Human papillomavirus as a risk factor for oral squamous cell carcinoma: A meta-analysis, 1982-1997

Objective. Human papillomavirus (HPV) infection is a significant risk facto r for uterine cervical carcinoma. However, the role of HPV infection in ora l squamous cell carcinoma (OSCC] is less well defined. To determine the sig nificance of the relationship of this virus in the progressive development of oral cancer, we estimated the risk of HPV detection in normal oral mucos a, precancerous oral tissue, and oral carcinoma using meta-analysis. Study design. Case reports and clinical series published in English-languag e journals were retrieved by searching MEDLINE (January 1980-August 1998). Review articles were also examined to identify additional studies. Studies that used biochemical, immunologic, microscopic, or molecular analyses to d etect HPV in tissue or cells derived from normal oral mucosa (n = 25), beni gn leukoplakia (n = 21), intraepithelial neoplasia tie, dysplasia and carci noma in situ; n = 27), and oral cancer (n = 94) were included in the meta-a nalysis. Information on sample size, age, sex, method of tissue preservatio n (ie, fresh, frozen, paraffin-embedded), assay, primer amplification regio n (early, late), high-risk versus low-risk genotype, and use of tobacco or alcohol was abstracted by one author (C.S.M.). Results. Data from 94 reports that analyzed 4680 samples were included in t he meta-analysis. Analyses made by means of a random-effects model with and without adjustments for assay sensitivity showed increased probability of HPV detection in tissue with precancerous and cancerous features compared w ith normal mucosa. The likelihood of detecting HPV in normal oral mucosa (1 0.0%; 95% confidence interval [CI], 6.1%-14.6%) was significantly less than of detecting benign leukoplakia (22.2%; 95% Cl, 15.7%-29.9%), intraepithel ial neoplasia (26.2%; 95% CI, 19.6%-33.6%), verrucous carcinoma (29.5%; 95% CI, 23%-36.8%), and OSCC (46.5%; 95% CI, 37.6%-55.5%). Adjustment of findi ngs For differences in assay sensitivity indicated that these estimates may be conservative. Overall, HPV was between 2 and 3 times more likely to be detected in precancerous oral mucosa and 4.7 times more likely to be detect ed in oral carcinoma than in normal mucosa. The pooled odds ratio for the s ubset of studies directly comparing the prevalence of HPV in normal mucosa and OSCC was 5.37, confirming the trend observed in the overall sample. The probability of detecting high-risk HPVs in OSCCs was 2.8 times greater tha n that of low-risk HPVs. Conclusion. This mete-analysis indicates that HPV is detected with increase d frequency in oral dysplastic and carcinomatous epithelium in comparison w ith normal oral mucosa. The findings provide further quantitative evidence that oral infection with HPV, particularly with high-risk genotypes, is a s ignificant independent risk factor fur OSCC.