Timing and Interpretation of Tests for Diagnosing Perinatally Acquired Hepatitis C Virus Infection

The diagnosis of hepatitis C virus (HCV) infection in children born to HCV-infected women is based on serologic assays and HCV RNA measurement by PCR. Interpretation of the results of these tests is hampered by uncertainty about the age distribution of loss of maternal antibody and the sensitivity and specificity of PCR at different ages. On the basis of findings from a recent vertical transmission study, we estimated the posttest probability of a child’s being infected or uninfected under several test result scenarios. These estimates may assist clinicians in assessing the likelihood of infection in an individual child and in using the currently available assays cost effectively. The diagnosis of hepatitis C virus (HCV) infection in children born to HCV-infected women is currently based on the detection of specific (nonmaternal) IgG antibody and/or the detection of HCV RNA by PCR. 1 However, the age at which passively acquired maternal antibody declines to undetectable levels and the sensitivity and specificity of PCR at different ages have not been accurately determined, with two important consequences for clinical management: (1) it is unclear how confidently infection can be excluded or inferred after a negative and positive PCR result, respectively, or when infection can be definitively diagnosed on the basis of a positive antibody test; (2) it hinders consideration of the most effective diagnostic test strategy, i.e., which assays should be performed at different ages. The high cost of PCR tests, the possibility of viral clearance and potential difficulties in maintaining follow-up of infants of indeterminate infection status, are other important factors that should be taken into account in this decision. Methods. The posttest probability of a child’s being truly infected or uninfected was calculated as described by Magder and Hughes, 2 assuming that infected children always test antibody positive and that PCR and antibody test results are independent (conditional on infection status). Estimates of PCR sensitivity and specificity and the distribution of loss of maternal antibody were obtained from a recent vertical transmission study conducted on 441 mother-child pairs in the United Kingdom and Ireland (Table 1). 3 Serologic testing was performed by second or third generation enzyme-linked immunosorbent assay with recombinant immunoblot assay confirmation. In Ireland HCV RNA PCR was performed in a single laboratory; in the United Kingdom most samples were tested in a reference laboratory under the auspices of the Public Health Laboratory Service, following national standard operating procedures (H Harris, personal communication). A key finding was the markedly lower PCR sensitivity in the first month of life than in older ages; in contrast PCR specificity appeared to be unrelated to age. 3Table 1: Estimates of PCR sensitivity, PCR specificity and distribution of loss of maternal antibodyThe predictive value of a diagnostic test depends on the underlying risk of transmission, which is influenced by several factors. Maternal HIV infection status is one of the most important of these and two analyses are presented, one based on the estimated HCV transmission risk among HIV-negative women (6.4%) and the other based on that among HIV-positive women (18.6%). In principle other factors such as the presence of maternal viremia could also be taken into account. Results. Figure 1 shows the probability that an antibody-positive child is truly infected or uninfected under several scenarios of PCR test results. The increase in the probability of infection with age, eventually reaching 100% for all curves, reflects the pattern of antibody disappearance in uninfected children, i.e. the later antibody is detected, the more likely it is to be endogenous rather than maternal.Fig. 1: Posttest probability of infection for an antibody-positive child according to age at last antibody test and results of PCR tests:A, no PCR test;B, one negative PCR before age 1 month;C, one negative PCR after age 1 month;D, one positive PCR before age 1 month;E, one positive PCR after age 1 month.Discussion. Figure 1 provides several insights into the interpretation and optimal timing of antibody and PCR tests for HCV infection. Although a negative serologic test at any age is very valuable because this allows infection to be excluded (with the caveat that there have been case reports of HCV carriers with undetectable levels of HCV antibody in children with 4 and without 5 concomitant HIV infection), such a finding is unlikely before 6 months of age because most children still have detectable amounts of maternally acquired antibody (Table 1). Moreover an early positive antibody test only slightly increases the probability of infection. It is not until ∼18 months of age that infection can be securely diagnosed based on serologic findings alone. Thus PCR is essential for the early diagnosis of infection. One strategy is to perform PCR soon after birth, before the infant and mother leave the hospital, but a negative result provides very little information because of the low sensitivity of PCR at this age. In view of the relatively high cost of PCR assays, it is preferable to delay testing until after 4 weeks of age, ideally to coincide with a neonatal checkup. A negative result on a delayed test strongly suggests that the child has escaped infection, although a confirmatory serologic test after 18 months of age is advisable. After a single positive PCR result after 1 month of age, the revised (i.e. posttest) probability of infection is 73% for a child born to an HIV-negative woman and 90% for a child born to an HIV-positive woman. These values are largely determined by PCR specificity, which encompasses false positive laboratory errors and the possibility of viral clearance. 1 Regardless of the explanation a single positive PCR is insufficient to demonstrate durable infection conclusively, and a repeat test should be performed. A recommendation on how long to delay the repeat test requires further information on the frequency and timing of viral clearance. It may also be prudent to confirm a virologically diagnosed infection with a serologic test after 18 months of age. The posttest estimates of infection we have presented depend critically on the assumed values for PCR sensitivity and specificity and, to a lesser extent, the serologic assays used. The values in this analysis pertain to techniques used in laboratories in the United Kingdom and Ireland between 1994 and 1999. However, quality control studies conducted in Europe have demonstrated substantial between-laboratory variability in test accuracy. 6 If sensitivity and specificity data for an individual laboratory are available, our method can be reapplied with this more relevant information; if such data are not available our estimates may still provide a useful basis for clinical decision making. Acknowledgments. We thank A. Conroy, S. Dooley, W. Hall, M. Beckett, T. Clarke, M. White and F. Gorman for their contribution to collection of the data in Ireland; Richard Lynn from the British Paediatric Surveillance Unit (BPSU) and reporting pediatricians and other health professionals who completed questionnaires; and members of the BPSU HCV study steering group for their support and collaboration: P. Tookey (National Study of HIV in Pregnancy and Childhood), M. Ramsay, K. Balogun, H. Harris (Communicable Disease Surveillance Centre), D. Goldberg (Scottish Centre for Infection and Environmental Health), Professor C. Peckham and A. E. Ades (Institute of Child Health); and clinical advisors to the study, D. Kelly and G. Mieli-Vergani. The UK component of data collection and the statistical analysis were funded by the UK Department of Health. All views and any errors are the responsibility of the authors alone. Appendix. In the calculation of posttest probability of infection, let P denote the pretest probability of infection and sens and spec the sensitivity and specificity of PCR (which may depend on age). Let A (t) be the proportion of uninfected children with detectable antibody at age t. Given a single PCR test with outcome Y (0 = negative, 1 = positive) and a positive antibody result when last tested at age T, the posttest probability of infection is given by MATH