Genomic epidemiology of the 2025 mpox epidemic in Sierra Leone

Categories: Disease & Virus

May 12, 2026

New G.1 lineage drives the outbreak in SLE

To investigate the evolutionary relationship between SLE’s epidemic lineage and global MPXV diversity, we generated 338 high-quality MPXV genomes from cases across 14 districts between 10 January and 3 August 2025 (Fig. 1a,b), produced through collaboration among the Central Public Health Reference Laboratory (CPHRL, n = 128), the Kenema Government Hospital (KGH) Viral Hemorrhagic Fever (VHF) Laboratory (n = 105) and the Institut Pasteur de Dakar (IPD) mobile laboratory in Port Loko (n = 105). With two additional genomes from the Beijing Institute of Technology³⁰, the total was 340 high-quality genomes (>60% coverage at 25–8,100× depth), representing 6.9% of all PCR-confirmed cases up to August 2025 (Fig. 1a,b).

**Fig. 1: Phylogenetic origin and emergence of mpox lineage G.1 in SLE.**

We reconstructed the clade IIb phylogeny to determine the relationship between the SLE sequences and the clade IIb/sh2017 lineage. In our phylogeny, 339 of 340 genomes clustered within clade IIb/sh2017, which emerged in humans in southern Nigeria in 2014 and drives the ongoing human epidemic in West Africa^7,8. According to the nomenclature proposed in refs. ^3,4, clade IIb/sh2017 is designated as lineage A (or clade IIb/sh2017/lineage A), with direct descendants designated as, for example, A.1 and subsequent subdivisions as, for example, A.1.1—analogous to the Pango nomenclature used for SARS-CoV-2³¹. Within lineage A, our sequences form a well-supported monophyletic group (100% bootstrap support) descended from lineage A.2.2 (Fig. 1c). In accordance with the nomenclature, we designated the new SLE lineage as G.1, the alias of A.2.2.1, or clade IIb/sh2017/lineage G.1 (Fig. 1c).

The enrichment—rather than the mere presence—of APOBEC3-context mutations distinguishes sustained human transmission from zoonotic spillover⁸. To test whether G.1 fits this profile, we quantified mutational biases by reconstructing ancestral SNPs across the G.1 phylogeny. Approximately 85% (90 of 106) of reconstructed SNPs were consistent with APOBEC3 editing (TC → TT or GA → AA transitions driven by APOBEC3F³²), providing strong genomic evidence that G.1 arose through sustained human transmission rather than repeated zoonotic introductions (Fig. 1c and Extended Data Fig. 1).

Historical circulation of clade IIa in West Africa

We identified one nonclade IIb genome: a clade IIa sequence sampled in mid-January 2025 in WAU. Phylogenetically, it clusters with two clade IIa genomes from Guinea (August and December 2024), forming a sister clade to the 1965 Rotterdam zoo epizootic and near-contemporary museum orangutan sequences^33,34 (Fig. 2). The broader clade encompassing these groups shares a deep common ancestor with the lineage containing the 1958 Copenhagen captive-monkey sequence³⁵, reinforcing that clade IIa has historically circulated in West Africa. APOBEC3 substitutions were scarce across all three sequences (≤2 per genome), in sharp contrast to the heavily enriched G.1 lineage (Fig. 2 and Extended Data Fig. 1), supporting zoonotic spillover rather than sustained human transmission. Whether this case represents importation from Guinea or an independent zoonotic event in SLE requires additional sampling to resolve.

**Fig. 2: Clade IIa phylogeny with reconstructed SNPs mapped on to branches.**

G.1 descended from lineages circulating in West Africa

The geographical origin of the G.1 lineage remains unresolved. It is unclear whether it arose directly from endemic clade IIb/sh2017 sublineages in Nigeria or was introduced into SLE via an intermediate source. In our phylogenetic analyses, G.1 is nested within lineage A.2.2, which is primarily sampled from the USA (Fig. 1c). G.1 is separated from its closest relative, a sequence from Togo, by nine APOBEC-like mutations and two non-APOBEC3-like mutations along its stem branch. Previous studies have estimated that APOBEC3-like mutations accumulate at approximately 6 per year (95% confidence interval 5–7), a rate that is consistent across MPXV clades and lineages with sustained human transmission^7,8,12. The number of APOBEC3-like mutations along the G.1 stem branch suggests that it diverged from its ancestor, the A.2.2 sublineage, approximately 18 months ago. This provides a lower bound on the timing of the lineage’s introduction into SLE.

The 11 A.2.2 genomes from the USA were sampled between June 2024 and June 2025 from Illinois (2), California, Massachusetts (2), Georgia, Pennsylvania, Tennessee, Minnesota, Michigan and Virginia, with confirmed travel history to Nigeria for at least 4 of the sequences (Fig. 1c). Based on the long internal branches separating the US A.2.2 genomes, they most likely represent independent viral imports from the Nigerian A.2.2 lineage rather than an established lineage that has been cryptically diversifying in the USA. The closest publicly available Nigerian A.2.2 sequence to the A.2.2 lineage from which the US and G.1 sequences descend is PP853012, sampled in Rivers State in September 2022. Taken together, these findings suggest that lineage A.2.2 in the USA and the G.1 lineage in SLE both descend from the clade IIb/sh2017 lineage A.2.2, which originated in Nigeria and continues to seed localized epidemics outside the country. However, due to limited sampling in the region and the divergence time inferred from APOBEC3 mutations along the G.1 stem branch, we cannot rule out the possibility of viral importation from an intermediate location.

