Assessment of African swine fever vaccine candidate ASFV-G-AMGF in a reversion to virulence study

African swine fever (ASF), caused by ASF virus (ASFV), has spread globally, posing major threats to domestic pigs and endangered wild suids. In the absence of licensed vaccines, live attenuated vaccines (LAVs) currently offer the most promising protection against lethal field strains, but their replication competence raises safety concerns, including potential genetic mutations and reversion to virulence. This study evaluates the ASFV-G-AMGF (AMGF), a genetically modified, efficacious LAV candidate, for reversion to virulence and genetic stability using a standardized in vivo passaging protocol (VICH GL41) in naïve weaner pigs across five serial passages.

Prior work identifies LAVs as the only approach consistently conferring complete protection against highly virulent ASFV strains, yet safety concerns persist due to residual virulence and risks of genetic change during replication. Historical experiences in Iberia using attenuated field isolates and recent unregulated vaccine use reports highlight risks of prolonged, subclinical infections and sustained virus shedding. Genetic alterations, especially in multigene family (MGF) regions, are implicated in host interactions and interferon signaling. Naturally occurring large terminal deletions and rearrangements have previously been described (e.g., ASFV Estonia 2014) and associated with phenotype changes. These references frame the need for comprehensive safety evaluation, including reversion-to-virulence testing under worst-case conditions.

Study design followed VICH GL41 for reversion to virulence in target animals using the master seed virus (MSV) at maximum expected release titer and serial passaging five times. Animals: Five groups of ten 6–10-week-old naïve weaner pigs (FLI). Housing in high containment; daily clinical scoring and rectal temperature recording. Sampling: P1–P4 euthanized/sampled at 7 dpi; P5 observed to 21 dpi with weekly oral/nasal/rectal swabs. Full necropsy with standardized pathology; tissues collected included spleen, lung, liver, tonsil, kidney, salivary gland, and gastro-hepatic and mandibular lymph nodes; EDTA blood collected. Inoculation: Intramuscular injection into neck (1 ml). P1 received AMGF MSV at 1.75×10^5 HAD50/ml. For passaging, at each passage tissues with the lowest cq (highest genome load) were selected, pooled equally, homogenized (10% w/v in PBS), clarified, and used as inoculum. Back titers/inoculum titers: P1 tissue homogenate back-titrated 10^4.25 HAD50/ml; P2 10^2.25; P3 10^4; P4 10^5.75 HAD50/ml. Tissue sources for inocula: P2 spleens from pigs #1081, #1082, #1083, #1085; P3 spleen+lung (#1073) and spleen+retropharyngeal LN (#1078); P4 spleens (#1072, #1074, #1082, #45, #46); P5 spleens (#1091, #1092) and retropharyngeal LNs (#28, #30). Virology and molecular methods: ASFV genome detection by qPCR (King et al. protocol or virotype 2.0) with in-house full-genome standards for gc quantification; swab processing as per SOP. Whole-genome sequencing on Illumina NovaSeq (PE150) after library prep; reads trimmed, mapped to MSV reference (Newbler), assembled (SPAdes), curated (Geneious), and validated by remapping. Tailored qPCRs designed to span the 5′ reorganization site: FAM-labeled assay for AMGF MSV and HEX-labeled assay for the novel variant AMGFnV. Virus isolation and titration on PBMC-derived macrophages with hemadsorption readout; HAD50 calculated by Reed–Muench. Growth kinetics: Macrophage infections at MOI 0.1 with AMGF MSV, AMGFnV, or both; sampling at −2, 0, 4, 8, 12, 24, 48, 72 hpi for qPCR differentiation and titration. Ethics: Approved by LALFF M-V (7221.3-1-020/21); EU Directive 2010/63/EC compliance.

