DISCUSSION
In our study, tofacitinib in addition to standard of care therapy did not reduce the composite end-point of invasive mechanical ventilation or death among hospitalised patients with moderately severe COVID-19. Most patients had signs of systemic inflammation, that is, persisting fever and/or elevated CRP. The primary end-point was not met both in patients requiring oxygen supply at the time of tofacitinib initiation and those with normal oxygen saturation. In multivariate Cox regression analysis adjusted for inverse propensity score weighting, the addition of tofacitinib to the standard of care therapy improved outcomes neither in hypoxic patients nor in those receiving no supplemental oxygen.
In both groups, analysis of the composite end-point rates non-significantly favoured tofacitinib. The percentages of patients who started mechanical ventilation or died during hospitalisation in the tofacitinib and control groups were 12.5% vs. 14.1%, respectively, among patients who required respiratory support, and 1.7% vs. 4.4%, respectively, in those with normal oxygen saturation. These differences could be related to the use of glucocorticoids that was two- to three-fold higher in the tofacitinib groups. The RECOVERY trial showed the efficacy of dexamethasone in reducing mortality only among hospitalised patients with more severe COVID-19 [4]. However, a favourable effect of glucocorticoids on the course of COVID-19 might also be present in cases with signs of excessive inflammatory response, even in the absence of hypoxia at the time of hospitalisation.
In the ACTT-2 trial, the beneficial effects of the combination treatment with baricitinib and remdesivir included a 1-day shorter time to recovery and a greater improvement in clinical status as assessed on the ordinal scale [10]. In contrast, we did not evaluate the time to clinical improvement or recovery, since we could not validate these data that were collected retrospectively for control patients. The assessment of time to recovery was even more challenging in patients with normal oxygen saturation. Therefore, the hard end-points of death or invasive ventilation seemed to be more suitable criteria of efficacy for our non-randomised clinical study.
The use of glucocorticoids was prohibited by the protocol of the ACTT-2 trial, although these medications were permitted for standard indications including septic shock and acute respiratory distress syndrome. In the ACTT-2 study, dexamethasone was administered to only 6.0% of patients in the baricitinib group. On the contrary, systemic glucocorticoids were used in 82% of patients who were enrolled in the tocilizumab arm of the RECOVERY study. Addition of tocilizumab to glucocorticoids resulted in 20% reduction in the risk of all-cause mortality, whereas this benefit was not seen in patients who did not receive glucocorticoids [7]. These findings suggest that in patients with COVID-19 tocilizumab and probably other immunomodulators should be considered in addition to glucocorticoids, particularly in those who do not respond to initial anti-inflammatory therapy or present with severe or progressive disease. In our study, 31.2% of patients received intravenous dexamethasone. Addition of tofacitinib to glucocorticoids was associated with a more than two-fold reduction in the occurrence of the composite end-point of death or mechanical ventilation compared to controls among patients with low oxygen saturation (12.9% vs. 28.2%). Both the need of mechanical ventilation and all-cause mortality rates were lower in the tofacitinib group. However, the differences between the two groups were not significant, probably as a result of limited number of enrolled patients. The incidence of the primary and secondary end-points was low in patients with normal oxygen saturation and did not differ between the two groups. Tofacitinib was well tolerated in the studied population and was discontinued in only 5.4% of patients.
Our findings are in contrast with the results of the STOP-COVID Trial, in which treatment with tofacitinib compared to placebo resulted in a lower cumulative incidence of death or respiratory failure through day 28 (risk ratio, 0.63; 95% CI, 0.41 to 0.97; p = 0.04) and reduction in the proportional odds of having a worse score on the eight-level ordinal scale (0.54, 95% CI, 0.27 to 1.06) at day 28, whereas the difference in the mortality rates between the tofacitinib and placebo groups did not reach statistical significance [11]. A higher rate of glucocorticoids administration (78.5%) and a higher dose of tofacitinib (10 mg twice daily) are the possible explanations of the better outcomes of JAK-inhibitor use in the STOP-COVID Trial.
Our study has several limitations. First, we could not account for confounders inherent to the study design, although we adjusted the baseline model for inverse propensity score weighting. Control patients were selected randomly from the population of COVID-19 patients who were hospitalised during the first wave of the pandemic in Russia. Nevertheless, selection biases cannot be ruled out. Second, the statistical power of our study was limited, particularly among patients with normal oxygen saturation given the low incidence of events. However, we evaluated tofacitinib’ efficacy in a relatively large sample of patients with COVID-19.
In summary, tofacitinib in addition to standard of care therapy did not reduce the risk of invasive mechanical ventilation or death in patients with moderately severe COVID-19. Analysis of the composite primary end-point and the secondary end-points favoured tofacitinib, particularly among patients with low oxygen saturation who received intravenous dexamethasone. However, all the differences between tofacitinib users and controls were not significant.