Aims-To investigate the role of human papillomavirus (HPV) in the association between HLA DQw3 and squamous cell cancer of the cervix (SCCC). Methods-Tissue from 194 cervical samples, ranging from normal, through cervical intraepithelial neoplasia, to SCCC, were typed for HPV by amplification of the LI gene using degenerate consensus primers, followed by oligonucleotide probing. HLA DQw3 typing was undertaken in the same samples using a new PCR amplification system using primers common to all DQ loci, followed by restriction digestion with Mlu 1 to differentiate HLA DQw3 types-null, heterozygous, and homozygous. The data were analysed using X2 analysis and by calculating relative risks with the 95% confidence interval. Results-Samples (n = 188) were successfully typed for HPV and 177 were typed for HLA DQw3. There was a nonsignificant rise in the prevalence of HLA DQw3 in SCCC (64.3%) compared with the group with normal histology (53.2%). Analysis of the prevalence ofHLA DQw3 on the basis ofHPV infection rather than histology showed that 63 of 95 (66-3%) of the HPV positive samples contained HLA DQw3 alleles, compared with 39 of 78 (50.0%) of the HPV negative samples (2 4'06; p < 0-05). Conclusions-There was a significant association between HLA DQw3 and cervical HPV infection. This may be because people with HLA DQw3 are less able to mount an effective immune response to HPV, which predisposes them to the development ofSCCC. (3 Clin Pathol 1994;47:1077-1081) Squamous cell cancer of the cervix (SCCC) accounts for over 450 000 new cases annually.' The epidemiology of SCCC has been extensively studied.2 Early age at first sexual intercourse and multiple sexual partners have been established as independent risk factors for this disease.34 These associations suggest that a sexually transmitted agent may have a role in the development of SCCC, and this is corroborated by studies showing a higher number of sexual partners for husbands of patients with SCCC, compared with healthy people,5 and evidence for the clustering of cervical and penile cancer.6 Herpes simplex virus (HSV) and human papillomavirus (HPV) have both been implicated as infectious agents involved in cervical carcinogenesis. Case control studies have shown a significantly higher incidence of antibodies to HSV in cases than controls, but whether this effect is secondary to the greater number of sexual partners among the cases of SCCC is controversial.78 Papillomaviruses cause benign warts, and Franceschi et al showed that patients with genital warts were more likely to have cervical dyskaryosis than women with trichomoniasis or gonorrhoea.9 Since this report, HPV has been detected in 75-100% of SCCC cases, depending on the methoa used'0 11; the strongest associations are for certain "high risk" HPV subtypes (HPV 16, 18, 31, 33, 35).12 The incidence of high risk HPV types in cervical tissue increases with increasing dysplasia of the cervical epithelium,"3 but it is clear that HPV infection is not the only factor in the development of cervical cancer. Between 10 and 20% of women without cervical abnormalities have cervical infection with "high risk" HPV subtypes,'4 15 and a proportion of women with cervical dysplasia and HPV spontaneously revert to normal histology, while some progress to SCCC.'6 Recently, an increased prevalence of HLA DQw3 was demonstrated in a group of patients with SCCC, compared with a normal population.'7 As the HLA system is known to have an important role in viral clearance'8 and may be involved in possible immune surveillance against certain neoplasms,'9 the association between DQw3 and SCCC could be a result ofDQw3 positive subjects having either (a) a relative inability to clear HPV infection; or (b) a relative inability to provide immune surveillance against dysplastic tissue. To investigate these possibilities, we developed a rapid system for DQw3 typing based on primer directed mutagenesis and restriction digestion.