DOI:10.1093/jnci/djt460 First published online February 19, 2014
© The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected].
Article
Early Impact of Human Papillomavirus Vaccination on Cervical Neoplasia—Nationwide Follow-up of Young Danish Women Birgitte Baldur-Felskov, Christian Dehlendorff, Christian Munk, Susanne K. Kjaer Manuscript received July 26, 2013; revised November 27, 2013; accepted December 9, 2013. Correspondence to: Susanne Krüger Kjaer, MD, Virus, Lifestyle and Genes, Danish Cancer Society Research Center, DK-2100 Copenhagen, Denmark (e-mail: [email protected]).
In clinical trials, vaccines against human papillomavirus (HPV) have been highly effective against HPV16- or HPV18-associated cervical lesions. The quadrivalent HPV vaccine was licensed in 2006 and subsequently implemented in the Danish vaccination program. The study aim was to use individual information on HPV vaccination status to assess subsequent risk of cervical lesions.
Methods
Using a cohort study design, we identified all girls and women born in Denmark in the period from 1989 to 1999 and obtained information on individual HPV vaccination status in the period from 2006 to 2012 from nationwide registries. Incident cases of cervical lesions were identified by linkage to the nationwide Pathology Data Bank. We compared vaccinated and unvaccinated girls and women stratified by birth cohort in Cox proportional hazards models.
Results
Risk of atypia or worse (atypia+) and of cervical intraepithelial neoplasia grade 2 or 3 (CIN2/3) were statistically significantly reduced among vaccinated women in birth cohorts 1991 to 1994 (1991–1992atypia+: hazard ratio [HR] = 0.46, two-sided 95% confidence interval [CI] = 0.39 to 0.56; 1991–1992CIN2/3: HR = 0.56, 95% CI = 0.37 to 0.84; 1993–1994atypia+: HR = 0.40, 95% CI = 0.29 to 0.56; 1993–1994 CIN2/3: HR = 0.27, 95% CI = 0.10 to 0.67). The birth cohort 1989 to 1990 had a statistically significantly reduced risk of atypia+ (HR = 0.75; 95% CI = 0.65 to 0.86); the risk of CIN2/3 was also decreased but not statistically significant. No events occurred among girls in the birth cohort 1997 to 1999, whereas for the birth cohort 1995 to 1996 a hazard ratio could be calculated only for atypia+.
Conclusions
Six years after licensure of the quadrivalent HPV vaccine in Denmark, a reduced risk of cervical lesions is observed at the population level.
JNCI J Natl Cancer Inst (2014) 106(3): djt460 doi:10.1093/jnci/djt460
Two vaccines are currently available against human papillomavirus (HPV) (1,2), the main etiological factor for cervical cancer (3). The bivalent vaccine contains viruslike particles of HPV types 16 and 18 (1), whereas the quadrivalent vaccine contains viruslike particles of HPV types 6, 11, 16, and 18 (2). Both vaccines have been shown to be highly effective against HPV16/18–associated cervical intraepithelial neoplasia grades 2 and 3 (CIN2/3) and adenocarcinoma in situ, endpoints accepted in trials for vaccine efficacy against cervical cancer (4–8). The quadrivalent HPV vaccine was licensed in Denmark in 2006. Subsequently, the vaccine was chosen by tender and included in the free-of-charge general childhood vaccination program for girls aged 12 years on January 1, 2009 (birth cohort 1996–1997). A free-of-charge catch-up vaccination program for girls aged 13 to 15-years was added in fall 2008 (birth cohorts 1993–1995) (9). Older girls and women and boys and men are not covered by the HPV vaccination program but can pay to receive the vaccine. HPV types are categorized into high risk and low risk based on their biological properties and oncogenic potential (10). HPV16 1 of 7 Article | JNCI
