The active site of the PA protein (P3 in C and D viruses) is nearly identical in all 4 influenza virus types (1,8). Therefore, we hypothesized that baloxavir would inhibit replication of not only influenza A and B viruses but also influenza C and D viruses. First, we tested 2 viruses of each type by using a virus yield reduction assay. We used baloxavir acid (baloxavir) in experiments and included favipiravir as a control.

Baloxavir broadly inhibited virus replication of all 4 types (Table 1). On the basis of 90% effective concentration values determined at 48 hours postinfection, influenza A viruses were most susceptible to baloxavir and influenza D viruses least susceptible. Baloxavir susceptibility for influenza B viruses was ≈3-fold lower and that for influenza C viruses was ≈6-fold lower than that for influenza A viruses. Analysis of 34 P3 sequences of influenza D virus and 221 of influenza C virus (retrieved from GISAID, https://www.gisaid.org, and GenBank) showed that all influenza D viruses have valine at position 38, whereas influenza C viruses have isoleucine, similar to most influenza A and B viruses. Nevertheless, valine at 38 in influenza A and B viruses had little or no effect (<3-fold) on baloxavir susceptibility (10–12). Favipiravir also showed inhibitory effects against all virus types, although much higher concentrations were required to achieve similar levels of reduction (Table 1).

Although the virus yield reduction assay has been used to assess baloxavir susceptibility of seasonal and avian viruses (10,13), other phenotypic assays, such as the focus-reduction assay (FRA) and the high-content imaging neutralization test (HINT), offer an improved throughput (12,14,15). Regardless of the assay used, baloxavir effective concentrations for influenza A viruses were similar (≈0.1–3 nmol/L) (10,12–15). Unlike the FRA, HINT relies on single-cycle virus replication, which is achieved by withdrawing trypsin needed to activate infectivity of progeny virus. HINT eliminates variance caused by different replication kinetics. However, the FRA is optimal for testing highly pathogenic avian viruses because multicycle replication of these viruses is trypsin independent. We used 2 seasonal A(H1N1)pdm09 viruses, one of which contains the naturally occurring substitution PA-I38L, for reference purposes (12) (Table 2).

First, we tested 25 influenza viruses of avian origin, representing H5, H6, H7, H9, and H10 subtypes, by using FRA or HINT as described (12) (Table 2; Appendix Table 1). Most viruses were isolated from infected humans. Most viruses had markers of M2 resistance and some had NAI-resistance markers. Data showed that these diverse viruses were susceptible to baloxavir and had 50% effective concentration (EC₅₀) values in a low nanomolar range (Table 2; Appendix Table 1). In the FRA, favipiravir EC₅₀ values were much higher than those for baloxavir (Appendix Table 1). However, favipiravir did not produce a measurable antiviral effect by HINT because this drug requires several hours for activation in cells. Baloxavir susceptibility of 30 swine-origin viruses, representing different lineages and subtypes and collected over many years, demonstrated HINT EC₅₀ values comparable to avian and seasonal influenza A viruses (Table 2; Appendix Table 2) (10,12,13).

It is prudent to analyze PA sequences of emerging influenza A viruses for markers previously associated with reduced baloxavir susceptibility (11,12). Among swine-origin viruses available for testing in this study, polymorphism PA-38I/M was detected in A/Iowa/33/2017 (H1N1)v. Virus populations with either PA-I38 or PA-I38M were recovered by biologic cloning and tested by using HINT. Substitution PA-I38M conferred 12-fold reduced baloxavir susceptibility, consistent with previous reports for PA-I38M–containing H3N2 viruses (11,12). Analysis of PA sequences from 2,485 H7N9 viruses (from GISAID and GenBank) showed 1 virus with PA-I38M, 2 with PA-E199G, and 1 with PA-A36V (11,12). The effect of these substitutions on baloxavir susceptibility for H7N9 viruses is currently unknown. Moreover, PA sequence of 1 swine influenza A virus showed PA-I38T, a marker associated with clinically relevant baloxavir resistance (11). None of these viruses were available for phenotypic testing.

Baloxavir displayed broad antiviral activity against diverse influenza viruses, including all 4 types and animal-origin influenza A viruses with pandemic potential. Our findings suggest that baloxavir might offer the first therapeutic option against influenza C virus infections. Further studies are needed to provide comprehensive assessment of baloxavir susceptibility by using a large panel of representative influenza C viruses. Ongoing monitoring of baloxavir susceptibility of emerging avian and swine influenza A viruses with pandemic potential is needed to inform clinical management and public health preparedness efforts.