Response to: Demarteau N, Standaert B. Modelling the economic value of cross- and sustained-protection in vaccines against cervical cancer. J Med Econ 2010;13:324–38

Sir

Demarteau and Standaert1, have developed a comprehensive model to assess the impact of different levels of cross-protection and of the duration of vaccine efficacy on the cost-effectiveness of human papillomavirus (HPV) vaccination. The Demarteau and Standaert model is based on two hypothetical HPV vaccines (designated Vaccine A and Vaccine B) that have the essential characteristics of the existing licensed bivalent (Vaccine A) and quadrivalent (Vaccine B) vaccines. This model generates several results given different scenarios that can be useful for the readers of your journal or healthcare decision makers. We are concerned, however, by some of the assumptions used in the model which are not supported by evidence. We believe that these assumptions may have resulted in some misleading conclusions.

First, in the case of considering cross-protection within the cost-effectiveness analysis as done by Demarteau and Standaert, cross-protection has to be assumed for both vaccines where a class-effect against HPV 31 can be considered. There is no scientific evidence to consider cross-protection efficacy for the bivalent vaccine only. The cross-protection data cited for the bivalent vaccine (64% against cervical intraepithelial neoplasia (CIN) 2+ lesions related to ten non-vaccine HPV types)2 may be an overestimate. Analyses undertaken in response to a request from the Food and Drug Administration in the USA concluded that the efficacy against CIN2+ due to any of 12 non-vaccine HPV types was 37%3. In addition, the product information of the bivalent vaccine4 states that HPV 31 is the only non-vaccine HPV type for which consistent cross-protection has been demonstrated. For the quadrivalent vaccine, statistically significant efficacy against disease was demonstrated against HPV types phylogenetically related to HPV 16 (primarily HPV 31) with statistical significance for the individual HPV types reached for HPV 31⁵. Importantly, no direct comparison of the cross-protection effect between the bivalent and the quadrivalent vaccine has been carried out. Furthermore, it is not known if cross-protection is long-lasting.

Second, the evidence to support the proposition that sustained protection is provided only by the bivalent vaccine is flawed. The head-to-head comparison of the immunogenicity of the bivalent and quadrivalent vaccines involved assays that were based on the virus-like particles included in the bivalent vaccine6, and so may have biased the results in its favour. Furthermore, it should be emphasised, as the authors stated in the Introduction section of their article, that there is no immune correlate for clinical efficacy against HPV disease. Clinical efficacy against cancers and precancerous lesions of the cervix, vagina and vulva is the only basis for comparison of HPV vaccines. Evidence to date suggests that if the HPV 16 component of the quadrivalent vaccine is administered as a monovalent vaccine it remains effective after an average of 8.5 years post-immunisation7.

Third, some of the data used to generate the modelled outcomes may also confound the results. For example, Demarteau and Standaert state that the efficacy of both vaccines ‘was conservatively set’ at 95%. However, the actual efficacy rates observed in phase III clinical trials were higher with the quadrivalent vaccine (efficacy in the per-protocol group was: 99%, 96% and 99% against HPV 6/11/16/18-related CIN2/3, CIN1 and genital warts, respectively8,9), and lower with the bivalent vaccine (efficacy in the according to protocol group was: 92% and 93% against HPV 16/18-related CIN1+ and CIN2/3, respectively3,10.

Fourth, the model proposed by Demarteau and Standaert also underestimates the burden of genital warts. It assumes that the prevalence of HPV 6/11 in genital warts is 76%, but the study by Aubin et al.11 used to support this assumption found a cumulative prevalence of 83%, whereas the commonly accepted prevalence value is 90%12. More specifically, the projected lifetime numbers of cases of genital warts in Italy without vaccination generated by the model (8563 per 100,000) appear low compared with the estimated annual incidence rate of 4.3 per 1000 women reported in a recent retrospective, observational analysis13. Furthermore, the costs for the management of genital warts in Italy that are used in the model (€144) are lower than the direct medical costs for management of genital warts in women given in the supporting reference (€332)14. Finally, one of the references cited in support of the risk of developing genital warts among women in Ireland15 relates to the costs of screening for cervical cancer and management of CIN in Italy.

Therefore, we think that the data used for vaccine efficacy and burden of genital warts may have biased the results in favour of the bivalent vaccine.

Taken together, these issues make it difficult for the reader to accept the conclusion that ‘Better cross- and longer sustained-protection against oncogenic HPV types in one vaccine could completely offset the clinical and financial benefits of protection against genital warts by the other vaccine’. The model does show that, under baseline conditions, the quadrivalent vaccine is superior to the bivalent one on the basis of efficacy and cost effectiveness. Current evidence supports the greater cost effectiveness of the quadrivalent vaccine16–19.