and HPV18 are the most common high-risk types, accounting for about 70% of all cervical cancer cases and about 50% of all CIN3 cases (10,11), whereas HPV6 and HPV11 are low-risk types associated with nearly 90% of all genital warts cases (12). Thus, it is estimated that HPV vaccination could prevent up to 70% of cervical cancer cases (13,14). Because progression from infection with a high-risk HPV to invasive cervical cancer can take decades (10), it may be many years before the effect of HPV vaccination on the incidence of cervical cancer can be assessed. Nevertheless, recent ecological studies have shown a decrease in the incidence of genital warts among young women since introduction of the vaccine (15–18). In another ecological study, Brotherton et al. (19) found a statistically significant decrease in the incidence of high-grade cervical lesions among girls aged less than 18 years in Australia, which began shortly after introduction of a nationally funded program for vaccination with the quadrivalent vaccine. An important limitation of these studies is that the data were not analyzed by individual vaccination status. To our knowledge, data on individual vaccination status was used in only four studies (20–23) besides the studies Vol. 106, Issue 3 | djt460 | March 12, 2014
Downloaded from http://jnci.oxfordjournals.org/ at Purdue University Libraries ADMN on June 13, 2015
Background
based on the vaccination trials. All showed a protective effect of vaccination. However, so far no nationwide population-based study has been published on the occurrence of cervical abnormalities in vaccinated and nonvaccinated women based on information about vaccination status on the individual level. The aim of the study reported herein was to determine the association between individual HPV vaccination status and subsequent risk of cervical lesions in a nationwide study.
Methods
Collection of Socioeconomic Data Lastly, we investigated socioeconomic factors for vaccinees and nonvaccinees. To obtain this information, our cohort was linked with data from the population-based Statistics Denmark. Socioeconomic position variables comprised mothers’ highest attained education at the beginning of follow-up (basic school/ high school, vocational education, higher education, unknown) and disposable income (categorized in tertiles). Statistical Analysis A Cox proportional hazard regression model stratified by five birth cohorts (1989–1990, 1991–1992, 1993–1994, 1995–1996, and 1997– 1999) was used to estimate hazard ratios (HRs) and corresponding
Figure 1. Study design. HPV, human papillomavirus. jnci.oxfordjournals.org
JNCI | Article 2 of 7
Downloaded from http://jnci.oxfordjournals.org/ at Purdue University Libraries ADMN on June 13, 2015
Cohort Data Collection The study design is shown in Figure 1. In Denmark, all citizens are assigned a unique personal identification number, which is used throughout society, including in all health registries, and is registered in the computerized Civil Registration System, which was established in 1968 (24). From this system, we identified all women born in the period from 1961 to 1999 and living in Denmark and obtained information on their date of emigration, death, or disappearance. To obtain information on HPV vaccination status, we linked this cohort with the National Health Service Registry and the Danish National Prescription Registry. The National Health Service Registry holds data since 1990 on the activities of primary health professionals and includes information about citizens, providers, and health services (25). From this registry, we identified girls who had received HPV vaccination within the general childhood vaccination or catch-up program since it started in 2008 up to July 1, 2011. Data on all girls and women who purchased an HPV vaccine on prescription between October 1, 2006, and March 1, 2012 were obtained from the Danish National Prescription Registry, which holds information on all prescription drugs sold in Danish community pharmacies since 1994 (26) and information on more than 90% of HPV vaccinations given outside the program. Hence, we were able to identify almost all women in each birth cohort who had been vaccinated and those who had not and the date of vaccination for each vaccinated woman. We included only birth cohorts with a vaccination coverage rate greater than 10%, corresponding to birth cohorts 1989 to 1999. Because we examined only the effect of the quadrivalent HPV vaccine used in the Danish program, we excluded women who had received the bivalent vaccine (n = 148).