Our phylogeny also indicates G.1 export from SLE (Fig. 1c). This evidence includes four sequences sampled in the USA between March and June 2025, three of which have confirmed travel histories to SLE. We also identified two sequences from Germany sampled from a single patient in March 2025 and two sequences from Guinea sampled in June 2025, with travel histories to SLE confirmed for all but the Guinea sequences. The detection of multiple viral export events indicates active local transmission in SLE at a prevalence sufficient to generate repeated international spread.

G.1 emerged in late September 2024

The timescale inferred from accumulated APOBEC3-like mutations along the G.1 stem branch suggests that the lineage circulated undetected in SLE or in an unsampled external location for an extended period of time before detection. To estimate the timing of G.1’s emergence, we used Bayesian phylogenetic reconstructions in BEAST^36,37 with a nested exponential model. Under this model, we applied an exponential growth model to the G.1 lineage while allowing the rest of the tree to evolve under an independent exponential growth model, capturing the distinct epidemiological dynamics observed in SLE.

We estimated that the time to the most recent common ancestor (tMRCA) of the G.1 lineage was 27 September 2024 (95% highest posterior density (HPD) 12 August 2024 to 11 November 2024) (Fig. 2a). This tMRCA represents the lower bound when the sampled G.1 lineage became established in SLE. This suggests that the G.1 lineage may have circulated for approximately 3 months (95% HPD 2–5 months) before its detection in early January 2025.

The inferred growth rate corresponds to a doubling time of approximately 3 weeks (95% HPD 2.4–3.9 weeks) (Fig. 3b,c), consistent with the sharp increase in incidence observed between late April and May 2025¹⁸. To confirm these observations, we estimated the time-varying reproductive number (R_t) from available case counts and obtained an R_t of 4.4 (95% confidence interval (CI): 3.1–5.9) in early March, consistent with both epidemiological data and the inferred doubling time (Fig. 1a)³⁸. We also applied JUNIPER³⁹ to infer the transmission dynamics of the outbreak. JUNIPER estimated an outbreak start date of 11 September 2024 (95% HPD 2 June 2025 to 11 November 2024), consistent with BEAST results, and an overall reproductive number of the outbreak, including the later decline, of 1.1 (95% HPD 1.1–1.2). In contrast, the estimated growth rate for the Nigerian epidemic, represented by the remainder of lineage A from which G.1 descended (Fig. 3a), corresponded to a doubling time of approximately 2.6 years (95% HPD 1.9–3.4 years)^7,40, indicative of a low-level endemic transmission (Extended Data Fig. 2). Our phylogenetic reconstruction under a nonparametric, Skygrid, coalescent model also supports a rapid and sustained increase in the effective population size of the G.1 lineage between April and July 2025 (Fig. 3b). We additionally used the size of clusters of identical sequences to infer a similar reproductive number (1.2, 95% CI 1.1–1.3) and an overdispersion parameter of 0.4 (95% CI 0.2–1.0) (Extended Data Fig. 3), consistent with previous reports of mpox transmission⁹.

**Fig. 3: Timing, growth dynamics and establishment of G.1 within SLE.**

Notably, JUNIPER estimated that the sampling proportion (the fraction of cases sequenced) was 3.3% (95% HPD 2.3–4.8%). This corresponds to an inferred total epidemic size of approximately 10,400 cases (95% HPD 7,000–15,200), compared with 5,096 confirmed cases (Fig. 3d). In addition, we estimated that the tMRCA between the G.1 lineage and the closest A.2.2.1 relative from Togo (that is, the stem branch age) was 9 August 2023 (95% HPD 3 April 2023 to 19 December 2023) (Fig. 3a). This represents an upper bound on the time of the viral introduction into SLE. However, the rapid exponential growth observed and the absence of closely related A.2.2 samples from endemic Nigeria suggest that the introduction most likely occurred shortly before the tMRCA of G.1 in late September 2024.

G.1 was established in the WAU district

The WAU and WAR districts are the epicenter of the outbreak, accounting for approximately 80% of the mpox cases in SLE. These districts include the capital, Freetown (~1.4 million of the country’s 8.6 million residents; Fig. 1b)¹⁸. However, it remains unclear whether the epidemic originated in this region or whether there was substantial underascertained transmission elsewhere during the early phase of the outbreak. To address these uncertainties, we performed discrete and continuous phylogeographical analyses at the district level to characterize the spatiotemporal spread of lineage G.1 within SLE.

