This study demonstrates the utility of an HCV RNA and anti-HCV testing algorithm to identify acute HCV infections from cross-sectional screening, as well as the use of this testing algorithm to estimate HCV incidence. For four populations with various risks of HCV, the proportion of detected viremic seronegative HCV infections, assumed to represent incident infections, increased in parallel, with the lowest proportion detected among blood donors and the highest detected among long-term IDU. The HCV incidence estimated from the number of detected viremic seronegative infections was most accurate among young IDU, for whom the estimated incidence was 39.8%, comparable to the observed incidence of 33.4%. Younger IDU are among those at highest risk of HCV infection (21
) and, not surprisingly, demonstrated the highest rate of HCV infection of all groups included here. The observed HCV incidence in blood donors and in older drug users was lower than that projected from cross-sectional samples; however, confidence intervals for projected and observed incidence overlapped across all groups. Among the blood donors, the discrepancy between projected and observed incidence was likely attributable to donor status and possibly to differences in incidence in different demographic strata (53
). HCV window-phase infections are known to be three times more likely to be detected in first-time donors than in repeat donors (47
Various factors may contribute to discrepancy between projected and observed rates of HCV infection, including the testing interval and follow-up rates within groups. Overestimation of incidence may result if some viremic seronegative infections are misclassified as acute infections. Such “immunosilent” infections have been documented for immunocompromised hosts who are incapable of generating a detectable level of HCV-specific antibody and, rarely, for immunocompetent hosts (41
). Viral factors may also contribute to delayed antibody responses in rare circumstances, as recently documented by our group (5
). Variability in HCV RNA detection and natural history during early infection may also result in differences between projected and observed incidence rates for these groups, as discussed below in more detail.
Based on systematic HCV NAT screening of blood donors, immunosilent infections appear to be rare; the vast majority (>99.9%) of HCV-viremic donors are detected by antibody screening (46
), and only 3 (4.5%) of 67 RNA-positive/antibody-negative donors who enrolled in a prospective follow-up study conducted following implementation of NAT screening failed to seroconvert by EIA 3 within 1 year of RNA detection (47
). It is noteworthy that these three immunosilent cases were detected during the first year of donor NAT screening, with no confirmed immunosilent cases detected over the subsequent 6 years of screening of over 12 million U.S. whole-blood donations annually, indicating that persistent viremia in the absence of seroconversion is a very rare phenomenon in immunocompetent populations (S. Stramer, personal communication). Transient infections have also been documented among the young IDU population, although they appear to be rare, with two confirmed events among 121 infections detected over a 6-year period (K. Page-Shafer, unpublished data). Intercalations of HCV RNA positivity which may occur in the early natural history of HCV infection (40
) may also result in a small proportion of participants being misclassified as cleared or uninfected and may result in differences between projected and observed incidence rates.
Regarding human immunodeficiency virus (HIV) infection, none was detected among the blood donor group members, who were tested concurrently. VA patients were not tested for HIV in this study, although a recent study found 8.4% HIV prevalence among VA patients with chronic HCV infection (6
). Among the older IDU group members, HIV has been documented overall at 11.9% (51
), and among the young IDU tested for this paper, 881 were tested for HIV and 3.1% were HIV infected. Delayed antibody responses have been documented for HCV-viremic individuals, including those coinfected with HIV, in several settings (9
), including drug users (4
). Older drug users and those with the greatest number of years of injection exposure have a much higher HCV prevalence (34
), and many of these infections may have occurred 20 or more years previously. Coinfection with HIV is more likely in this older group (30
), possibly resulting in impaired antibody responses to acute HCV infection or HCV EIA seroreversion after prolonged HCV infection as HIV-induced immunosuppression evolves. Although the number can be expected to be low, aberrant antibody responses associated with HIV in these populations and the potential misclassification of acute infection may result in overestimation of the projected HCV incidence rates, especially compared to the observed incidence based on anti-HCV detection methods. Since 5% or more of HCV/HIV-coinfected patients may be anti-HCV negative and potentially misclassified as acutely infected (7
), caution is advised in the use of our proposed testing strategy to estimate incidence in HIV-infected or other immunosuppressed populations.
