Uncovering the spatiotemporal patterns of the largest vaccine-derived poliovirus serotype 2 outbreak in the Democratic Republic of the Congo

Vaccine-derived poliovirus serotype 2 (VDPV2) outbreaks are a major challenge to polio eradication, seeded by rare reversions of oral polio vaccine to a neurovirulent strain. The Democratic Republic of the Congo has experienced more VDPV2 outbreaks than any other country; however, little is known about the subnational dynamics of VDPV2 spread, and understanding of the quality of VDPV2 surveillance and outbreak response in the country is incomplete. We used phylodynamic and phylogeographic methods to analyse 116 VP1 sequences from cases of poliomyelitis associated with RDC-MAN-3, the largest VDPV2 outbreak to date in the DRC with cases reported between October 2021 and August 2023, alongside programmatic data from the RDC-MAN-3 outbreak. We identified a pair of vaccination campaigns in Tshopo in October 2020 as the most plausible seeding event, making the delay between seeding and first discovery by environmental sampling approximately 12 months. Phylogeographic reconstruction showed predominantly local spread, with a median of 74% (20/27) of inferred cross-province movements occurring between neighbouring provinces. and identified pairs of provinces with the most important contributions to spatial dispersal of the outbreak; in particular, transmission from Tanganyika to Haut Lomami was highly frequent (9/27 cross-province movements). We inferred from the effective population size trajectory that surveillance was likely less intensive in the first wave of the outbreak, and that in some provinces RDC-MAN-3 was introduced at least two months before detection, while in Équateur province RDC-MAN-3 was never detected by culture testing. Finally, our analysis found that outbreak response vaccination campaigns were conducted 2-13 months (modal average across provinces of 5 months) after the estimated importation of poliovirus into each province. These results show how campaign scope and timing were insufficient to stop the outbreak early and highlight how routine genomic data can retrospectively assess the adequacy of surveillance and response implementation.