1. RESULTS
  2. Assessment of novel PA/I38X substitutions detected in clinical trials
Resistance monitoring in phase 2 (T0821) and pediatric [T0822 (Japic CTI-163417)] trials revealed treatment-emergent I38T/F/M substitutions in PA which confer reduced susceptibility to BXA (Table 1 and Supplementary Table 1).10 In order to assess treatment-emergent AA substitutions associated with reduced susceptibility to BXA in phase 3 [T0831, CAPSTONE-1 (NCT02954354)] and [T0832, CAPSTONE-2 (NCT02949011)], Sanger sequencing was conducted with paired pre- and post-treatment swab samples from BXM-treated subjects. While in the CAPSTONE-1 study only I38T/M substitutions were detected, PA/I38N-substituted A(H1N1) viruses were newly identified from one BXM-treated patient in the CAPSTONE-2 study. Additionally, PA/I38T, I38T/I, and I38M substituted A(H3N2) viruses were detected from 10, 2 and 1 BXM-treated subject, respectively, and PA/I38T substituted type B viruses were detected from 1 BXM-treated patient. Moreover, PA/I38R, a single nucleotide substitution at isoleucine 38 (ATA to AGA), was temporary detected by next generation sequencing analysis at day 5 with 7.6% of proportion in the virus population among one of the 10 subjects with PA/I38T-substituted A(H3N2) viruses . Finally, PA/I38S-substituted A(H1N1) viruses and polymorphic PA/I38V and PA/I38L were reported in the literature during 2018/19 influenza season.18 25,26
In order to assess the impact of these detected I38 substitutions on BXA susceptibility, the recombinant A(H1N1) and A(H3N2) viruses harboring the individual substitution were generated and subjected to susceptibility testing. Plaque reduction assay revealed that the A(H1N1) viruses with polymorphic I38V and L substitutions displayed reduced BXA susceptibility by 2-fold and 6-fold, respectively (Table 1), consistent with a previous report.25 The replicative capacities of the A(H1N1) and A(H3N2) viruses with I38V and L substitutions were comparable to those of the wild-type viruses in canine MDCK and human RPMI2650 cells (Figure 1 and Supplementary Figure 1). Viruses bearing the I38N and I38S substitutions showed reduced BXA susceptibility by 24-fold and 12-fold, respectively (Table 1), but also, the recombinant A(H1N1) and A(H3N2) viruses with I38N and I38S substitutions exhibited significant fitness cost in MDCK and RPMI2650 cells (Figure 1 and Supplementary Figure 1). The I38R virus could not be obtained by reverse genetics, suggesting I38R conferred severe growth defect to the virus.
Assessment of non-I38 PA substitutions and PB1/2 substitutions detected in clinical trials
Impact of PA substitutions at other positions than I38 PA on BXA susceptibility was also assessed. Results for AA substitutions identified in T0821 and T0822 clinical trials were previously reported.10 None of the newly tested AA substitutions identified in the clinical trials T0831 and T0832 did significantly impact BXA susceptibility (< 3-fold change in EC50) (Supplementary Table 1).
Treatment-emergent AA substitutions in PB1 and PB2 subunits were analyzed in T0821, T0822 and T0831 clinical trials. Sanger sequencing was conducted with all paired pre- and post-treatment swab samples from BXM-treated subjects in studies T0821 and T0822. In study T0831, sequencing of the PB1 and PB2 genes was performed on samples from BXM-treated patients not exhibiting a treatment-emergent substitution at position 38 in the PA gene and identified as non-responders based on the following criteria: i) virus rebound (virus titer rise of ≥0.6 log10 TCID50/mL between consecutive time points), or ii) continued virus shedding (virus titer >1.5 log10 TCID50/mL at day 5 and beyond), or iii) no reduction in virus titer (no change or rise in virus titer between consecutive time points). All detected PB1/2 substitutions were then subjected to susceptibility testing using the plaque reduction assay. None of the tested AA substitutions in PB1/2 did significantly impact BXA susceptibility (EC50 fold change ranged from 0.53 to 1.70) (Supplementary Table 1).
Assessment of PA and PB2 substitutions identified in extended analyses of clinical trial data
AA substitutions at positions associated with baloxavir resistance were identified from NCBI database influenza sequences and from extended analyses of virologic data from clinical trials (Supplementary Table 2). Substitutions potentially associated with a reduced virologic response (defined as a significantly reduced change from baseline on day 2 in virus titer relative to the virus type/subtype subset distribution), virus rebound, or elevated (≥90 percentile) baseline EC50 values of virus isolated from clinical specimens were assessed for their impact on BXA susceptibility using recombinant viruses and the ViraDot assay. None of the identified and tested 21 AA substitutions in PA and PB2 significantly affected BXA susceptibility (< 1.5-fold change by means of EC50 values) (Supplementary Table 2).
DISCUSSION
In this study, we characterized PA/I38 substitutions detected in clinical studies (I38T/F/M/N/R/S) and as naturally occurring polymorphisms (I38V/L). PA/I38N was newly identified in the clinical setting and we demonstrated that BXA susceptibility of I38N viruses was reduced compared to wild-type virus (24-fold for A(H1N1) and 10-fold for A(H3N2)). In addition, the replicative capacity of I38N viruses was reduced to a comparable level to I38T viruses. The genetic barrier to the development of reduced susceptibility is often defined as the number of nucleotide changes required for the AA change. All detected PA/I38 substitutions can develop through a single nucleotide change, suggesting that the introduction of two nucleotide changes may make it difficult for other I38-substituted viruses to appear.
An arbitrary 3-fold threshold has been recently used in surveillance screening to define reduced BXA susceptibility.19Although we have comprehensively tested individual AA substitutions in PA, PB1 and PB2 subunits in viral RNA polymerase complex, only known AA positions were detected as substitutions that confer reduced susceptibility by more than 3-fold change in EC50. The body of data supports that I38 substitutions are the major pathway for reduced BXA susceptibility. Rare changes at E23, A37 and E199, found with 0.07% to 0.44% frequency in clinical treatment trials, should be monitored as non-I38 substitutions. However, the cut-off value at 3-fold can exceed dependent on robustness of assay systems, and therefore standardization of susceptibility testing with BXA may be important.
Since AA substitutions in PA protein combined with I38 substitutions may affect functional compensation for fitness cost, we further investigated whether AA substitutions associated with I38 substitutions identified in the clinical trial were compensatory mutations. The replicative capacity was previously evaluated for substitutions A20S+I38F and I38T+E623K,10 and I38T+S60P and I38T+I201T were tested in this study (Supplementary Figure 2). However, these substitutions were unable to restore the growth impairment of the I38T-subsituted viruses. Given that I38-substituted viruses show different patterns in terms of replicative capacity dependent on the type/subtype or isolated year,10 18 2527 different substitutions are likely to be involved in restoration of the replicative capacity of the I38 mutant viruses. In addition, currently circulating strains may have a different genetic backround compared to the recombinant viruses used in this study. While in vitro results of the replicative capacity of I38X mutant viruses from different studies vary, further investigation of potential compensatory mutations that could recover the fitness cost of I38T substitution will be needed.
It is important for disease management to understand the risk of treatment-emergent resistance to BXA, and therefore continuous surveillance and exploration of mutations that affect BXA susceptibility and viral fitness are important. This study provides the characteristics of clinically identified I38-substituted viruses, and extensive information on the impacts of further non-I38 AA substitutions in the trimeric viral polymerase complex to BXA susceptibility. This additional information will further support influenza surveillance.