Abstract
INTRODUCTION: Australia has provided unrestricted subsidized access to direct-acting antiviral (DAA) treatment for hepatitis C virus (HCV) infection since 2016. Epidemiological surveillance estimates suggest prevalence of chronic HCV infection has declined since 2016, but these estimates are not separated by genotype and may not capture 'hidden' infected populations, notably the most marginalized groups affected, including people who inject drugs and people in prison. This study used phylogenetics to assess whether epidemiological estimates of declining HCV prevalence in the prisons of New South Wales, Australia due to DAA scale up could be reproduced. METHOD: Near-full-length 280 HCV consensus sequences (GT1a: n = 140, GT3a: n = 140) sampled between 2006 - 2019 from two prison-based cohort studies in NSW were used for phylogenetic estimates. These included 110 acute infection sequences (GT1a: n = 48, GT3a: n = 62) which were considered in a separate sensitivity analysis given the differences in virus mutation rates in acute and chronic infection. Changes in the effective population size of infected people for each genotype were explored with BEAST software suite (v1.10) using a coalescent Bayesian skyline approach. RESULTS: Both the main and sensitivity analyses for GT3a showed a reduction in the effective population size with the latter showing a 36 % decline between 2011-2019 which is more concordant with the decline estimated from non-phylogenetic methods. A decline of similar magnitude was not demonstrated for GT1a. Overall, the analyses using acute infection sequences only were closer to the trends of independent epidemiological estimates. CONCLUSIONS: An adequately powered Bayesian evolutionary analysis using acute stage infection sequences may reproduce the decline in HCV infections observed by traditional epidemiological methods during DAA scale up.