In Reply We appreciate Dr Franceschi’s Letter to the Editor regarding our article. In the light of associations that we observed between several oral β- and γ-human papillomaviruses (HPVs) and risk of head and neck squamous cell carcinoma (HNSCC), Dr Franceschi requested that analyses of oral β- or γ-HPVs and risk of HNSCC be stratified by smoking status.
First, we would like to point out that the sample size of our study is relatively small (132 cases of incident HNSCC and 396 controls) for stratified analysis by smoking habits or by alcohol consumption, which are the other 2 main risk factors for HNSCC.1,2 However, given this request, we now present data stratified by never vs ever smoking for any α-, β-, or γ-HPVs, and specifically for HPV16, β1-HPV5 type, and any γ11 and γ12 species that were statistically significantly associated with risk of HNSCC (Table). Among controls, the prevalence of oral α-, β-, or γ-HPVs was similar between never and ever smokers. Although the prevalence of any γ12 species was slightly higher in never vs ever smoker controls (7% vs 4%, respectively), none of the differences observed for prevalence of oral HPVs were statistically significant at P < .05. In addition, as noted in our article, we did not detect any statistically significant multiplicative interactions between oral β- or γ-HPVs and smoking status with risk of HNSCC. Because most of our HNSCC cases were ever smokers (86%), the conditional logistic regression models had low statistical power among the never smokers. We also stratified the data by alcohol consumption (never or <1 vs ≥1 drinks/wk) and observed similar results (data not shown). The stratified analyses by smoking habits or alcohol consumption in our data suggest that oral HPVs act as independent risk factors for HNSCC overall. However, larger sample size studies are needed to better explore the potential interaction that may exist between oral β- or γ-HPVs, smoking status, pack-years of smoking, or alcohol consumption and risk of HNSCC.