Baloxavir marboxil is used for influenza treatment in Japan. However, the clinical and virologic significance of baloxavir-associated PA variants in children remains unclear. We conducted a four-season, multicenter study from 2018-2019 to 2023-2024 in pediatric patients under 19 years old with influenza A who were treated with baloxavir. Clinical data were collected at the first visit; diary cards were maintained for up to 8 days. Clinical specimens were collected twice, with a median interval of 4 days. PA substitutions were screened by real-time RT-PCR and confirmed by next-generation sequencing and Sanger methods. Among 247 patients, 24 (9.7%) harbored PA variants, one pre-treatment and 23 post-treatment. PA/I38T was the most common substitution. A(H3N2) patients had a higher frequency of PA variants than A(H1N1)pdm09 (13.8% vs. 5.1%). Although PA variants were associated with slower viral RNA decline, no significant differences were observed in fever duration (median 22.6 vs. 15.6 h) or symptom duration (median 105.0 vs. 113.3 h) compared to wild-type virus infections. PA variants, especially PA/I38T, emerged more frequently in A(H3N2) infections but had minimal clinical impact. The rarity of pre-treatment variants supports limited community transmission. Continued surveillance is warranted, particularly in younger children.

We characterized influenza A(H1N1)pdm09, A(H3N2), and B/Victoria viruses circulating in Japan during 2023–2024, focusing on lineage placement relative to WHO-recommended vaccine strains and on baloxavir resistance (PA/I38T substitutions). We enrolled 210 outpatients with influenza-like illness across eight clinics in six prefectures (October 2023–September 2024). Of these, 209 had an analyzable pre-treatment respiratory specimen for RT-PCR; hemagglutinin (HA) and neuraminidase (NA) genes were sequenced by next-generation sequencing (NGS). PA/I38T substitutions that confer baloxavir resistance were assessed by cycling-probe RT-PCR, Sanger sequencing, and NGS. HA phylogenies were constructed with global datasets and WHO vaccine reference strains. Of 209 pre-treatment specimens, 181 were influenza-positive (A(H1N1)pdm09 44.2%, A(H3N2) 37.6%, B/Victoria 18.2%); 51 follow-up specimens were collected ≈4–5 days after baloxavir or neuraminidase inhibitor therapy. HA phylogeny placed A(H1N1)pdm09 in clades 5a.2a/5a.2a.1 with predominance of subclade D.2. A(H3N2) clustered exclusively in clade 2a.3a.1 (J lineage, mostly J.1), indicating a mismatch with the season’s A/Darwin/9/2021 vaccine component and supporting the subsequent J-lineage update. All B/Victoria genomes fell within V1A.3a.2 on a C.5 backbone (C.5.1 and C.5.7). No PA/I38T variant was detected in any pre-treatment specimen. Post-baloxavir, PA/I38T emerged in one A(H3N2) case (confirmed by all three methods) and in one B/Victoria case detected by NGS only (minority variant in a low-load sample). NA genes showed no substitutions associated with reduced susceptibility to laninamivir (e.g., E119A, G147E). During 2023–2024, A(H1N1)pdm09 and B/Victoria remained genetically aligned with their vaccine components, whereas A(H3N2) shifted to the J lineage, consistent with the 2024–2025 vaccine update. Although pre-treatment PA/I38T was absent, low-frequency on-therapy selection was observed, including a rare PA/I38T in influenza B/Victoria detected by NGS, suggesting the value of deep sequencing when viral loads are low. These integrated genomic–clinical data support vaccine strain realignment for H3N2 and continued monitoring of baloxavir resistance in outpatient care.