Transparency

The author has disclosed that he has received grant/research funds from and is a consultant/adviser for Sanofi Pasteur MSD and GlaxoSmithKline and that he is on the Speakers’ Bureau of Sanofi Pasteur MSD.

Yours sincerely

Professor Dr. med. Peter Hillemanns

Department of Obstetrics and Gynecology

Hannover Medical School

Hannover, Germany

[email protected]

References

• Demarteau N, Standaert B. Modelling the economic value of cross- and sustained-protection in vaccines against cervical cancer. J Med Econ 2010;13:324-38
   PubMedGoogle Scholar
• Skinner R, Apter D, Chow SN, et al. Cross-protective efficacy of Cervarix against oncogenic HPV types beyond HPV-16/18. 25th International Papillomavirus Conference, Malmö, Sweden, 8–14 May 2009. Abstract O-29.01
   Google Scholar
• Cervarix, GlaxoSmithKline Biologicals. FDA Briefing Document. Vaccines & Related Biological Products Advisory Committee Meeting 9 September 2009. Available at: http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/BloodVaccinesandOtherBiologics/VaccinesandRelated BiologicalProductsAdvisoryCommittee/UCM181425.pdf. Accessed 17 December 2010
   Google Scholar
• Cervarix. Summary of Product Characteristics. GlaxoSmithKline Biologicals. 6 August 2010. Available at: http://www.medicines.org.uk/emc/medicine/20204/SPC/Cervarix/
   Google Scholar
• Gardasil. Summary of product characteristics. Sanofi Pasteur MSD. August 2010. Available at: http://www.medicines.org.uk/emc/document.aspx?documentid=19016
   Google Scholar
• Einstein MH, Baron M, Levin MJ, et al. Comparison of the immunogenicity and safety of Cervarix and Gardasil human papillomavirus (HPV) cervical cancer vaccines in healthy women aged 18–45 years. Hum Vaccin 2009;5:705-19. Available at: http://www.landesbioscience.com/journals/vaccines/article/9518
   Google Scholar
• Rowhani-Rahbar A, Mao C, Hughes JP, et al. Longer term efficacy of a prophylactic monovalent human papillomavirus type 16 vaccine. Vaccine 2009;27:5612–19. Available at: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2749988/?tool=pubmed
   Google Scholar
• The FUTURE II Study Group. Effect of prophylactic human papillomavirus L1 virus-like-particle vaccine on risk of cervical intraepithelial neoplasia grade 2, grade 3, and adenocarcinoma in situ: a combined analysis of four randomised clinical trials. Lancet 2007;369:1861-8
   PubMedGoogle Scholar
• The FUTURE I/II Study Group. Four year efficacy of prophylactic human papillomavirus quadrivalent vaccine against low grade cervical, vulvar, and vaginal intraepithelial neoplasia and anogenital warts: randomised controlled trial. Br Med J 2010;340:c3493
   Google Scholar
• Paavonen J, Naud P, Salmerón J, et al. Efficacy of human papillomavirus (HPV)-16/18 AS04-adjuvanted vaccine against cervical infection and precancer caused by oncogenic HPV types (PATRICIA): final analysis of a double-blind, randomised study in young women. Lancet 2009;374:301-14 [Erratum in: Lancet 2010;376:1054]
   PubMed Web of Science ®Google Scholar
• Aubin F, Prétet JL, Jacquard AC, et al. Human papillomavirus genotype distribution in external acuminata condylomata: a large French national study (EDiTH IV). Clin Infect Dis 2008;47:610-15. Available at: http://cid.oxfordjournals.org/content/47/5/610.long
   Google Scholar
• von Krogh G. Management of anogenital warts (condylomata acuminate). Eur J Dermatol 2001;11:598-603. Available at: http://www.john-libbey-eurotext.fr/en/revues/medecine/ejd/e-docs/00/01/86/B6/article.phtml
   Google Scholar
• Vittori G, Matteelli A, Boselli F, et al. A new approach to estimate genital warts incidence and prevalence in the Italian general female population. It J Gynaecol Obstet 2008;20:33-42. Available at: http://www.sigo.it/Documenti/Journal/italianjournal1_08.pdf
   Google Scholar
• Merito M, Largeron N, Cohet C, et al. Treatment patterns and associated costs for genital warts in Italy. Curr Med Res Opin 2008;24:3175-83. Available at: http://informahealthcare.com/doi/abs/10.1185/03007990802485694%20
   Google Scholar
• Giorgi Rossi P, Ricciardi A, Cohet C, et al. Epidemiology and costs of cervical cancer screening and cervical dysplasia in Italy. BMC Public Health 2009;9:71. Available at: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2651166/?tool=pubmed
   Google Scholar
• Jit M, Choi YH, Edmunds WJ. Economic evaluation of human papillomavirus vaccination in the United Kingdom. Br Med J 2008;337:331-5. Available at: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2500202/?tool=pubmed
   Google Scholar
• Brisson M, Van de Velde N, De Wals P, et al. The potential cost-effectiveness of prophylactic human papillomavirus vaccines in Canada. Vaccine 2007;25:5399-408
   PubMed Web of Science ®Google Scholar
• Chesson HW, Ekwueme DU, Saraiya M, et al. Cost-effectiveness of human papillomavirus vaccination in the United States. Emerg Infect Dis 2008;14:244-51. Available at: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2600200/?tool=pubmed
   Google Scholar
• Dee A, Howell F. A cost–utility analysis of adding a bivalent or quadrivalent HPV vaccine to the Irish cervical screening programme. Eur J Public Health. 2010;20:213-19. Available at: http://eurpub.oxfordjournals.org/content/20/2/213.long
   Google Scholar

