Reply

We read with great interest the letter by Yin et al,¹ where the authors comment the results of our recently published study.²

The authors raised some methodologic issues on our study design, suggesting a possible selection bias because we included both patients receiving homecare (survival rate of 100.0%; 82.1% received tacrolimus) and those requiring hospitalization (survival rate of 76.0%; 63.7% received tacrolimus) in the analysis. Against this argument we emphasize that “place of management” indicates the place with the highest intensity of care required by symptomatic patients during their disease course, and it should be considered as an intermediate outcome between onset of coronavirus disease 2019 (COVID-19) and death or recovery or a proxy for disease progression. If we selected patients based on place of management, we would, in contrast, introduce a selection bias that would affect the assessment of any protective effect of the baseline use of tacrolimus. In the end, tacrolimus was used more frequently in patients receiving homecare who never experienced a worsening of the disease, clearly supporting our finding of a protective effect, particularly in younger patients without comorbidities.

A second issue refers to changes of the calcineurin inhibitors (CNI) doses in hospitalized patients. Of the 57 inpatients who underwent withdrawal of CNI or a 25%–50% dose reduction, 18 (31.5%) were on lopinavir therapy, dose modifications being justified by the interference between the 2 drugs. For the remaining 39 patients, the modifications of CNI doses were proportionally distributed between cyclosporine and tacrolimus, making any selection bias unlikely. Further, 12 of 13 patients who stopped tacrolimus had interstitial pneumonitis requiring oxygen supplementation.

An additional issue points to the discordant results between our study and the only other study published on COVID-19 liver transplant patients by Colmenero et al.³ In this latter study from Spain, a different composite outcome was used, defined by the need of mechanical ventilation, intensive care, and/or death, thereby supporting our previous statement regarding place of management (intensive care unit) as an outcome and not as a risk factor. Notably, Colmenero et al³ reported a similar protective role of tacrolimus on development of severe COVID-19 with a relative risk of RR 0.54 (95% CI, 0.29–1.07; P = .08) in the univariate analysis and a relative risk of 0.19 (95% CI, 0.05–0.68; P = .011) in the initial multivariate model. However, a statistically significant association between tacrolimus use and severe COVID-19 was not confirmed in the final multivariate model. In the end, both studies suggest a protective role of tacrolimus, keeping in mind that the ELITA/ELTR study could benefit of a much larger sample size (243 patients vs 111 patients).

A final comment refers to the results of an unpublished meta-analysis from Yin et al¹ not showing a protective effect of tacrolimus. The meta-analysis was mainly focused on kidney transplant recipients, 7 of 11 studies, where the use of CNI is different from that on liver transplant recipients, with tacrolimus being more frequently used among long-term kidney transplant recipients (85% vs 68%). In addition, it is well-known that kidney transplant recipients have a higher level of immunosuppression compared with liver transplant recipients in general. The only 2 studies in liver transplant recipients included in the metanalysis have been already extensively commented on elsewhere in this letter. We therefore believe that pooling results from various types of solid organ transplants may bias the interpretation of the results

Having clarified these factors, we agree with the conclusions of Yin et al that tacrolimus is not a risk factor for mortality among all solid organ transplant recipients with COVID-19. For liver transplant recipients, however, the evidence emerging from the only 2 studies published to date, is that there may be some additional benefit. We hope further studies will better clarify this point.

Additional relevant comments come by Ruiz et al, who suggest that the lack of beneficial effect of cyclosporin A/other group may be explained by the higher prevalence of many confounders such as male gender, older age, longer interval from liver transplantation, more comorbidities, and mycophenolate mofetil use. However, the protective effect of tacrolimus persisted after including all the confounders in the multivariable analysis, the biological explanation for this finding remaining unclear. In the end we agree with the conclusions from Ruiz et al that tacrolimus has a beneficial effect in severe acute respiratory syndrome coronavirus 2 infection and therefore tacrolimus doses should not be modified. For the time being there is no evidence that shifting from cyclosporine A/other to tacrolimus could be of any benefit, thus it should not be done.