Clinical: No abnormalities or fever in P1. In P2, 3 pigs had transient temperature rises up to 40.4°C. From P3 onward, frequent fever with peaks around 5–6 dpi; maximum 42.1°C (pig #31, P4, 6 dpi). In P5, 9/10 pigs reached ≥41°C at least once. Clinical signs were mild (reduced appetite, apathy), with a maximum cumulative score of 3 (one P4 pig); most others scored 0–1. Temperatures normalized and animals were clinically healthy by day 21 in P5; one pig had transient non-ASFV lameness (P5). One pig euthanized for unrelated nerve injury at 7 dpi. Pathology: Very few lesions (slightly enlarged lymph nodes, pulmonary consolidation) with no correlation to passage or time point. Virology (qPCR genome loads): P1 ≤7.2×10^3 gc/5 µl across samples; virus undetected in two pigs. P2: ≤5×10^1 gc in a single sample; virus not detected in five pigs. P3: all pigs positive in ≥1 sample; up to 1.8×10^3 gc. P4: all pigs positive; maximum 2.6×10^3 gc. P5: all pigs positive with slightly lower tissue loads than P4 but similar blood loads; pig #21 up to 2.6×10^3 gc (euthanized 7 dpi). Swabs: Low loads at 7 dpi (up to 4 gc) in 7/10 pigs; 14 dpi 6/9 positive; 21 dpi 3/9 positive, with higher loads at 21 dpi than earlier. Genomics: Detected a novel variant (AMGFnV) in P4 samples characterized by an 11,197 bp 5′-terminal deletion (loss of 18 genes) and a 18,592 bp inverted duplication from the 3′-end resulting in duplication of 29 genes; recombination sites did not involve the engineered MGF deletions; core genome and engineered MGF deletions remained stable across passages. Emergence and spread: AMGFnV traced back to pig #1081 in P1 (mixed infection). P2: detected in two pigs (one mono-, one mixed). P3: all positives carried AMGFnV (one mono-, nine mixed). P4: all AMGFnV-positive (one mono-, nine mixed). P5: four monoinfections and six coinfections (AMGFnV + MSV). In vitro growth: Similar kinetics for AMGF MSV and AMGFnV in macrophages; final titers up to 10^6.75 HAD50/ml in monoinfections and 10^7 HAD50/ml in coinfection; tailored qPCR Cq values indicated comparable replication (AMGFnV ~17 vs nV ~19–21 at 48–72 hpi). Overall outcome: Despite increased replication and transient fever associated with AMGFnV emergence, no reversion to significant virulence was observed; all animals were clinically normal at study completion, and virus loads in tissues were low or absent.

The study directly addressed reversion-to-virulence risk by subjecting AMGF to worst-case, forced serial passaging in pigs. The emergence of AMGFnV with large terminal rearrangements indicates that ASFV can adapt under selection pressure without affecting the engineered attenuation loci. Clinically, the variant correlated with transient fever and somewhat increased replication and shedding, but there was no progression to high virulence akin to parental Georgia07. In vitro kinetics did not reveal a replication advantage, suggesting that host-specific factors in vivo may drive the modest phenotype change, potentially involving interferon-related pathways linked to MGF regions, though initial assays did not show clear interferon modulation differences. The findings underscore the necessity for comprehensive safety assessments of LAV candidates, including genetic stability monitoring, shedding, and transmission studies. The acceptability of brief, mild fever for a first-generation vaccine must be weighed against benefits in a stringent benefit–risk analysis, provided genetic stability of the variant is confirmed and further adaptation is unlikely. The identified genomic rearrangement resembles naturally occurring events (e.g., Estonia 2014), suggesting a potential common mechanism of ASFV adaptation under selective pressures.

Under stringent reversion-to-virulence testing, AMGF maintained its engineered stability but yielded a novel terminally rearranged variant (AMGFnV) early in passaging that modestly increased in vivo replication and transiently induced fever, without reversion to significant virulence. These findings highlight both the promise and the safety challenges of LAV-based ASF vaccines. Future work should focus on: (i) long-term follow-up beyond 21 days, (ii) shedding and transmission (including contact studies), (iii) stability of AMGFnV over passages and under field-relevant conditions, (iv) immunogenicity and host-response characterization to both MSV and AMGFnV, and (v) development and deployment of genetic DIVA tools. Conditional, controlled field use could be considered only after a thorough benefit–risk assessment incorporating these data.

The passaging approach and intramuscular inoculation of pooled high-load tissues represent an artificial, worst-case transmission scenario that may not reflect field conditions. Increasing inoculum titers across passages could confound clinical outcomes (e.g., fever). The observation period extended to 21 days only in P5, limiting insights into long-term persistence or chronic outcomes. No transmission to naïve contacts was evaluated. In vitro macrophage assays may not capture in vivo host–virus interactions; mechanisms underlying the in vivo replication advantage remain unresolved. Genomic analyses were limited to samples meeting input thresholds; subtle minority variants may have been missed.