We acknowledge other potential limitations of this testing algorithm as well. First, the 50.9-day mean length of the viremic preseroconversion window period used in this testing algorithm was derived from a cohort of acutely infected plasma donors. This sample group is comprised of individuals with community-acquired HCV infection, generally assumed to be IDU (20
), and hence should be relevant to the populations represented in our analyses. Other estimates of this window period differ somewhat: Glynn et al. (20
) reported a mean of 56.3 days (95% CI, 44.8 to 67.8 days) for plasma donors, and Busch reported a mean of 60 days for the blood transfusion recipient setting (9
). It is possible that in addition to the infection route, other factors, including the amount or size of the inoculum (39
), exposure frequency, and even demographic factors, could influence the natural history of acute infection and hence the duration of the viremic preseroconversion phase. Differing distributions of viral genotypes within these populations could potentially influence window period estimates as well. Viral genotypes are known to influence disease outcome and the response to treatment (24
), and although to date there is little evidence of variation in acute-phase HCV by genotype (39
), data are limited, and further studies are warranted to assess window period estimates by viral subtype.
NAT testing for HCV RNA and other blood-borne viruses is now regularly and effectively used in the blood, plasma, and organ donor screening settings to reduce the residual risk of transfusion- and transplant-transmitted infections (47
). In donor screening, pooled sample testing, as was done for most of the groups assessed here, is now routinely employed for HCV, HIV, HBV, and West Nile virus NAT. Among such low-prevalence/incidence populations, this approach is the most viable in terms of testing capacity, turn-around time, and cost-effectiveness. Among populations with a high prevalence and incidence of HCV, such as IDU, pooled testing may be less efficient. First, a large proportion of anti-HCV-negative samples can be expected to be reactive, leading to the majority of pooled tests having to be retested and resolved individually. Second, in high-incidence populations, such as young IDU, a small but significant proportion (10% at a 1:10 dilution and 32% at a 1:100 dilution) of window-phase infections may be missed using the pooled sample approach. The HCV inoculum size may vary by route of infection, possibly affecting viral load during the window phase such that pooled or diluted sample approaches may be less effective at detecting viremia in early infection. Community-acquired infection is believed to be associated with a differing immune response as well (29
). Further studies of viral kinetics and immune responses during early HCV infection following different exposure routes are needed to address these issues.
Aarons et al. (1
) used a similar approach to detect acute HCV infections among IDU in London. Seronegative samples were tested retrospectively for HCV RNA, using minipools of 20 samples each; positive tests were resolved using individual dilution (1:20) testing. HCV viremic seronegative infections were identified, and incidence was estimated based on a 58-day window period (9
). The estimated incidence from cross-sectional testing (12.5%) was compared to the observed incidence (16.1%; n
= 2) among IDU who were tested multiple times during the same time period. The authors noted that poor storage and the dilution factor may have contributed to the underestimation of estimated incidence. Our results showing that 10% to 30% of acute infections may go undetected upon dilution support this possibility. In both the above-mentioned study and this one, 5% to 7% of anti-HCV-negative IDU were confirmed to have HCV infection. All of these infections would have been missed by conventional serological screening, and many would have been missed by commercially available PCR-based quantitative HCV viremia assays, due to the relatively low sensitivities of these tests. Together, these results support the use of sensitive HCV RNA screening of high-risk populations.
In conclusion, the utility of pooled NAT screening to detect window-phase infections is well recognized for blood supply safety. With respect to screening for acute HCV infection in high-risk populations, individual screening will provide the best estimates of HCV incidence and can be an effective tool for public health surveillance and case-finding purposes. On an individual level, identifying acute HCV infection may help in reducing transmission risk from acutely infected individuals, who may be more infectious due to high-titer viremia, as seen with HIV (54
). Since it is now recognized that treatment of acute HCV infection is highly efficacious (26
), this testing approach may be used to inform and improve health care needs of IDU. Finally, this testing strategy will be highly useful for identifying high-incidence populations for future intervention studies, such as preventive HCV vaccine trials.