Vaccination against human papillomavirus

A recent meta-analysis in CMAJ about prophylactic vaccination against HPV reported a reduction in the frequency of high-grade cervical lesions caused by vaccine-type HPV strains compared with control groups: Peto odds ratio 0.14 (95% confidence interval [CI] 0.09–0.21) from combined per-protocol analyses and 0.52 (95% CI 0.43–0.63) from modified intention-to-treat analyses.1 The magnitude and statistical significance of the difference between per-protocol and modified intention-to-treat analyses speak to the issues involved in translating efficacy to effectiveness.

Even more uncertainty abounds when translating results from the controlled settings of randomized trials to the real world. As most cases of cervical cancer occur in women who have not undergone preventive Pap smear screening, an enhanced public health program, possibly with mandatory screening and improved educational initiatives, may well attain health benefits equal or superior to those attainable with a generalized vaccination program, with better cost-effectiveness. This, of course, remains to be studied.

Although Lisa Rambout and colleagues provide a clear justification for their use of surrogate end points,1 the use of such outcomes does mandate a word of caution. Here lessons learned in cardiology 30 years ago may be pertinent. The association of premature ventricular beats with adverse outcomes following myocardial infarction was firmly established, like the causal association of oncogenic HPV strains, high-grade lesions and cervical cancer. Moreover, certain antiarrhythmic drugs were shown to suppress this ventricular ectopy, much as the HPV vaccine has been shown to decrease the risk for high-grade cervical lesions. However, later randomized trials showed that these antiarrhythmic drugs were associated not with an improved survival rate, but rather with a worsening one. These points would appear to reinforce the sagacious message of the commentary by Abby Lippman and colleagues that careful evaluation of the evidence, much still lacking, is required before intelligent decisions regarding public policy can be made.2