptoms. Interestingly, leflunomide produces a stronger and widespread decrease in pro-inflammatory ILs than classical DMDARDs, such as methotrexate with limited effects in ILs expression (Kraan et al., 2004). On the other hand, the exposure to chloroquine/hydroxychloroquine was not associated with a statistically significant protective effect of hsCOVID-19 symptoms (RR=0.76; CI95% 0.36, 1.62). These compounds have been widely used to treat COVID-19, although their benefit/risk was questioned due to the side-effects and moderate efficacy in disease progression, in agreement with our results. Accordingly, clinical trials have been now withdrawn due to the recently reported decreased in-hospital survival associated with increased risk of ventricular arrhythmia during hospitalization (Mehra et al., 2020; Adhanom Ghebreyesus, 2020).
The severity of COVID-19 symptoms shows clear sex differences with more severe cases and higher mortality reported in men than women (Zhang et al, 2020). Similar sex differences were revealed in the outcome of influenza A virus infections probably mediated by the impact of sex steroid hormones in immune responses (Vom Steeg and Klein, 2019). Although the studied population was not sex-balanced (1592 womenvs 902 men) our analyses stratified by sex also revealed potential sex differences in the effects of several immunomodulatory compounds. Indeed, anti-TNFα compounds show higher protective effects in women (RR=0.33, 95%CI 0.17, 0.64) than in men (RR=0.76, 95%CI 0.41, 1.43). Although a possible sex influence in the therapeutic effects of anti-TNFα compounds is controversial, a better prognosis of ulcerative colitis has been reported in women treated with infliximab, an anti-TNFα monoclonal antibody (Nasuno and Miyakawa, 2017). Sex differences were also revealed in our study in glucocorticoid effects. Taken into account the high variability of the doses of glucocorticoids used in these patients (Ruiz-Irastorza and Danza, 2012) and the differential effects depending on dose exposure (Meng et al., 2020), we have stratified glucocorticoid treatment in low (≤10 mg of prednisone or equivalent) and high doses (>10 mg). Women exposed to low glucocorticoids doses had a reduced RR of hsCOVID-19 symptoms (RR=0.72, 95%CI 0.42, 1.22), whereas high doses produce the opposite effect (RR=1.62, 95%CI 0.75, 3.52). Considering the high availability and safety profile of low doses of glucocorticoids, it would be of interest to evaluate the potential use of such low doses in women as a potential treatment in early periods of SARS-CoV-2 infection to prevent disease progression.
In spite of the potential beneficial effects of biological and synthetic DMARDs, those patients receiving a combination of both groups of compounds (n=298) show enhanced incidence of hsCOVID-19 symptoms (RR=4.30, CI95% 2.0, 9.3). The strong immunosuppression that should result by the combination of these treatments and the severity of the diseases targeted by these drug combinations may explain this paradoxical effect. Indeed, previous studies have reported that more patients experienced infectious adverse events when increasing doses of synthetic DMARDs were combined with anti-TNFα compounds (Burmester et al., 2015; Honkila et al., 2019). In addition, the main reason for combining both treatments is related to the lack of efficacy in these particular patients (Van Vollenhoven et al., 2012), which could also have influenced our results.
Finally, we made a sub-analysis of patients aged 60 or older taking into account their worse prognosis of the disease (Wu and Mc Googan, 2020). Biological DMARDs showed the same protective effect in both groups of patients (RR<60=0.47 and RR>60=0.47), although the subgroup of anti-IL6/12/17/23 ILs showed higher effects in patients under 60 (RR=0.36) than in older patients (RR=0.77). Synthetic DMARDs effects were higher in patients aged 60 or older (RR=0.48) than in younger individuals (RR=0.77). In contrast, protective effects of low doses of glucocorticoids were higher in women under 60 (RR=0.50) than in older women (RR=0.91) underlying the potential interest of this treatment in this particular population.
