Remdesivir for COVID-19 in Hospitalized Children: A Phase 2/3 Study

Children and adolescents ≥28 days to 17 years who had SARS-CoV-2 infection confirmed by polymerase chain reaction (PCR) and were hospitalized and required medical care for COVID-19 were eligible. Cohort 1 participants were aged ≥12 to 17 years and weighed ≥40 kg (Fig 1). Cohort 2 through 4 participants were aged ≥28 days to 17 years and weighed ≥3 kg to <40 kg. Cohorts 5 through 7 (term and preterm neonates) are not included in this manuscript. Cohort 8 participants were aged <12 years and weighed ≥40 kg. Exclusion criteria included concurrent treatment with agents with known or possible direct antiviral activity against SARS-CoV-2 within 24 hours before study drug dosing; alanine aminotransferase (ALT) or aspartate aminotransferase (AST) >5X the upper limit of normal; estimated glomerular filtration rate <30 mL/min/1.73 m² using the Schwartz formula for participants aged ≥1 year; and creatinine values exceeding the 97.5th percentile¹⁵ or upper limit for age¹⁶ in those <1 year.

This was a phase 2/3, multicenter, open-label study. A total of 6 cohorts were enrolled in parallel between July 2020 and May 2021. RDV was dosed intravenously once daily for up to 10 days. Discontinuation was allowed before 10 days in cases of clinical improvement as judged by the investigator or clinical team. The weight-based dosing regimen tested in this study was derived using physiologically based pharmacokinetic (PK) modeling.¹⁷ For participants aged ≥12 years and weighing ≥40 kg, RDV 200 mg on day 1 was followed by 100 mg daily. For those aged ≥28 days and weighing ≥3 kg to <40 kg, RDV 5 mg/kg on day 1 was followed by 2.5 mg/kg daily.

The study protocol (https://clinicaltrials.gov/ProvidedDocs/81/NCT03022981/Prot_000.pdf) was approved by the institutional review board or ethics committee of each institution or by a centralized institutional review board, as applicable. Parents or guardians provided informed consent, and older children and youth provided assent when able, per local regulations. The study was conducted in accordance with the International Conference on Harmonization Good Clinical Practice Guidelines and the Declaration of Helsinki.

Twelve subjects for each cohort (cohorts 1–4) were to be enrolled in this study. Pharmacokinetic data from these participants were to provide >99% power for each cohort to conclude exposure equivalence of RDV AUC_τ in adolescents and children versus 25 healthy adult subjects in the study GS-US-399-5505 using two 1-sided tests with each performed at an α level of 0.05. In this power analysis, it was assumed that the expected geometric mean ratio of AUC_τ between the adolescents and children group versus the adult group is equal to 1, the intersubject SD (natural log scale) of AUC_τ is 0.18, and the equivalency boundary is 70% to 143%.

Twelve subjects from each cohort (cohorts 1–4) were also to provide >99% power to conclude exposure equivalence of RDV C_max in adolescents and children, compared with 26 healthy adult subjects in study GS-US-399-5505, assuming the expected geometric mean ratio of C_max between the adolescents and children group and the adult group is equal to 1, the intersubject SD (natural log scale) of C_max is 0.19, and the equivalency boundary is 70% to 143%.

The primary objectives of this study were to evaluate the safety and PK of RDV among participants aged 28 days to 17 years with laboratory-confirmed COVID-19. The primary outcomes were the proportions of participants with treatment-emergent adverse events or laboratory abnormalities, as well as the plasma pharmacokinetics of RDV and its metabolites.

Secondary objectives were to determine clinical improvement among participants and the antiviral activity of RDV. Endpoints were change from baseline in SARS-CoV-2 viral load as well as clinical improvement as based on the 7-point ordinal scale and the Pediatric Early Warning Score (PEWS) Scale (Supplemental Appendix). When possible, plasma concentrations of sulfobutylether β-cyclodextrin (SBECD) sodium were evaluated.

