Cervical cancer incidence can increase despite HPV vaccination — Author's reply ( BTW- paid for by GlaxoSmithKline makers of Cervarix)

Our Personal View1 presents a mathematical model to determine the conditions under which a decline in screening adherence in vaccinated women would outweigh the benefits of a human papillomvirus (HPV) immunisation programme, and thus result in a net increase in cervical cancer incidence. In their refutation, Diane Harper and colleagues’2 create a misrepresentation of our model and then refute that misrepresentation. First, Harper and colleagues interpret our model as showing that human papillomavirus (HPV) vaccination will prevent any future increase in incidence of cervical cancer. We find that such an increase is indeed possible, especially when screening coverage is high before vaccine introduction, however, it does not occur unless the decline in screening adherence is steep. Second, Harper and colleagues incorrectly cite the opportunistic screening scenario; for comparison to the situation in Finland, for which the organised screening scenario (high coverage) should be cited. Our model1 predicts that a rise in cervical cancer incidence in vaccinated women is more likely with organised screening than with opportunistic screening, and so data from Finland might actually validate our model.

Harper and colleagues use data showing increasing cervical cancer incidence in Finland and the UK, either directly related to declining screening coverage or despite rising screening coverage. They conclude that incidence of cervical cancer can increase in vaccinated populations and our model must therefore be wrong; however, this reasoning is also questionable. The data used are from the pre-vaccine era when there was no other public health interventions to counterbalance surges in cervical cancer cases caused by reduced screening adherence or changes in sexual behaviour. Of course cervical cancer incidence will increase in such conditions. Our model was devised to understand when immunisation would be sufficient to outweigh the effects of a decline in screening adherence in vaccinated women.

Every model makes assumptions and ours is no exception. It assumes that cervical cancer incidence is constant or already declining before the HPV vaccine is introduced. This is true for many countries,3, 4 but the data presented by Harper and colleagues show that this assumption does not apply to certain age groups in Finland. To address this example adequately, our model would have to include a scenario of a pre-existing rise in incidence before vaccine introduction. Even without this inclusion, the organised screening scenario of our model suggests that Finland might be more vulnerable than other countries to increases in cervical cancer incidence in vaccinated women who reduce their cervical screening adherence. This insight—together with the pre-vaccine data presented by Harper and colleagues—suggests that HPV immunisation programmes implemented in Finland should be monitored closely to ensure that this increase in incidence does not occur.