The results obtained here may be interpreted in the light of the following limitations. First, taking into account that the main interest of our study was focused on the early stages of COVID-19 pandemic in Spain, we included patients showing hsCOVID-19 symptoms due to the scarcity of COVID-19 tests that for ethical reasons were only reserved to patients showing more severe disease symptoms. Indeed, large cohorts of patients with mild symptoms and confirmed COVID-19 tests were not available in Spain at these early moments. Therefore, we and others (Michelena et al., 2020) have used a similar research strategy based on patients showing hsCOVID-19 symptoms as the unique possible research approach available to perform this kind of early studies in Spain on the protective effects of medication pre-exposure. It is also important to underline that the symptoms were recorded from 14 days before the COVID-19 alarm was announced in Spain (March 16th) when patients could be supposed to protect themselves more if they are at risk. Therefore, this potential self-protection would not represent any possible bias for the interpretation of our results considering the time schedule of our symptoms recording. Second, the indications for each treatment not only depend on the underlying pathology, but also on the specific clinical manifestations of each patient, and some of the indications are risk factors of COVID-19 (Sawalha and Zhao, 2020). Given the heterogeneity of the studied treatments and underlying pathologies, it is difficult to analyse all the factors that could cause confounding by indication. However, RR estimates of hsCOVID-19 symptoms after propensity score matching with some of the covariates that predict receiving anti-TNFα and other treatments were similar than RR estimates in the unmatched sample (Supplementary Table 6). The slightly different RRs found with these treatments matching the above mention covariates suggest that some of these IMID may represent an increased risk for hsCOVID-19 symptoms. Indeed, these particular comorbidities have been reported to increase COVID-19 susceptibility (Sawalha and Zhao, 2020). Furthermore, patients receiving these immunomodulatory treatments have an enhanced propensity to bacterial infection (Chiu and Chen, 2020) that could eventually provide manifestations similar to hsCOVID-19 symptoms. In spite of this possible bias that would impair the results obtained with these treatments, we have obtained promising RRs with these compounds that suggest significant protective effects on COVID-19 symptoms.
In summary, our results report that pre-exposure to DMARDs does not enhance the incidence of COVID-19 symptoms. In contrast, we identify specific DMARDs with different immune-depressor mechanisms that decrease hsCOVID-19 symptoms incidence with potential sex differences. These results underline the interest of starting clinical trials in women with anti-TNFα compounds in early periods of SARS-CoV-2 infection to evaluate their possible efficacy in minimizing the severity of COVID-19 progression. We have also revealed protective effects of low doses of glucocorticoids in women, which may also open new possibilities for low cost and safe early COVID-19 treatment mainly considering the current progression of the disease in countries with limited health resources. Interestingly, the protective effects of anti- IL6/12/17/23 compounds were mainly revealed in patients younger than 60, which underlies the need of an appropriate patient stratification for optimizing the results of the future clinical trials. The protective effects reveal with these immunomodulatory compounds may open novel therapeutic strategies to avoid serious COVID-19 manifestations, future deaths and, ultimately, the collapse of the health system.
Acknowledgements: We thank all patients who participated in the study, Gemma Vilagut (PhD) for her guidance and support on the statistical analysis and Mònica Gratacós (PhD) for translating the clinical protocol into English, for English language support and proofreading of the manuscript.
Competing interests : All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf, and declare: the submitted work was supported by the Hospital del Mar; outside the submitted work, NSD has received funding from Centro de Información Cerveza y Salud (CICS); AG has received research grants or consulting fees from Astrazeneca and Bioiberica S.A.U., RM has received research grants or consulting fees from Aelis, Almirall, Boehringer Ingelheim, BrainCo, Esteve, Ferrer, GlaxoSmithKline, Grünenthal, GW Pharmaceuticals, Janus, Lundbeck, Pharmaleads, Phytoplant, Rhodes, Sanofi, Spherium, Union de Pharmacologie Scientifique Appliquée, Upjohn, and Uriach; JM has received grants or consulting fees from Procare Health Iberia S.L, Esteve, Labhra, Bioibérica S.A.U, Grunenthal Pharma S.A, Pfizer, OPKO Heath Spain S.L.U and Roche Pharma S.A. LT, JL-O, PN, EM-G and RdlT declare no competing interests.
Funding : “Ministerio de Ciencia, Innovación y Universidades” (#AEI-SAF2017-84060-R FEDER to RM, #DPI2016-80283-C2-2-R), “Ministerio de Sanidad, Servicios Sociales e Igualdad” (#RD16/0017/0020 & #PNSD-2017I068 to RM, #PI18/00059 to TCS-M) and “Generalitat de Catalunya” (#2017-SGR-669 & #ICREA-Acadèmia 2015 to RM, #2017-SGR-138 to RdlT). NSD is recipient of predoctoral fellowship #2019-DI-47 from the DIUE-AGAUR of the “Generalitat de Catalunya”. There is not a private sponsor for this study.
Ethical approval : The observational study was approved by the Parc de Salut Mar Ethical Committee on Clinical Studies (ref. 2020/9246) before it started, and was monitored by the Clinical Trial Unit of Rheumatology Service at Hospital del Mar. Due to the nature of the study (all the data are completely anonymous), the importance of immediate results, and their implication for the treatment of the SARS-COV-2 pandemic, it is not planned to obtain informed consent from the participants. The Clinical Protocol and the STROBE checklist are attached as Supplementary files.
Data sharing : Requests to access data should be addressed tojmonfort@parcdesalutmar.cat. Deidentified individual participant data (including data dictionary) will be available to medical researchers by request in accordance with local registration and ethical approval when the article has been published until 30/4/2030. All proposals requesting data access will need to specify an analysis plan and will need approval of the scientific board before any data can be released.
The lead author (RM) affirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned and registered have been explained