Response from Demarteau et al.

Nadia Demarteau, Michael Zaiac, Baudouin Standaert, GSK Biologicals, Wavre, Belgium

Address for correspondence: Nadia Demarteau, GlaxoSmithKline Biologicals, Rue de Fleming 20, 1300 Wavre, Belgium. Tel.: [phone]; [email protected]

Sir

We thank you very much for giving us the chance to respond to Prof. Hillemanns’ comments on our research paper. We very much welcome his comments as they give us the opportunity to further clarify our work. We will focus on the five major points Prof. Hillemanns made, which include the intent of the modelling exercise, the level of cross-protection and sustained-protection considered, the overall vaccine efficacy value selected, and the data used on genital warts.

First we want to reiterate as Prof. Hillemanns did, that this paper is about a modelling exercise with two hypothetical vaccines (a quadri- and a bivalent vaccine). Because the composition of each hypothetical vaccine is different (the bivalent vaccine uses a specific adjuvant and different VLPs) differences in cross- and sustained-protection might be expected which would lead to a different outcome for each vaccine1–3. Hence the economic question of how big a difference must be to compensate for the additional protection against low risk HPV from the hypothetical quadrivalent vaccine appears to be a critical one for decision makers. The level of cross- and sustained-protection is tested in our model over a wide range of combinations because of the uncertainty in those values due to differences in HPV type distributions, disease burden and management in different countries. However, we clearly reported in Table 3 that if there is no difference in cross- and sustained-protection between these two hypothetical vaccines, a quadrivalent vaccine will dominate the other for obvious reasons. Only in the Discussion section do we review published information on cross-protection for HPV vaccines on the market today for which Prof. Hillemanns has concerns. The intent of the exercise is therefore clearly about how large a difference in cross- and sustained-protection must be between the two hypothetical vaccines to be economically interesting and not, as Prof. Hillemanns suggests, to derive at definitive economic result about real vaccines.

Prof. Hillemanns reports that we only consider cross-protection for the hypothetical bivalent vaccine. Our model assumes a net difference in cross-protection between the two hypothetical vaccines that is the result of calculating the difference between each level of cross-protection in each vaccine shown as a practical example included in the Discussion section. An average point estimate of the net difference in cross-protection is around 25% in our model and not 64% as Prof. Hillemanns suggested. Moreover Prof. Hillemanns suggested that in the absence of having a head-to-head comparison there could be no difference in cross-protection between the two vaccines. One should, however, be aware that in the US label of Gardasil the vaccine has no cross-protection against HPV 31, while Cervarix does4,5 indicating there could be a perceived or real net difference in cross-protection between the two marketed vaccines supporting the effort to model how big such a difference hypothetical would have to be to be of economical value.

Moreover and provided that the vaccine efficacy point estimates are high, EMA has decided to accept 6-month persistent infection (PI) as a surrogate measure for cervical cancer to better evaluate the vaccine efficacy against non-vaccine HPV types. Using 6-month PI overcomes the issue of co-infection and the results presented in the SPC for Cervarix6 indicate significant results for three non-vaccine HPV types 31 [77.5% (95% CI 68.3–84.4)], 33 [45.1% (21.7–61.9)], and 45 [76.1% (59.1–86.7)]. The Gardasil SPC states a significant result for HPV 31 [55.6% (95% CI 26.2–74.1)] only7.

Finally and contrary to Prof. Hillemanns’ statement, because we do not know, from real life data, if protection or indeed cross-protection will last forever, we linearly over time decreased the vaccine efficacy for HPV 18 and for all cross-protection HPV types as mentioned in the Methods section on p. 328.

The same reasoning has been applied with regard to sustained protection where the net difference between the two products is included in the model and not an all-or-nothing approach, suggested by Prof. Hillemanns. In addition different scenarios were included in the sensitivity analysis to assess the effect of different baseline sustained protection values. Meanwhile, Prof. Hillemanns’ view on the assumption of a potential difference in sustained protection is that it cannot be supported by real data because the head-to-head comparison of immunogenicity over time is biased as the assay used for that evaluation favours Cervarix.

