Antiviral efficacy of fluoxetine in early symptomatic COVID-19: an open-label, randomised, controlled, adaptive platform trial (PLATCOV)

Abstract

Background: The selective serotonin reuptake inhibitors (SSRIs) fluoxetine and fluvoxamine were repurposed for the treatment of early COVID-19 based on their antiviral activity in vitro, and observational and clinical trial evidence suggesting they prevented progression to severe disease. However, these SSRIs have not been recommended in therapeutic guidelines and their antiviral activity in vivo has not been characterised.

Methods: PLATCOV is an open-label, multicentre, phase 2, randomised, controlled, adaptive pharmacometric platform trial running in Thailand, Brazil, Pakistan, and Laos. We recruited low-risk adult outpatients aged 18-50 with early symptomatic COVID-19 (symptoms <4 days) between 5 April 2022 and 8 May 2023. Patients were assigned using block randomisation to one of eleven treatment arms including oral fluoxetine (40 mg/day for 7 days), or no study drug. Uniform randomisation ratios were applied across the active treatment groups while the no study drug group comprised ≥20% of patients at all times. The primary endpoint was the rate of oropharyngeal viral clearance assessed until day 7. Measurements were taken daily between days 0 and 7 and analysed in a modified intention-to-treat population (>2 days follow-up).The viral clearance rate was estimated under a Bayesian hierarchical linear model fitted to the log₁₀ viral densities measured in standardised duplicate oropharyngeal swab eluates taken daily over one week (18 measurements per patient). Secondary endpoints were all-cause hospital admission at 28 days, and time to resolution of fever and symptoms. This ongoing trial is registered at ClinicalTrials.gov (NCT05041907).

Findings: 271 patients were concurrently randomised to either fluoxetine (n = 120) or no study drug (n = 151). All patients had received at least one COVID-19 vaccine dose and 67% were female (182/271). In the primary analysis, viral clearance rates following fluoxetine were compatible with a small or no increase relative to the no study drug arm (15% increase; 95% credible interval (CrI): -2 to 34%). There were no deaths or hospitalisations in either arm. There were no significant differences in times to symptom resolution or fever clearance between the fluoxetine and the no study drug arms (although only a quarter of patients were febrile at baseline). Fluoxetine was well tolerated, there were no serious adverse events and only one grade 3 adverse event in the intervention arm.

Interpretation: Overall, the evidence from this study is compatible with fluoxetine having a weak in vivo antiviral activity against SARS-CoV-2, although the primary endpoint is also compatible with no effect. This level of antiviral efficacy is substantially less than with other currently available antiviral drugs.

Funding: Wellcome Trust Grant ref: 223195/Z/21/Z through the COVID-19 Therapeutics Accelerator.

Keywords: Antivirals; COVID-19; SARS-CoV-2; Viral clearance.

Fig. 1

Study CONSORT diagram for the fluoxetine versus no study drug analysis. In Thailand, pre-screening occurred in the Acute Respiratory Infection (ARI) unit of the Hospital for Tropical Diseases, Bangkok. Potentially eligible patients were selected by the ARI Nurses to be contacted by the study team. Therefore, a high proportion of those assessed for eligibility participated in the study.

Fig. 2

Antiviral effect of fluoxetine in early COVID-19. Panel A: individual viral densities data (fluoxetine: dark green; no study drug: light green). Triangles show the daily median oropharyngeal eluate viral densities by arm. Text annotations indicate the proportions of samples with viral densities below the limit of quantification (NSD: No study drug; FLX: Fluoxetine). Panel B: posterior estimates of the treatment effects of fluoxetine relative to no study drug, under the linear and non-linear models (orange: viral clearance assessed over 7 days; green: viral clearance assessed over 5 days). Thick and thin error bars represent the 80% and 95% CrIs, respectively.

Fig. 3

Estimated SARS-CoV-2 clearance half-lives (in hours) estimated over 7 days for individual patients in the fluoxetine arm (dark green), and the no-study-drug arm (light green). The median estimates (circles) and 80% credible intervals (error bars) are displayed. Vertical dashed lines indicate the median half-lives of each group.

Fig. 4

Individual patient data meta-analysis of antiviral interventions relative to the no study drug arm (n = 783 patients enrolled into the trial between the 30th September 2021 and the 23rd May 2023, not all concurrently). Circles show the median posterior estimates of the change in viral clearance rate under the linear model, adjusting for calendar time and virus variant (orange: viral clearance assessed over 7 days; green: viral clearance assessed over 5 days). Thick and thin error bars represent 80% and 95% centred credible intervals, respectively. The shaded area indicates the futility zone (<20% increase in viral clearance rate). Ivermectin was evaluated in the study between 30th September 2021 and 18th of April 2022; 95% had received at least one COVID-19 vaccine. Favipiravir was evaluated between 30th September 2021 and 31st October 2022; 97.5% had been vaccinated. Casirivimab/imdevimab was evaluated between 30th September 2021 and 24th August 2022. Remdesivir was evaluated between 30th September 2021 and 10th June 2022. Molnupiravir and ritonavir-boosted nirmatrelvir were both evaluated in the study from 6th June 2022. Molnupiravir was stopped on 23rd February 2023 and ritonavir -boosted nirmatrelvir remains as the positive control. The proportions vaccinated were 100% and 99% respectively.