Jorge Quarleri, Ana Silvia Haedo, Ana Ceballos, Veronica Lidia Mathet, Lidia Espínola, Alessandra Maríncola, Marcela L. Natal, Liliana Martínez Peralta, Jose Raul Oubiña, Vanesa Ruiz, and Daniel O. Sánchez
To investigate GB virus C (GBV-C)/hepatitis G virus (HGV) mother-to-infant transmission when only 5′ noncoding region (NCR) sequences were available, a retrospective study was carried out. Although the 5′ NCR is widely used for GBV-C/HGV detection due to the high degree of sequence conservation among isolates (8), this fact clearly limits its potential use to trace chains of transmission, despite the presence of scattered subregions with higher heterogeneity. Nevertheless, since the 5′ NCR has been reported to be unmodified after 8.4 years of evolution (5), a mother-to-infant chain was expected to show (almost) identical sequences unless specific 5′ NCR variants were involved in selective transmission. Analyzed sequences were obtained from (i) a study group which encompassed plasma samples simultaneously obtained from 34 human immunodeficiency virus (HIV)-infected mothers (13 intravenous-drug users [IVDU] and 21 nonaddict women putatively infected by a heterosexual route [IHSR]) as well as from their respective infants (1 to 7 months old, without history of blood transfusion) born to either HIV-positive and GBV-C/HGV-infected mothers (n = 8) or to HIV-positive but GBV-C/HGV nonviremic mothers (n = 26) and (ii) a control group which comprised 31 5′ NCR sequences derived from contemporary isolates circulating in the city of Buenos Aires and 6 others from the rest of the world. Several precautions (3) as well as appropriate controls were included to minimize the possibility that sequence identity among samples was the mere result of PCR contamination: (i) samples from mothers and infants were analyzed separately (elapsed time, 2 months); (ii) each positive result was confirmed in duplicate; and (iii) one negative control (SDW) was included for every three samples. To estimate the potential influence of Taq-dependent misincorporated nucleotides during PCR amplification (4), three cDNAs from local isolates of GBV-C/HGV whose sequences (325 bp) had been previously established (6) were amplified in duplicate and bidirectionally sequenced. A total of 4 of 13 IVDU and 4 of 21 IHSR mothers (30.8 and 19.0%, respectively; P not significant) proved positive for GBV-C/HGV RNA ascribed to genogroup 2a (n = 4), 2b (n = 2), or 3 (n = 2). Only three of eight infants born from viremic mothers (and none from 26 nonviremic mothers) exhibited GBV-C/HGV RNA (37.5%; one born to an IVDU and two to IHSR GBV-C/HGV-positive mothers); their genogroups, as established by both restriction fragment length polymorphism and cDNA sequencing (8), were the same as those of their respective mothers. The phylogenetic analysis was based on both genetic distance (neighbor joining) (Fig. (Fig.1)1) and maximum parsimony (data available upon request) methodologies. Both procedures showed unambiguously a genetic relatedness (i.e., no mutation between them) in each GBV-C/HGV mother-infant pair. In contrast to such a lack of nucleotide differences among the three mother-infant pairs, a mean ± standard deviation of 11.56 ± 6.17 nucleotide (nt) replacements was found among unrelated chosen sequences (n = 155 pairs of sequences analyzed). These control sequences corresponded to simultaneously circulating strains in the city of Buenos Aires which were ascribed to the same genogroups putatively associated with transmission chains in this study (i.e., genogroups 2a [n = 17 isolates] and 3 [n = 9]). None of these 26 sequences were identical. When nucleotide replacements were analyzed separately among all included (sub)group 2a (n = 119) and group 3 (n = 36) pairs of sequences, means ± standard deviations of 10.13 ± 3.16 and 16.77 ± 9.77 changes, respectively, were recorded (P < 0.00001; Student t test [two-sided test]). However, one pair of epidemiologically unrelated sequences (Arg15/Arg25) exhibited only one nucleotide difference (a 1-nt gap), seemingly appearing genetically related by the neighbor-joining analysis (Fig. (Fig.1).1). When maximum parsimony was carried out, only the three pairs of mother-infant 5′ NCR sequences remained undoubtedly related, since the lengths of their respective tree segments (in units of expected nucleotide substitutions per site) were equal to 0 and those between the above-mentioned sequences exhibiting a 1-nt gap were equal to 0.1073. Such a discrepancy between the two phylogenetic methodologies might be attributable to the reported (7) known loss of information with neighbor joining compared with maximum parsimony (distance versus discrete methods, respectively). FIG. 1. Phylogenetic analysis of GBV-C/HGV sequences (nt positions 132 to 337) from mother A (MA) and infant A (IA), MB and IB, and MC and IC compared with those obtained from simultaneously circulating strains in Argentina (ARG) and the rest of the world (W). ... The use of one mutation as a cutoff to discriminate related versus unrelated 5′ NCR sequences has the inherent risk of being Taq dependent. However, sequencing of the three control cDNAs (2,150 nt) rendered uniformly conserved (and repeatedly obtained) results. Thus, considering one mutation as the best cutoff value to discriminate between the related and unrelated 5′ NCR sequences analyzed in this study, the Youden's Index (sensitivity + specificity − 1) was equal to 1, the maximum possible value (1). In agreement with this view, the probability that three pairs of 5′ NCR sequences were identical (between members of pairs) due to random events is less than 1/17,576 (1/26 × 1/26 × 1/26), or 0.000056895, since none of the 26 simultaneously circulating local isolates ascribed to the same genomic group or subgroup implicated in putative chains of transmission were identical. Such mathematical support is coincident with the epidemiological data, since the studied babies lacked any risk factor for GBV-C/HGV infection except the corresponding mother's ongoing infection. As a whole, results obtained provided strong support to suggest the occurrence of GBV-C/HGV mother-to-infant transmission. The 5′ NCR was recently studied in transmission chains among hemodialyzed patients (2). To our knowledge, this is the first report to use both mathematical and phylogeneticanalyses to examine the eventual use of 5′ NCR sequences—when they are the only available data—to evaluate putative transmission chains.