Human papillomavirus type 16 variants and risk of cervical cancer.

There are substantial data demonstrating that human papillomaviruses (HPVs) are the sexually transmitted etiologic agents of cervical cancer (1). HPV type 16 (HPV16) is the most common HPV type detected in tumors, accounting for 50% of cancers and their precursors, called high-grade squamous intraepithelial lesions (HSILs) (2). Preliminary studies (3–16) have suggested that variants of HPV16 may show varying degrees of association with cervical neoplasia. This may partially explain why some HPV16 infections progress to HSIL or cancer, while others do not. If causal, these associations may be explained by differences in the transcriptional regulation of the virus by different variants, in the biologic activities of the proteins encoded by HPV16 variants (e.g., enhanced transforming abilities of E6/E7), or in the ability of the host to respond immunologically to specific viral epitopes encoded by variants. This last effect is likely to be mediated through human leukocyte antigen (HLA) presentation of viral antigens (17–21). Previous studies of viral variants and cervical neoplasia have relied on convenience samples, limiting their interpretation, while others have had small sample size. We conducted a prevalent case– control study within a 10 000-woman, population-based cohort in Costa Rica to examine the association between HPV16 variants and cervical neoplasia. In addition to its size and populationbased design, our study has the advantage of having been conducted in a highly admixed population. Our investigation was approved by institutional review boards in the United States and in Costa Rica. All participants provided written informed consent. The cohort from which this study derives has been described (22–24) and consists of 10 077 women (>93% response rate) who were interviewed and screened for cervical abnormalities by use of conventional cytology and newer screening methods (ThinPrep, Papnet, and cervicography). The cohort consists of 10 049 women randomly selected from the general population and an additional 28 women (22) from the same ascertainment area diagnosed with cancer and representing 90% of all cervical cancer cases diagnosed among women from Guanacaste, Costa Rica, during the cohort ascertainment period. Those women with an abnormal screening test were referred to colposcopy, where biopsy specimens were taken. Review of cytohistologic data revealed 40 cancers (12 within our random sample plus the 28 cancer cases described above), 128 HSILs, 189 lowgrade squamous intraepithelial lesions (LSILs), 661 equivocal lesions, and 7564 cytologically normal results (22,23,25). The entire cohort was screened for HPV DNA by the use of the Hybrid Capture Tube test (Digene Corp., Gaithersburg, MD) (24,26). Polymerase chain reaction (PCR)-based testing was also performed on more than 40% of cohort members, including all women with any screening abnormality, those who were positive by hybrid capture, and a sample of the remaining cohort (23). HPV16 DNA by PCR was detected in 190 subjects. Sequencing of the long control region (nucleotides 7408– 7891) of HPV16, which contains the highest degree of variation in the viral genome, was performed in a blinded fashion on 176 subjects (16 cancers, 56 HSILs, 20 LSILs, and 84 with equivocal lesions or normal diagnosis). PCR products were confirmed by agarose gel electrophoresis and purified by use of the Quickstep PCR kit (Edge BioSystems, Gaithersburg, MD). Sequences were determined by cycle sequencing. Host genotyping was also performed to determine the degree of genetic relatedness of individuals in our study with the use of a set of commercially available microsatellite markers (AmpFISTRTM ProfilerPlus; PE Corporation, Foster City, CA) (27). Specimens were tested by the use of the kit reagents and a modified protocol (28). Genotyping was performed on a subset of 140 women, including all 16 cancers, 55 HSILs, and 69 women with LSIL, equivocal lesions, or normal cytology, and was successful for all specimens but two tested. The genetic distance between groups was computed by Nei’s method (29) to determine whether subjects with prototype and variant forms of HPV16 were genetically heterogeneous (29) so that we could address concerns of population stratification arising from the coevolution of HPV with the human species. At the time of this analysis, high-resolution HLA class II DRB1 and DQB1 testing was available for 95 and 98, respectively, of the 176 study participants from a parallel study within our cohort (manuscript in preparation). HLA testing was performed by a PCR-based sequence-specific oligonucleotide probe method as described previously (30–32). The European-derived HPV16 prototype virus (EP[g]) was detected in 36 subjects. In addition, three distinct variant groups were observed. The most common variant (EP[a]) was a G to A substitution at position 7521. In contrast to EP[g], EP[a] was not associated with disease (relative risk [RR] for HSIL 1.0; RR for cancer 1.3; overall P .94) and was combined with EP[g] for analysis. Twenty subjects were found to contain variants that, other than at posi-