Safety assessments conducted at screening and every day of treatment administration included monitoring of adverse events and concomitant medications, vital signs measurements, respiratory status documentation, weight measurement, and physical examinations. The Medical Dictionary for Regulatory Activities, version 24.0, was used to code treatment-emergent adverse events. Severity grades were defined by the Division of AIDS Table for Grading the Severity of Adult and Pediatric Adverse Events, version 2.1, July 2017. Electrocardiograms were done during screening. Clinical laboratory analyses were done at screening and days 2, 5, 8, and 10. Documentation of respiratory status was measured using either respiratory rate (if not on a ventilator), oxygen supplementation, or oxygenation.

Plasma samples for measuring concentrations of RDV, RDV metabolites (GS-704277 and GS-441524), and SBECD sodium (where possible) were obtained on days 2 and 3, with 5 being an optional collection day. A newly developed liquid chromatography tandem mass spectrometry method was used to analyze SBECD sodium plasma concentrations (lower limit of quantification [LLOQ] = 10 µg/mL and %Bias = 0.4–1.7 µg/mL),¹⁸ and previously described methods were used to analyze RDV and metabolites concentrations.¹⁹ Cohort 1 and 8 participants receiving RDV 100 mg had an estimated SBECD dose of ∼3000 mg. Cohorts 2 through 4 receiving RDV 2.5 mg/kg had an estimated SBECD dose of ∼75 mg/kg. SBECD exposures in this study were compared with 36 adults without COVID-19 with normal renal function (n = 16) or mild (n = 10) or moderate (n = 10) renal impairment.

Sparse PK samples were analyzed using population-PK (Pop-PK) models for RDV and its metabolites GS-704277 and GS-441524¹⁷ (data on file). PK data used to develop the Pop-PK models for RDV and its metabolites were obtained from the following: (1) sparse sampling PK data from the majority of pediatric participants enrolled in CARAVAN (n = 51); (2) intensive sampling PK data from 3 phase 1 healthy volunteer studies: GS-US-399-1812 (n = 78), GS-US-399-1954 (n = 16), and GS-US-399-5505 (n = 29); and (3) PK data from 2 phase 3 studies in adult COVID-19: sparse/intensive sampling from GS-US-540-9012, a randomized, double-blind, placebo-controlled study that evaluated RDV in adults with COVID-19 in the outpatient setting (n = 148), and sparse sampling from CO-US-540-5844, a randomized, double-blind study that compared the combination of RDV and tocilizumab to RDV alone in hospitalized adults with severe COVID-19 pneumonia (n = 289). Pediatric exposures were compared with adult phase 3 exposures, and geometric mean ratios and 90% confidence intervals were estimated. Correlation of pharmacokinetic exposures with adverse events (7 most common adverse events: constipation, pyrexia, ALT increased, hyperglycemia, hypertension, hypomagnesemia, and acute kidney injury) was assessed. The exposure-response relationship was explored for recovery by day 10 as well as last available assessment based on the 7-point ordinal scale; the PEWS scale (behavior, cardiovascular, respiratory domains) on day 10; and time to discharge alive from hospital. The PK-pharmacodynamic analysis set included all participants aged 28 days to 17 years with COVID-19 who received at least 1 dose of remdesivir and had evaluable Pop-PK–based PK parameter estimates (n = 50) and the pharmacodynamic measurement available.

For participants weighing ≥12 kg, SARS-CoV-2 serology (IgG, IgM, and IgA) was completed at screening, day 5, and day 10. Samples were collected for SARS-CoV-2 reverse transcriptase quantitative PCR (RT-qPCR) testing and viral resistance testing on days 1, 3, 5, 7, and 10 as applicable (collection described further in the Supplemental Appendix). Viral load was analyzed via the central laboratory, except for fecal and rectal swab data, which are not included in this manuscript because assays are being developed. The serology for SARS-CoV-2 and SARS-CoV-2 samples/swabs were collected on the day of discharge if before day 10. Thereafter, these samples could be collected by study staff at the participant’s home or as an outpatient on designated days.

