Isolated in 1989, the hepatitis C virus (HCV) has since been found to be a globally diverse, positive sense, RNA virus, belonging to the Flaviviridae family, and is classified into at least 7 genotypes and numerous subtypes [1]–[4]. Each genotype, and subtype, displays different patterns of endemic and epidemic epidemiology. For example, subtype strains such as 1a, 1b, 2a, 2b, 3a, and 4a, are linked with 20th century outbreaks caused by the use of unscreened blood products, illicit drug injecting, or non-sterile medical injections [5]–[10]. In contrast, other local endemic strains appear to have been present in particular populations for many centuries, such as genotypes 1 and 2 in West Africa and genotype 6 in South East Asia. Genotype 6, for example, exhibits a very high degree of genetic diversity, and evolutionary molecular clock analyses suggest that the subtypes of this genotype diverged at least 1000 years before present [6].

HCV exposure in 50–80% of cases (depending on host and virus factors) [11]–[14] leads to a chronic infection that can in turn result in chronic liver disease, liver cirrhosis and liver cancer, each resulting in considerable morbidity and mortality. The high prevalence of HCV infection makes it a health problem of international importance. Studies over the last 23 years have identified that clinical outcomes depend on both host and viral factors. For example, individuals infected with genotype 1 are less likely to clear viral infection upon treatment [15]. Recent genome-wide association studies findings have identified haplotypes of single nucleotide polymorphic markers (SNPs) strongly associated with spontaneous and treatment-induced clearance in genotypes 1 and 4 (i.e., the C in haplotype rs12979680, and the T in haplotype rs8099917) [16]–[18].

The primary risk factor for HCV infection is exposure to infected blood or blood products, for example, via unscreened blood transfusions or cutaneous injury with HCV contaminated needles, piercing instruments, tattoo pens, or surgical and dental instruments. Throughout the 21^st century the causes of local epidemics have included unsafe medical practices, illicit injecting drug users (IDUs), and unsafe vaccination practises. Whilst rare, HCV can also be transmitted via sexual exposure. HCV has also been detected in saliva, breast milk, urine, and faeces, as well as semen and cervico-vaginal secretions [10].

Any effort to prevent transmission and control HCV requires accurate epidemiological data. Serological screening assays have played a major role in understanding HCV epidemiology and have reduced the risk of HCV transmission and the global disease burden considerably. Unfortunately there is much debate with regards to their most efficacious usage and interpretation [19]–[24]. The many tests vary in cost, sensitivity, and efficiency resulting in a plethora of recommendations on appropriate use and interpretation. The implementation of efficacious screening programs in the developing South Pacific Island Countries (SPICs) is primarily hindered by cost, but is also problematic because these assays commonly produce biological false-positive reactions, particularly in populations of low HCV prevalence and in tropical environments where the frequent level of exposure to other pathogens (such as malaria and dengue virus) can cause cross reactivity [11], [25]–[28]. Nonetheless, there have been a few HCV prevalence surveys in SPICs; the earlier screening surveys (from serum samples collected as early as 1984, but predominantly from 1990–91) presented anti-HCV sero-prevalence rates up to 8% [29]–[34]. More recent studies, circa 2006, have estimated lower infection rates of 2–4% [35]. Despite HCV being endemic world-wide and despite there being strong indicators of HCV presence in SPICs, the true epidemiology of HCV infection in the developing nations of the 22 SPICs is not known, as the resources for high-quality studies are not readily available and national screening procedures are not in place.

Here we estimate the frequency of HCV sero-reactivity and RNA carriage in three SPICs Papua New Guinea (PNG), Fiji and Kiribati, amongst otherwise healthy adults. These three low-to-middle income nations (as classified by the World Bank) carry a high burden of Hepatitis B Virus (HBV) [36], [37]. The use of marijuana and alcohol is widespread in these Pacific Island societies and is closely associated with unprotected sexual activity and acts of sexual violence, however, injecting drug use is not recognised to be prevalent in the region. The incidence of injecting drug abuse is considered to be extremely rare (pers. comm. Dr Kelly, PNG Institute of Medical Research); it has only been recorded as the mode of transmission in one case of HIV transmission. Interestingly tattooing is also not considered to be an important risk factor [39], it remains to be seen if penile injecting, which is wide spread in the region, is similarly not associated. Fiji introduced HCV serological screening for transfused blood in 2006 but neither PNG nor Kiribati screen, therefore in these countries exposure via unscreened blood transfusion remains the primary risk factor for HCV transmission.

