Reply to Van Howe

To the Editor—We thank Van Howe [1] for his interest in our article on the effect that male circumcision has on high-risk human papillomavirus (HR-HPV) prevalence among young men [2]. In his letter, he raises an important issue about the yield of HPV culture, which is dependent on the penile site sampled

We agree with Van Howe that detection of HPV at the glans and corona sulcus differs significantly between circumcised and uncircumcised men [3–5 ]. Indeed, because of the proximity between the glans and the foreskin, which presents an inner mucosal surface vulnerable to HPV infection, autoinoculation can occur: HPV acquired via the foreskin during sexual intercourse can be subsequently transmitted to the glans and corona sulcus on contact. Apart from the glans and corona sulcus, there is no clear indication from available reports of any other anatomical site that would yield differing HPV detection between circumcised and uncircumcised men [3, 4]

In our study, we collected samples only from the urethra. We chose the urethral site for the detection of HR-HPV specifically because there was no anatomical reason that could explain a differential effect of circumcision status on the detection of HPV or associated lesions at this site [6]. To ensure that the detection of HR-HPV was not affected by circumcision status, we compared the prevalence of HR-HPV among 371 participants of the control group who were enrolled in a nested study designed to compare 2 circumcision methods. These participants underwent urethral swab sampling before and after male circumcision. The average (median) duration between the 2 swab-sample collections was 59 (43) days. HR-HPV prevalence did not differ between the 2 samplings (23.7% vs 23.9%; P=.99, sign test). This clearly demonstrates that the as-treated effect of male circumcision on HR-HPV prevalence reported in our study (adjusted prevalence rate ratio, 0.68 [95% confidence interval {CI}, 0.52–0.89]) cannot be attributed to detection bias for HR-HPV by urethral swab sampling in uncircumcised men. Moreover, this result suggests that there is no risk of differential misclassification between the 2 groups

Finally, we stated in our article that urethral sampling was likely to miss infections [7]. Thus, the prevalence of HR-HPV infections in our cohort is likely to be underestimated. However, HR-HPV infections would be underestimated equally in the 2 randomization arms, and this underestimation would have no effect on the findings. Despite the loss of power, our study found that male circumcision has a significant protective effect against HPV infection, which reinforces the robustness of our findings

We believe that it is a mistake to apply HPV detection rates from one study to another to compensate for supposed underdetection of HPV among circumcised men caused by not sampling other penile sites, such as the shaft or the glans. In his compensation calculation for HPV detection rates, Van Howe uses only data on the detection of HPV on the glans. Incidentally, using the same data from Weaver et al [3] but for detection of HPV on the penile shaft, we found that, among men who had HPV recovered from any location, the results of penile shaft culture were positive for 77% of circumcised men and 53% of uncircumcised men. This means that an infected man with a circumcised penis for whom only a sample from the penile shaft was cultured would be 1.45 times more likely to have a positive culture result than would an infected man with an uncircumcised penis. According to Van Howe’s method, the adjustment of our results for this difference in yield reveals a statistically significant protective effect of a magnitude greater than that reported in our study (intention-to-treat odds ratio, 0.36 [95% CI, 0.27–0.48; as-treated odds ratio, 0.34 [95% CI, 0.26–0.45]). Hence, it is clear that, if there is a sampling bias, it is not unidirectional, unlike what Van Howe argues. Use of such a method is certainly a source of bias, because the populations studied are different. This method, which has been previously used by Van Howe in a meta-analysis of HPV and circumcision [8], has been strongly criticized elsewhere [9]

The results of our study, obtained among young men in a country in which circumcision prevalence is low, demonstrates that circumcision protects against HR-HPV acquisition. With the generation of consistent data on the protective effect that circumcision has on HPV infection [2, 5, 10, 11], circumcision should be regarded as a possible method to prevent HPV infection, one that should be implemented in developing countries in which HPV vaccination—notably for men—is either not available or not accessible

Potential conflicts of interest: none reported

Financial support: none reported