Resistance testing (details in the Supplemental Appendix) was planned for all participants enrolled in the study who received at least 1 dose of remdesivir. Baseline and postbaseline (during and/or after treatment) samples with SARS-CoV-2 viral load above the LLOQ were sequenced. The whole genome of SARS-CoV-2 was amplified using the ARTIC version 3 primer set at 63 °C annealing temperature, and the nucleotide sequence was determined by next-generation sequencing at DDL Diagnostic Laboratory (Rijswijk, The Netherlands). Amino acid variants and indels were reported as a change from reference when present ≥15%. Sequencing analyses focused on amino acid substitutions in the SARS-CoV-2 nsp12 RdRp (target of RDV) or other parts of the replication-transcription complex (Nsp8, Nsp10, Nsp13, and Nsp14).

Phenotyping of emergent substitutions in Nsp12 was conducted using site-directed mutants (SDMs) of SARS-CoV-2 in a replicon system adapted from Xie et al 2020²⁰ and Zhang et al 2021²¹ (Supplemental Appendix). Briefly, the 4 plasmids that encode SARS-CoV-2 genes for the nonstructural proteins and the nucleoprotein, with or without SDMs, were prepared. The 4 DNA fragments were isolated from SARS-CoV-2 SH01 (SARS-CoV-2/human/CHN/SH01/2020, GenBank MT121215). Huh7-1CN cells were mixed with RNA and subjected to immediate electroporation followed by luciferase assay. EC₅₀ values were calculated using a nonlinear 4-parameter variable slope regression model as the concentration at which there was a 50% decrease in the luciferase reporter signal relative to dimethylsulfoxide vehicle alone (0% virus inhibition) and uninfected control culture (100% virus inhibition). The fold-change values were calculated by dividing the variant mean EC₅₀ by the SH01 reference strain mean EC₅₀.

Clinical status on the 7-point ordinal scale (1 = death and 7 = not hospitalized; Supplemental Appendix) was evaluated at baseline and daily for days 1 through 10 or until hospital discharge. Recovery was defined as an improvement from a baseline score of 2 through 5 to a score of 6 or 7 or an improvement from a baseline score of 6 to a score of 7. To measure clinical improvement, PEWS^22,23 also was recorded at baseline and daily for days 1 through 10 or until hospital discharge (Supplemental Appendix).

From July 2020 to May 2021, 53 participants were enrolled and treated at 19 sites in Italy, Spain, the United States, and the United Kingdom. The median (Q1, Q3) age of participants was 7 (2, 12) years, with a range of 0.1 to 17 years (Table 1). Overall, 98% (52/53) of participants had a preexisting medical condition, with the most common being obesity (19/53, 37%), asthma (11/53, 21%), and cardiac disorders (11/53, 21%). Most participants (51 participants, 96%) received concomitant medications other than those used for treating COVID-19. Thirty-four participants (64%) received dexamethasone during the study.

At baseline/enrollment, participants had a median (Q1, Q3) of 5 days (3, 7) of COVID-19 symptoms and 1 day (1, 3) on line of hospitalization for COVID-19 (Table 2). The most common COVID-19–related disease manifestations were respiratory (44 participants, 83%), followed by gastrointestinal (27 participants, 51%); 12 participants (23%) were receiving invasive mechanical ventilation. Of 21 participants with an abnormal baseline electrocardiogram, all but 3 had results that were considered not clinically significant.

Of the 53 participants who received at least 1 dose of study drug, the median duration of treatment was 5 (4, 8) days with a range of 1 to 10 days; the median for each cohort was also 5 days (Fig 1). Reasons for discontinuation of study drug before 10 days were hospital discharge (22 participants, 42%), investigator discretion (13 participants, 25%), adverse event (2 participants, 4%), parent/guardian decision (2 participants, 4%), and patient decision (1 participant, 2%).