Our discrete phylogeographical reconstructions indicate that lineage G.1 became firmly established in the WAU region, with strong posterior support (posterior probability = 0.995; Fig. 4a), consistent with the index case being reported from this district. The long stem branch implies potentially ~3 months of undetected local circulation before identification, possibly reflecting cryptic circulation in a neighboring country or unsampled transmission within SLE. We cannot confirm the site of G.1’s initial emergence due to undersampling within the country and across West Africa. Nevertheless, G.1 most likely became established in WAU by late September 2024, with the district’s dense, mobile population driving nationwide spread (Fig. 4a).

**Fig. 4: Spatiotemporal dynamics of mpox lineage G.1 in SLE.**

To account for uncertainty from sparse sampling, we applied a Markov jump-counting approach^41,42 to reconstruct the history of location changes along phylogenetic branches, allowing us to estimate the timing and origin of geographical transmission chains^41,42, defined as continuous periods of viral lineage circulation within a specific district after an introduction. The WAU was the primary source of interdistrict dissemination throughout the outbreak, with an estimated 71 introductions (95% HPD 56–88) originating from the district, equivalent to ~117 exports per million residents (HPD 92–144), far exceeding any other district (Fig. 4b).

Most early viral exports originated in the WAU region and spread into Kenema, then to Bo, Port Loko, WAR and Bonthe (Fig. 4b,c). Although WAU remained the principal source throughout the epidemic, Kenema became an important secondary hub later in the outbreak, accounting for an estimated 10 exports (95% HPD 3–21), equivalent to ~13 per million (HPD 4–27 per million; population 772,472), with exports predominantly disseminating to Bo, Pujehun, Bonthe and Falaba (Fig. 4c). The remaining districts contributed only marginally to viral dissemination. Continuous phylogeographical analysis was consistent with the discrete analyses, indicating that G.1 was first established in the WAU region before spreading early into WAR, Kenema and Bonthe (Fig. 4d).

In our dataset, Port Loko was disproportionately well sequenced relative to other regions, including the WAU, with 21% of confirmed cases sequenced compared to 5% in the WAU. To account for this heterogeneity in sampling across districts, we repeated the discrete phylogeographical analysis at the regional level. The regional reconstructions were consistent with the district-level results, again indicating that G.1 most likely emerged or was first established in the western region (posterior probability = 0.99) (Extended Data Fig. 4).

Our regional analyses also supported the patterns of within-country, spatiotemporal spread observed at the district level. We found that the western region was the principal source of inter-regional viral introductions, with an estimated 52 introductions (95% HPD 39–67), equivalent to ~41 introductions per million residents (95% HPD 31–53 per million). The earliest introductions originated in the western region and disseminated to the eastern, southern and north-west regions (Extended Data Figs. 5 and 6). As the epidemic progressed, the introduction profile shifted, with subsequent spread into other regions increasingly driven by viral exports from the western, eastern and southern regions (Extended Data Fig. 6). The northern regions contributed only marginally to interdistrict spread, with transmission there primarily seeded by introductions from the western region (Extended Data Fig. 6).

Persistence in the WAU drove the outbreak

As shown above, repeated introductions from a limited number of districts drove the spread of lineage G.1 across SLE. However, the extent to which locally persistent transmission chains sustained ongoing transmission within districts remained unclear. To disentangle the relative contributions of local viral persistence versus new introductions, we quantified the duration of within-district viral persistence for each reconstructed transmission chain (Figs. 4a and 5a).

**Fig. 5: Persistence and local transmission dynamics of G.1 across SLE.**

We found evidence of persistent within-district circulation in the WAU district from the time of emergence onward (Fig. 5a). A large transmission chain originating in this district persisted continuously throughout the sampling period (Fig. 4a). We also observed an early introduction from the WAU into Port Loko in November 2024, with the resulting transmission chain sustaining local circulation throughout the epidemic, detectable despite sparse sampling (Fig. 5a). This pattern is consistent with the index case’s recent travel history to Lungi, located in the Port Loko district and home to the country’s international airport^30,43. By the time that the first cases were detected in early January 2025, active transmission chains had already been established in four districts: WAU, Port Loko, Bo and Kenema (Fig. 4a). Several districts, including Bo, Bombali and Kenema, harbored transmission chains seeded from WAU in early 2025 that persisted locally for months (Fig. 5a,b).

From March 2025 onward, co-circulating transmission chains increased sharply, consistent with the rise in incidence between late April and May 2025 (Fig. 5a). These chains were predominantly seeded from WAU, with repeated introductions establishing multiple co-circulating but shorter-lived chains in Kenema, WAR and Bonthe. The strong negative correlation between chain duration and epidemic phase (Pearson’s r = −0.93; Fig. 5b) reflects the impact of progressive public health interventions, including vaccination campaigns that began 26 March in WAU and WAR districts, expanded nationwide by 30 April and were complemented by ring vaccination of contacts in early June (Fig. 1a). The transition to mandatory isolation, initiated after home-based management was observed to be ineffective within this particular setting, is credited as a key driver of the subsequent decline (Fig. 5b). Although our sampling period limits inference beyond August 2025, WAU maintained persistent transmission throughout, whereas other districts were primarily fueled by repeated introductions from this dominant hub.

Source link