In this single centre, retrospective study we compared the efficacy of LPV/r and DRV/r combinations for the treatment of patients with COVID-19 during the second wave of pandemic. We found that the treatment with LPV/r was associated with better survival in hospitalised patients, but not with DRV/r. Comorbidities such as diabetes increased the risk of mortality in case of DRV/r but not with LPV/r. Lopinavir-ritonavir was particularly beneficial in patients with moderate-to-severe lesions on chest CT scan; however, arterial hypertension and the need for oxygen supplementation reduced its effectiveness.
Similarly to the HIV aspartic protease, the cysteine protease of SARS-CoV-2 was hypothesized to be a reasonable target of repurposed antiretroviral protease inhibitors [19]. In vitro studies demonstrated the efficacy of lopinavir-ritonavir against SARS-CoV-2 and its ability to bind to SARS-CoV 3C-like protease, thus inhibiting viral replication [4, 20]. Apart from lopinavir- ritonavir, another protease inhibitor combination darunavir-cobicistat was tested in silico and found to have theoretical affinity to 3C-like protease and potential inhibitory effect on viral replication [21]. Later, an in vitro research demonstrated the lack of antiviral effects of darunavir on SARS-CoV-2 [22]. In the early desperate times of the pandemic, protease inhibitors came into focus again after successful treatment of a patient with mild COVID-19 [18]. Although lopinavir-ritonavir and hydroxychloroquine reduced organ support-free days and worsened the clinical outcomes among critically ill patients at an intensive care unit [23], protease inhibitors remained still a potential option for hospitalized mild to severe COVID-19 patients. In a randomized, controlled, open-label trial by Cao et al, the authors suggested that lopinavir-ritonavir has no benefit in hospitalised patients compared to standard care [15]. Although the clinicians started to abandon lopinavir-ritonavir based on the results of this trial, several arguments should be taken into consideration: the trial was underpowered due to small sample size (N = 199) and arguably, the treatment started at a median time of 13 days after symptom onset. Of note, the same trial reported several interesting findings with regard to the secondary outcomes suggesting lopinavir-ritonavir may be associated with reduced all-cause mortality (19% of patients in the lopinavir-ritonavir group vs. 25% of control group), decreased risk of severe adverse events (20% in lopinavir-ritonavir vs 32% of control group) and lower risk of severe respiratory failure (13% vs. 27%, respectively) [24]. Another group reanalysed these data and found that lopinavir-ritonavir can contribute to clinical improvement [17]. In this aspect, our results were consistent with their observation that lopinavir-ritonavir reduced mortality. A further publication advised to consider starting lopinavir-ritonavir earlier and speculated about the favourable effects of LPV/r, supporting the need for further clinical studies in this field [16].
In the RECOVERY trial LPV/r was not associated with reduced 28-day mortality, duration of hospital stay or the risk of progression to invasive mechanical ventilation. However, we have to evaluate these results cautiously as an overall 74% of patients required respiratory support in form of oxygen therapy at baseline and both the usual care and LPV/r were started at an average 8 days after symptom onset [25]. In line with this, oxygen therapy was associated with worse outcomes in the LPV/r group in our study. Additionally, the WHO SOLIDARITY trial included primarily patients already being ventilated or using respiratory support at the time of their recruitment.
Patients assigned to antiviral treatment had severe COVID-19, most of them lived in Asia and Africa where limited treatment resources were available, and some patients would have needed respiratory support [26]. Although the final results concentrated on the effects of remdesivir therapy, the interim research results suggested that LPV/r can have benefits in a defined group of patients (< 50 years) [27]. Further studies compared LPV/r to other drug combinations, and one of them found that a triple combination with LPV/r, interferon beta − 1b and ribavirin significantly improved National Early Warning Score (NEWS2), Sequential Organ Failure Assessment (SOFA) scores, hospital stay and decreased the time to negative viral load in nasopharyngeal specimens only if the patient allocation was done prior to 7 days after symptom onset [28]. These data also pointed out that similar to MERS and SARS, the early antiviral treatment may be crucial against SARS-CoV-2, and one should consider the wide randomisation windows in the evaluation of clinical trial results [29]. Additional studies aimed to compare LPV/r alone to other drugs such as umifenovir, novaferon or hydroxychloroquine, and to their combinations. However, these studies had no standard care group, had small sample size, poor statistical power and different control groups with various study population heterogeneity, therefore they have to be interpreted cautiously [30–32]. The TOGETHER trial enrolled COVID-19 patients with at least one clinical criterion for high risk and compared the effects of LPV/r or hydroxychloroquine to placebo. Although the participants received the treatment less than 8 days from symptom onset, the trial confirmed that neither LPV/r, nor hydroxychloroquine showed associations with COVID-19 mortality or hospitalization [33]. Another study confirmed that all paediatric patients with mild or moderate COVID-19 receiving LPV/r were cured and had reduced hospital stay [34]. Taken together, these data suggest that LPV/r may have clinical benefits in a predefined subgroup of hospitalized patients without baseline respiratory support. Indeed, our data is consistent with this theory, as patients without severe comorbidities or oxygen supplementation receiving LPV/r showed association with lower risk of COVID-19 associated death.
Darunavir, another protease inhibitor, was identified as a promising hit by computational methods which indicated it to be more effective against COVID-19 than LPV/r [35]. Darunavir was mostly used in combination with cobicistat (DRV/c), but this drug combination did not meet the initial expectations. DRV/c was effective at high concentrations against SARS-CoV-2 in vitro and proved to be more tolerable and safer than LPV/r [12, 36]. However, a study by Milic et al. confirmed that patients on DRV/c had higher mortality rate and risk for mechanical ventilation, compared to patients received standard care [37]. A further study compared DRV/c with LPV/r and revealed that DRV/c was associated with 89% increased risk of death if the patients were women, older, had severe infection and received hydroxychloroquine. Additionally, a subgroup analysis of data from this study showed a lower risk of death in patients with mild disease treated with LPV/r [38]. Another observational study comparing the efficacy of early administered DRV/c versus LPV/r unravelled that LPV/r was associated with faster time to recovery and virological clearance, but not DRV/c [39]. These findings can be also attributed to the unfavourable toxic side-effects of DRV/c [40]. The majority of darunavir is bound to plasma proteins and metabolized by CYP3A4. As ritonavir is a known CYP3A4 inhibitor, their combination can be theoretically a more efficacious drug against COVID-19 as DRV/c, however, there is an increasing risk of drug interactions and severe side-effects [41, 42]. In fact, our study is the first that evaluated the clinical outcomes of DRV/r on COVID-19 patients, compared to LPV/r or standard care group. Based on our results, DRV/r was associated with increased mortality rates compared to LPV/r treated patients, showed higher rates of impaired glucose metabolism and oxygen support, but we did not observe higher frequency of cardiovascular alterations.
To our knowledge, our study is the first to compare the clinical effects of LPV/r and DRV/r protease inhibitors in hospitalised patients with mild-to-severe COVID-19, however, our results have some limitations. Firstly, we checked the effects in a retrospective observational nature. The treatment options were highly dependent on the accessibility of LPV/r and DRV/r in our centre. A placebo-controlled project would have been also not accepted during the early desperate time of the pandemic, so we tried to exclude the potential confounding factors with propensity-score matching method. Additionally, our study was launched relatively early during the pandemic and it would have been complicated to set up a randomized prospective study and get an ethical approval for these combinations.