In fact the lead assay used for comparing the immunogenicity of Gardasil and Cervarix was PBNA (developed by NIH in the US). ELISA (GSK) was used as a secondary assay and it correlated well with the PBNA. ELISA (GSK) also correlated well with the Merck CLIA/CLIC. For both HPV 16 and HPV 18 antigens, Cervarix recipients showed much higher antibody titres than Gardasil recipients independently of the assay tested. No bias was thus introduced by using either assay when comparing sera from individuals receiving either vaccine8. To our current understanding there is therefore no evidence for Prof. Hillmanns’ assertion that the assay technology used has biased results of the cited comparative immunogenicity study towards Cervarix.

Prof. Hillemanns states that the clinical efficacy against cancers and precancerous lesions of the cervix, vagina and vulva is the only basis for comparison of HPV vaccines. We agree that this is the best basis for a comparison but in the absence of data, or indeed the poor feasibility of comparing efficacy on these endpoints, other parameters such as persistent infection (see above) and immunogenicity are being discussed and used. The authors of the comparative study cited by Prof. Hillemanns conclude: Although the importance of differences in magnitude of immune response between these vaccines is unknown, they may represent determinants of duration of protection against HPV-16/18. Long-term studies evaluating duration of efficacy after vaccination are needed for both vaccines9.

The 8.5 years’ efficacy data have only been shown for an unlicensed HPV 16 monovalent vaccine10. These data could however be significantly different from those of a licensed formulation and cannot therefore be used to extrapolate to any other HPV type for long term efficacy or immunogenicity.

Modelling different durations of protection overall and between two hypothetical vaccines based on the above data appears therefore justifiable rather than misleading.

Regarding the selection value of overall vaccine efficacy for both vaccines, it is known and reported that the Cervarix efficacy value against CIN2+ lesions at the end of the study analysis (a time-point comparable to the Gardasil data quoted by Prof. Hillemanns) varies between 94.9 and 98.9% depending which analysis and cohort is selected 11. Meanwhile many independent evaluators of HPV vaccines in economic assessments have assumed no difference in vaccine type efficacy on HPV 16 and 1812–14. Hence our assumption of 95% being a conservative efficacy estimate for both vaccines against 16/18 causing lesions, appears justifiable.

Prof. Hillemanns asserts that the data on genital warts (GW) selected in the paper are biased towards favouring the results of the hypothetical bivalent vaccine. The HPV 6/11 prevalence rate in genital warts of 90% commonly cited as Prof. Hillemanns is referring to, is a study based on 41 patients15. We thought it would be better to refer to studies with a much larger sample size to increase the credibility of the data selected. So the study of Aubin et al. we refer to has a much larger number included (10 times more) with a precise figure for women only of 76% prevalence rate16. We could have added another study of Vandepapeliere17which also reports a lower rate of HPV 6/11 in genital warts of 72% with a sample size again of more than 400 women screened17,18. Also the modelled incidence was validated against observed incidence (see Figure 2 in the published article). Meanwhile the data selected by Prof. Hillemanns comes from a cross-sectional population study and should be adjusted for a cohort model that is different in its demographic composition19.

Finally, regarding the cost of genital warts in Italy, our study uses the Merito cost (€332) further adjusted to the proportion of GW actually treated in the public sector and therefore charged to the Italian NHS20,21.

The references for the level of regression from HPV-LR to GW in Ireland (Table 1) were indeed mistakenly included. Instead of being 54–56 it should be 54–55. Many thanks for noticing that error.

Having demonstrated that all assumptions for our hypothetical vaccines have a foundation we believe that the model constructed and tested, fully supports the conclusion of the paper that ‘Better cross- and longer sustained-protection against oncogenic HPV types in one vaccine could completely offset the clinical and financial benefits of protection against genital warts by the other vaccine’ for two hypothetical vaccines

Transparency

The authors have disclosed that the sponsor of the original study on which this response to the letter to the editor is based was supported by GlaxoSmithKline Biologicals. N.D. and B.S. have disclosed that they are employees of GlaxoSmithKline Biologicals, Rixensart, Belgium.

Yours sincerely,

Nadia Demarteau, Michael Zaiac, Baudouin Standaert

GSK Biologicals, Wavre, Belgium

Address for correspondence:

Nadia Demarteau, GlaxoSmithKline Biologicals, Rue de Fleming 20, 1300 Wavre, Belgium.

Tel.: [phone]; [email protected]

Michael Zaiac

Email: [email protected]

Baudouin Standaert

Email: [email protected]

Cervarix is trademark of the GSK group of companies

Gardasil is a trademark of the Merck group of companies