Our study revealed that multiple HCV lineages are transmitted at the time of liver transplant without a major decrease in viral genetic diversity. Although only some of the pretransplant lineages are identified within the first 4 months posttransplant, lineages are undoubtedly present because their ancestors are sampled at later time points. The data clearly argue against a bottleneck scenario in most transplant recipients and suggest that little restriction of diversity occurs in the new liver in which multiple lineages set up a new infection.
It could be suggested that a restriction in the viral population is unlikely to be observed in the 4 months posttransplant, since insufficient time has elapsed and patients are typically on immunosuppressive drugs during this period that may reduce the selective pressure on the virus. However, we show the absence of a transplantation bottleneck for as long as 1 to 2 years posttransplant. Although analysis of sample genetic diversity was consistent with a bottleneck in 3/10 patients (A, D, and I), subsequent population genetic analyses plainly demonstrated by multiple parameters that the genetic bottleneck signature failed to apply to the virus populations in these patients and that estimates based on sample diversities have underestimated the actual diversity of the viral population.
Furthermore, multiple distinct viral lineages sampled a year or more after transplant share a ancestor with viruses sampled well before (2 years) transplant, rather than with viruses sampled within 4 months posttransplant, as would be expected for a continuously replicating virus. Our sequencing strategy allowed for detection of viral variants at ~5% level (average of 24 clones/time point). Thus, it is possible—and indeed very likely—that variants belonging to these unobserved lineages were present in the serum at levels below our threshold of detection. However, a fundamental tenet arising from molecular evolutionary theory is that, in the absence of a active mechanism of maintenance, segregating variants in a population will be lost, either due to fixation (via random genetic drift or positive selection) or due to elimination (via genetic drift or negative selection). The mean survival time of very rare segregating variants (whether advantageous, neutral, or deleterious) is particularly short as a result of their high probability of stochastic extinction (see, for example, reference 38
). Thus, the persistence of rare variants over long periods of time requires an active evolutionary force, such as frequency-dependent selection (whereby the fitness of a variant declines as its frequency increases) or spatially conditioned selective pressure that results in local adaptations (and thus population structure) (31
). Our present study cannot distinguish between these two possibilities. In the latter, the population structure could arise from spatial segregation in the liver or from virus replication in additional anatomic compartments (39
). For example, hepatic lymph nodes (37
) and peripheral blood mononuclear cells (25
) have been suggested possible extra-hepatic sources of infection for the new liver. Other possible reservoirs include macrophages (23
), the central nervous system (10
), and/or B cells (6
). However, previous models have suggested that only a small fraction of posttransplant viruses originate from these sources (34
Primer/PCR bias is unlikely to be a factor in the detection of the major variants, as a bias would differentially affect the pool of variants such that one particular type would be preferentially amplified. In the present data set, however, entirely different variants were amplified at various time points. PCR misincorporation errors and/or recombination would not account for the deep lineage structure observed. Interestingly, an ongoing independent analysis of the clinical samples studied here using pyrosequencing of the NS3 region with a limit of detection of ~1% suggests a similar pattern in which clades contain sequences from well before and after transplant, but none from the time points surrounding the operation (G. Wang, unpublished data). This observation will be investigated further in future work.
A complex suite of selective pressure is likely to operate on the virus in this cohort, including the potentially strong selective pressure of a new liver expressing different HLA alleles, as well as a potential reduction in host immune pressure as a result of immunosuppressive therapy. Positively selected sites were detected in 7 of 10 patients, mainly in the HVRI region. These sites are most likely to be under diversifying selection as methods using the dN/dS test within a population are the most sensitive to this type of selection, and this region is known to be the target of T and B cell immune responses. Lineages containing sites under positive selection were distributed through time and in several cases led to one of the dominant variants posttransplant, indicating a potential selective advantage of those variants.
The results of the present study differ significantly from those reported previously. Our study included a longer temporal sampling strategy and used more extensive analysis methods. Although mean pairwise genetic distance measures of sample
diversity are one method for assessing bottlenecks, the statistic is readily confounded when the sampled population is structured into subpopulations (49
) because pairwise distance ignores the phylogenetic relationships among samples. In contrast, measures of population diversity explicitly take in account the phylogenetic relationships among samples and can infer the presence of unsampled lineages at earlier time points from the observation of their descendants at later times. In addition, a long-term sampling strategy that included viruses sampled from several years before and after the transplant revealed the nature and complexity of the viral population that reinfects the new liver, which would have been obscured by a simple cross-sectional analysis immediately after the transplant.
All virus populations in the present study were obtained from serum. Although such viruses are often assumed to represent the viral population in the liver, serum viruses may also contain variants from nonhepatic sites. Future studies should include both liver biopsies and long-term temporal sampling to understand the dynamics of transplant reinfection.