Fig. 6: Antibodies cross-reactivity of neuraminidase between A/Aichi/2/1968 (H3N2), A/Singapore/1/1957 (H2N2) and avian A(H9N2).

a The NI antibody profile against the NA of A/Aichi/2/1968 (H3N2) (AI68) and avian A(H9N2), and the HI antibody profile against AI68, as determined for Fig. 2, when stratified by year of birth. The age groups depicted in Fig. 2 are indicated on top of the graph. b HI-titer against AI68 in individuals born in between 1957 and 1968 (n = 11) and before 1957 (n = 20), as determined for Fig. 2. c The NI antibody profile against AI68 and avian A(H9N2) NA from CARES cohort samples, n = 43 individuals. d–g NI-titer against NA of A/Singapore/1/1957 (H2N2) (SG57), AI68 and two representative avian N2 from A/guinea fowl/Hong Kong/WF10/1999 (H9N2) (HK99), A/Pakistan/486/2015 (H9N2) (PA15) in the EPI-HK cohort. Analyzes were stratified to individuals after 1968 (n = 47), between 1957 to 1968 (n = 13), and individuals born before 1957 (n = 45). h The NI antibody profile of the EPI-HK participants against human N2; SG57 and AI68, and avian N2; HK99 and PA15, when stratified by year of birth. i Correlation of NI titers between SG57 N2 and the different neuraminidase subtypes in n = 58 individuals born in or before 1968 from EPI-HK cohort. Dashed lines indicate the 1957 A(H2N2) and 1968 A(H3N2) pandemics. Dotted lines indicate limits of detection. The bar graphs indicate the geometric mean antibody titer with 95% confidence intervals. Statistical significance in b was analyzed using two-sided unpaired t-test. Statistical significance between different age groups in d–g was analyzed using two-sided one-way ANOVA, adjusted with Tukey’s multiple comparisons test (***p = 0.0008 for 1957-1968 vs before 1957, ****p < 0.0001 for after 1968 vs 1957-1968 or 1968 vs before 1957 in (d); ****p < 0.0001 for after 1968 vs 1957-1968 or 1968 vs before 1957 in e, f or g). Correlation in i was reported by Spearman’s correlation for each comparison using two-sided test and visualized using the ‘corrplot’ package (R version 0.92), and p-values were adjusted by controlling for the False Discovery Rate using the Benjamini-Hochberg method (***p < 0.001 for each pairwise comparison).