Table 1 provides a summary of the results of the serological assays by age, sex and country, Table 2 provides a summary of the results by country alone, and Table 3 gives the results by town within PNG. Table S1 presents the serological results including the signal-to-cut-off (S/CO) ratios, the results of the nucleic acid testing (NATs) and the IL28B results. Fourteen PNG, one Fijian (an adult female) and zero I-Kiribati samples tested reactive for both the Ortho HCV3 assay and the anti-HCV Monolisa assay.

In PNG there were considerable provincial differences in HCV sero-reactivity. Samples were collected at sites representing 5 of the 19 provincial regions in PNG: Central Province (Port Moresby), Western Province (Daru), Madang Province (Madang), Eastern Highlands Province (Goroka), and Western Highlands Province (Mt Hagen). All samples collected from Daru, Goroka, and Mt Hagen were negative for HCV; 2.2% and 9.6% of the samples collected in Port Moresby and Madang respectively, tested positive. We found no significant difference in prevalence between men and women (2-tailed Fisher's Exact Test) nor between those <50 and those >50 years old (2-tailed Fisher's Exact Test). S/CO ratios (see Methods) from both assays were calculated and compared and we found inconsistent results between the assays. Five samples tested with the Ortho 3 assay consistently had S/CO ratios greater than ≥3.8 (samples P_047, P_130, P_242, PW018 and PW064) but only two tested with the Monolisa assay had a ratio ≥5 (samples P_109 and P_130). This lack of consistency reaffirms the difficulty in scoring HCV-positive samples in a resource-poor setting where the more specific recombinant immunoblot assay (RIBA) tests or HCV-NAT testing may not be available.

Duplicate NATs for HCV RNA were conducted on all samples that gave a reactive serologic screening response. No viral RNA was detectable in any sample. As part of the original HBV study we were required to return to the collection sites and provide follow-up support and feedback from the study. On our last visit in 2008, three of the participants that screened reactive for HCV returned for follow up; at this later time they were re-bled and retested and none were HCV RNA positive.

The IL28B haplotype results are listed in Table S1. There was considerable homogeneity in the populations sampled. All but one of the fifteen sero-positive samples were homozygous for the major allele C in haplotype rs12979860 and all were homozygous for the major allele T in rs8099917. The remaining sample, P_242, was heterozygous for the rs12979860 SNP. The CC/TT genotype is strongly associated with spontaneous and treatment-induced clearance in genotypes 1 and 4. [16], [17], [38]. A further eighty-eight individuals from the same three countries were typed for the rs8099917 SNP as part of another study and the results were also homogeneous (see Table S2).

Our results demonstrate a national HCV sero-prevalence of 2.2% in PNG, 0.1% in Fiji, and 0% in Kiribati, considerably less than previously reported but in line with the few published reports of the sero-prevalence in the SPICs [33], [35], [39]. However, we observed considerable variability in HCV sero-prevalence among the sample sites. Five sites were screened in PNG but HCV reactivity was found in only two locations (Port Moresby and Madang). These two towns are in lowland coastal areas with moderately well-supported hospitals and infrastructure, both have a notable influx of foreign workers and tourists, and both are towns that were established early in PNG colonial history. Any one of these factors may have introduced the virus to these particular provinces. HCV is unlikely to have become established by IDUs. A handful of IDUs from PNG have been reported in behavioural survey studies, but there is no known IDU cohort or culture, nor reliable estimates of IDU prevalence in PNG [40]). It is unlikely that the 9.6% sero-prevalence we found in Madang reflects the general population exposure, as closer inspection revealed that the sample population was enriched for high-risk exposure participants such as health-care workers or people that handled blood as part of their profession (e.g. nurses or phlebotomists). Indeed, individuals PW013, PW017, PW023, PW062, PW064, and PGM016 were all health-care workers of some kind. The possibility that HCV exposure might be so prevalent amongst health-care workers highlights the need for better awareness and control.