A total of 38 participants (72%) had at least 1 adverse event during the study, and 15 participants (28%) had a Grade 3 or higher adverse event (Table 3). Eleven participants had serious adverse events (Supplemental Table 4); none of which was deemed related to study drug, and all were consistent with either clinical manifestations of COVID-19 or underlying medical conditions. Grade 3 or 4 laboratory abnormalities were reported in 22 participants (42%) overall, including 9 participants (75%) in cohort 1, 2 participants (17%) in cohort 2, 4 participants (33%) in cohort 3, 4 participants (36%) in cohort 4, and 3 participants (60%) in cohort 8 (Supplemental Table 5). The most common grade 3–4 laboratory abnormalities were decreased hemoglobin (n = 9) and decreased estimated glomerular filtration rate (n = 7).

Two participants (4%), both in cohort 1, prematurely discontinued study drug because of adverse events. One participant prematurely discontinued after 5 doses of RDV because of adverse events of grade 3 ALT increase, grade 3 AST increase, grade 4 blood sodium increase, and grade 4 hyperbilirubinemia; this participant had elevated transaminases and hyperbilirubinemia at study entry and died on day 14 as a result of multisystem organ failure. With the exception of blood sodium being increased, the laboratory abnormalities were considered related to study drug by the investigator. One participant, who had elevated ALT and AST at baseline, prematurely discontinued study drug after 3 doses of RDV because of grade 3 ALT increase, which was considered related to study drug by the investigator. None of the adverse events that resulted in premature study drug discontinuation were reported as serious adverse events.

Three treatment-emergent deaths occurred during the study. The cause of death for 1 participant (cohort 1), a 16-year-old female, was reported as multisystem organ failure on day 14; this participant (described previously) discontinued RDV after 5 doses because of elevated transaminases. At baseline, the participant was receiving oxygen through invasive mechanical ventilation. The cause of death for the second participant (cohort 2), who had serious adverse events of hypotension, cardiorespiratory arrest, and respiratory failure, was reported as respiratory failure secondary to removal from support on day 35 of hospitalization/study course. The participant was a 16-year-old female who was receiving oxygen via invasive mechanical ventilation at study baseline. For a third participant (cohort 8), who had serious adverse events of pneumoperitoneum, fatal respiratory distress, hemodynamic instability, gastrointestinal necrosis, cardiac failure, and multiple organ dysfunction syndrome, the cause of death was reported as respiratory, cardiac, and kidney failure as well as acute blood loss from the abdomen (day 18). This participant was an 8-year-old male receiving high-flow oxygen at baseline.

Thirty-two days after the last dose of RDV, a fourth participant, a 13-year-old female, died. The participant (in cohort 1) was receiving high-flow oxygen at baseline and had serious adverse events of Gram-negative septic shock as well as pulmonary hemorrhage after completion of treatment. The cause of death was hypoxemic and hypercarbic respiratory failure secondary to SARS-CoV-2 acute respiratory distress syndrome and secondary cytomegalovirus pneumonitis in the setting of systemic lupus erythematosus and lupus nephritis.

The range of exposures of RDV and its metabolites were generally similar, though with high variability, across all evaluated cohorts (Fig 2; Supplemental Fig 4). No apparent trends in increasing PK exposures with decreasing weight or age were observed for RDV and its metabolites (Supplemental Tables 6 and 7).