Interestingly, in concordance with all previously published studies from SPICs, no HCV RNA could be confirmed in any of the samples [29], [30], [32], [35]. For our initial samples collected in 2005, inappropriate long-term storage could be a reason for the lack of detection of HCV-RNA. However, for the three individuals retested in 2008 this is not the case as each sample was tested for HCV-RNA within an hour of the blood samples being drawn. Neither HCV screening nor treatment are available in PNG and none of the HCV-positive individuals indicated in their initial interview that they were aware of having had HCV. Therefore the infected individuals will most likely have spontaneously cleared the infection. Spontaneous clearance estimates for HCV range from 15–50% depending on the population being sampled [11], [14], [41]. Clearance has been attributed to various factors, including host innate and adaptive immunity as well as being associated with particular genetic markers such as the Il28 SNPs [14], [17], [18]. We screened all our HCV antibody-positive samples for the rs12979860 and rs8099917 SNPs; all but one individual was homozygous for the “protective” C and T haplotypes respectively. Thus a high rate of HCV clearance in these individuals may have been enhanced by the presence of these favourable IL28B alleles - it is plausible that the individuals who were HCV sero-positive but PCR negative may have had only a transient infection that spontaneously resolved. Previous work by Thomas et al. 2009 [17] who screened for the rs12979860 SNP, and by Shebl et al. 2011 [42] who screened for rs12979860 and rs8099917 markers, and a separate parallel study of our own (where eighty-eight individuals from the same three countries where screened for the rs8099917 SNP; see Table S1) have demonstrated a 70–100% prevalence of the major ‘protective’ rs12979860 and rs8099917 haplotypes in indigenous SPIC populations. However, these haplotypes are only associated with spontaneous viral clearance of genotypes 1 and 4 and no information is available regarding the circulating genotypes in these populations. If the HCV strain circulating in PNG is an ancestral endemic strain then it is more likely to be genotype 6 (the endemic strain circulating in South East Asia) or perhaps a close relation. If HCV was recently introduced it is more likely to be one of the epidemic strains such as subtypes 1a, 1b, 2a, 2b or 3a [6].

The prevalence and preponderance of HCV makes it a global health problem and accurate epidemiological data must underpin any effort to prevent transmission and control the virus. Unfortunately the epidemiology of HCV infection in SPICs is not well characterised [29]–[35]. National blood transfusion services exist in the region, to ensure a safe and secure supply as well as use of blood and blood products but only a few SPICs screen for HCV. There is no information regarding circulating genotypes. In SPICs, IDUs are extremely rare, and to date scarification and tattooing has shown no association with HCV [36]. Consequently, blood transfusions very likely present the greatest risk of infection. Clearly the primary risk factor for HCV infection (i.e., exposure to infected blood or blood products) remains neglected. We were fortunate in having a sample bank of sera collected from mother-child pairs and healthy adults in PNG, Kiribati and Fiji that were collected for the purpose of viral hepatitis research. From this bank, 2224 serum samples were made available for screening and have contributed significantly to the knowledge of HCV in the region.

Finally, there is much debate over the correct interpretation of serological test results, particularly in a resource-poor setting where RIBA testing or HCV-NAT testing may not be available. In our study five samples scored “truly positive” in the Ortho-3 assay and two in the BioRad assay, but only one sample scored “truly positive” in both assays. This reaffirms the difficulty in scoring HCV positive samples. In our work we chose to follow recommendations from the WHO and Owusu-Ofori et al 2005 [23] and scored samples as either past, or presently positive if positive in two assays. We find this to be more practical than using the S/CO ratios.

This study has revealed that the HCV sero-prevalence rate in SPICs is variable but generally lower than previously estimated. We were unable to detect HCV RNA and define genotypes in this extensive survey. The ability to assess accurately the burden of HCV disease is clearly important in defining public health priorities, as well as, indicating the history and the likely future impact of this infection. The high rate of false-positives using a single screening test suggests that caution must be applied in interpreting such assays, particularly in low-prevalence settings where there is a high burden of other chronic infections. Applying such a cautious multi-layered approach will be of value in determining the impact of HCV endemic and epidemic strains in diverse global populations.

Finally, we note that, despite the relatively low frequency of HCV, the rate of liver cancer in this region is still very high. The simplest explanation for this is the high prevalence of Hepatitis B virus infection [31].