PK exposures for RDV and its metabolites were largely overlapping, although modestly increased relative to corresponding PK exposures in adults; the percent increases varied depending on which analyte and which PK parameter, with increases in estimated mean AUC_τ, C_max, and C_τ ranging from 33% to 129% for RDV PK parameters, 37% to 124% for GS-704277 PK parameters, and 0% to 60% for GS-441524 PK parameters (Fig 2). No trends were identified between PK exposures (RDV, GS-704277, and GS-441524) and the 7 most commonly reported adverse events (ie, constipation, acute kidney injury, hyperglycemia, pyrexia, ALT increased, hypertension, and hypomagnesemia), except for increased levels of GS-441524 in participants with acute kidney injury (Supplemental Fig 5). No trends were identified between PK parameters and recovery by day 10 on the 7-point ordinal scale (Supplemental Fig 6) or other efficacy endpoints evaluated in this study (data on file). Also, no trends were observed between common COVID-19–related disease manifestations at baseline such as systemic inflammatory response or circulatory manifestation or baseline oxygen support status (data on file). Altogether these analyses demonstrated that sufficient exposures were achieved in all participants regardless of baseline demographics or disease characteristics.

Among 47 participants, no trends were identified in SBECD exposures across age and weight bands (Supplemental Fig 7). SBECD PK exposures (AUC_0-4h, C_max) after administration of RDV appear similar when comparing pediatric participants with COVID-19 to adults with mild to moderate renal impairment or healthy matched controls. The half-life and clearance parameters were also consistent between the adult and pediatric participants receiving remdesivir.

Confirmed negative SARS-CoV-2 PCR results on day 2 through day 10 were reported for 42% (8/19) participants with nasal/oropharyngeal samples, 21% (6/28) participants with nasopharyngeal/oropharyngeal samples, and 22% (2/9) participants with endotracheal tube aspirates. The mean (SD) SARS-CoV-2 viral load changes from baseline up to day 10 were −1.59 (1.514) log₁₀ copies/mL for nasal/oropharyngeal samples (3/53 participants), −1.59 (1.697) log₁₀ copies/mL for nasopharyngeal/oropharyngeal samples (5/53 participants), and −5.94 log₁₀ copies/mL for endotracheal tube aspirates (1/53 participants).

Among the 23 participants with both baseline and postbaseline sequencing data available, 1 participant had 2 amino acid substitutions in RNA-dependent RNA polymerase Nsp12 (A656A/P and G670G/V) observed at day 3 (Supplemental Table 8). Based on analysis of previous cryo-EM structures of the SARS-CoV-2 polymerase complex, A656P and G670V have no direct interaction with the RNA or the incoming nucleotide. Phenotypic analysis of G670V showed no impact on susceptibility to RDV (≤0.96-fold change in EC₅₀; Supplemental Table 9). For the A656P substitution, transfection of the mutant replicon was attempted twice; however, phenotypic results were unable to be generated because of lack of replication. In this participant, viral load declined rapidly, and days 5, 7, and 10 were negative for SARS-CoV-2 RNA. At baseline, this participant had a clinical status of 3 (hospitalized, on noninvasive ventilation or high-flow oxygen devices) and experienced a ≥2-point improvement and recovery based on ordinal score and was released from the hospital on day 15.

Phenotypic analyses were also conducted on amino acid substitutions observed in other parts of the replication-transcription complex (Nsp8, Nsp10, Nsp13, and Nsp14). Substitutions were observed in Nsp10 and Nsp13 (Nsp10 D64D/Y, T101T/I, Nsp13 R248R/I, S259S/L, V266V/F) in 3 participants (Supplemental Table 8). No substitutions were observed in Nsp8 or Nsp14. The substitutions observed either did not impact RDV susceptibility in vitro (Nsp10 T101I, Nsp13 S259L, and V266F showed ≤0.74-fold change in EC₅₀) or could not be determined because of a lack of replication (Supplemental Table 9).

Recovery was reported for 62% of participants (95% CI, 48–75) on day 10 and 83% of participants (95% CI, 70–92) at the time of the last assessment. The median (Q1, Q3) time to recovery was 7 (5, 16) days. For participants who were discharged alive by day 30, the median (Q1, Q3) duration of hospitalization from day 1 was 7 (5, 12) days. The proportion of participants who were discharged from the hospital was 60% by day 10 and 83% by day 30. Among the participants who were not discharged by day 30, 7 were still hospitalized and 2 had died.

Participants showed improvement in clinical status based on the 7-point ordinal scale, discharge from hospitalization, viral load, oxygen requirement, and PEWS scale with treatment with RDV. For all cohorts, there was a trend in improvement of clinical status on the 7-point ordinal scale over time, as indicated by an increasing score (Fig 3). The proportion of participants with an ordinal score of ≤5 points at baseline who had a ≥2-point improvement in clinical status was 75% on day 10 and 85% at the time of the last assessment. The median (Q1, Q3) time to ≥2-point improvement for participants with an ordinal score of ≤5 points at baseline was 7 (5, 10) days. By day 10 or the time of the last assessment, the proportion of the total population with decreases in the PEWS total score (indicating improvement) was 38% of participants (20/53) by 1 to 3 points; 15% or participants (8/53) by 4 to 6 points; and 6% of participants (3/53) by 7 to 9 points.

We provide results of a phase 2/3 study of RDV treatment of infants and children aged ≥28 days hospitalized for COVID-19 infection. The study population included 98% who had a preexisting medical condition. In this group, RDV had a favorable tolerability profile with no new apparent safety trends. Pharmacokinetic exposures of RDV in this population of children were similar to those reported in adults. Participants had a high rate (85%) of clinical improvement.

Safety findings were consistent with participants having COVID-19 and/or underlying medical conditions. The most common adverse events were constipation (17%) and acute kidney injury (11%). In clinical studies of RDV in adults, the most common adverse event has been nausea. In clinical studies in healthy adults, no evidence of nephrotoxicity has been observed with single doses of RDV up to 225 mg or multiple once-daily doses of RDV 150 mg for up to 14 days (studies GS-US-399-1812 and GS-US-399-1954). The amount of SBECD administered to children was well within established safe doses for existing commercial products. Based on a review by the European Medicines Agency,²⁴ doses up to 250 mg/kg/day of SBECD are considered safe for children aged >2 years, and no significant toxicity has been seen in a small number of neonates treated with SBECD-containing products corresponding with up to 336 mg/kg/day for 18 to 24 days.

Given the high rates of recovery and absence of new safety concerns observed, the modest increases in RDV exposure observed in hospitalized pediatric participants were not deemed clinically relevant and support up to 10 days of dosing with an adult regimen (200 mg intravenous loading dose on day 1 followed by 100 mg intravenous maintenance doses) in children and adolescents weighing ≥40 kg. Weight-based dosing (5 mg/kg intravenous loading dose on day 1 followed by 2.5 mg/kg intravenous maintenance doses) is supported in children ≥28 days of age weighing ≥3 kg to <40 kg. No substitutions associated with resistance to RDV were observed in this study, supporting a high barrier to RDV resistance development in COVID-19 patients.

This was a single-arm study with small numbers of participants in each cohort, and comparisons between cohorts and extrapolation to larger populations are limited. The preexisting medical histories may also limit generalizability, although they are representative of populations at risk for severe COVID-19 outcomes.

RDV is the only approved COVID-19 treatment in children. This open-label study supports its use in the pediatric population.

The authors thank Leila Bianchi, Elizabeth Lang, and all site investigators and site staff for their contributions to the study. They also thank the participants and their families. Jennifer King, PhD, of August Editorial provided writing assistance. Deqing Xiao and Shalini Saxena supported sample bioanalysis. Eileen Kim and Luzelena Caro critically reviewed the manuscript. Sean Regan and Emma Hughes helped prepare figures.

ALT

alanine aminotransferase

AST

aspartate aminotransferase

CARAVAN

Clinical Administration of Remdesivir After COVID-19 Diagnosis in Children Study

LLOQ

lower limit of quantification

PCR

polymerase chain reaction

PEWS

Pediatric Early Warning Score

Pop-PK

population-pharmacokinetics

PK

pharmacokinetics

RDV

remdesivir

RT-qPCR

reverse transcription quantitative polymerase chain reaction

SBECD

sulfobutylether β-cyclodextrin

SDM

site-directed mutant