Cytology and Assessment of Outcomes In Denmark, cervical cytology testing is recommended every third year for women aged 23 to 49 years and every fifth year for women aged 50 to 64 years (9), but opportunistic screening is also common. To identify cases of cervical lesions, the cohort was linked to the nationwide Danish Pathology Data Bank, established in 1999, which holds information on all cervical cytology (organized and opportunistic) and all cervical biopsies and cones from all Danish pathology departments (27). Abnormal cervical diagnoses are usually reported as atypia, mild dysplasia, moderate dysplasia, severe dysplasia, carcinoma in situ, or cancer. The Bethesda nomenclature (28) is increasingly being used for cytological diagnoses, whereas the CIN nomenclature is used for histological diagnoses, with moderate dysplasia categorized as CIN2 and severe dysplasia and carcinoma in situ as CIN3 (29). We assessed three outcomes. The first, atypia or worse, comprised cytological diagnoses of atypia or atypical squamous cells of undetermined significance or worse. The second outcome, CIN2/3, comprised histological diagnoses of CIN2, CIN3, carcinoma in situ (CIS), and adenocarcinoma in situ (AIS). The third outcome, CIN3, consisted of diagnoses of CIN3, CIS, and AIS. No cervical cancers were detected during follow-up. To identify all cervical lesions in the study period and before, we collected information on all cervical examinations (cytology, biopsy, cone) undergone by the women in our cohort registered in the Pathology Data Bank and excluded women who had any cervical abnormalities before study start (n = 72).
Results From the Civil Registration System, we identified 399 244 Danish women in the birth cohorts 1989 to 1999. Characteristics of this population are shown in Table 1. A total of 247 313 (62%) women
were vaccinated during the period. The vaccination coverage was high (>85%) in the youngest birth cohorts of 1993 to 1999, which were included in the childhood and catch-up vaccination program, and much lower in the birth cohorts not covered by the program (1989–1990: 14%; 1991–1992: 27%). The total number of events was 3629 for atypia or worse, 708 for CIN2/3, and 365 for CIN3. The women constituting the study population contributed more than 2 million person-years of follow-up in the analysis for each outcome. Overall, the event rate for atypia or worse was 2.4 times higher among unvaccinated women (0.22%) than among vaccinated women (0.09%) and that for CIN2/3 and CIN3 were both increased twice (nonvaccinees: 0.04%; vaccinees: 0.02%; nonvaccinees: 0.02%; vaccinees: 0.01%, respectively). Screenings rates for nonvaccinees and vaccinees were estimated separately using the respective number of screenings records and person-years. Subsequently we calculated screenings rate ratios (RRs) by dividing the screenings rates of vaccinees by that of the nonvaccinees and found that in all birth cohorts, vaccinees had higher screenings rates than nonvaccinees (1989–1990: RR = 1.55, 95% CI = 1.48 to 1.63, P < .001; 1991–1992: RR = 1.22, 95% CI = 1.14 to 1.30, P < .001; 1993–1994: RR = 3.75, 95% CI = 3.25 to 4.33, P < .001; 1995– 1996: RR = 8.87, 95% CI = 5.19 to 15.13, P < .001; 1997–1999: RR = 17.03, 95% CI = 1.99 to 143.76, P = .005). Table 2 shows the hazard ratios for cervical lesions among vaccinated women compared with unvaccinated women in the birth cohorts in the two models. There were too few events in birth cohort 1995 to 1996 to estimate hazard ratios for CIN2/3 or for CIN3, and hazard ratios could not be calculated for any outcome for the birth cohorts 1997 to 1999 because there were no events. In model 1 (Table 2), the risk of atypia or worse was statistically significantly lower among vaccinated women in the birth cohorts 1989 to 1994 (1989–1990: HR = 0.75, 95% CI = 0.65 to 0.86, P < .001; 1991–1992:
Table 1. Characteristics and number of events in the Danish female birth cohorts 1989–1999, October 2006 to March 2012 Birth cohort
Total population
1989–1990 1991–1992 1993–1994 1995–1996 1997–1999
78 448 74 323 72 544 71 214 102 715
Number of vaccinees, (%)*
Screened population
Atypia or worse
CIN2/3†
CIN3
11 243 (14.3) 19 867 (26.7) 63 922 (88.1) 63 912 (89.8) 88 369(86.0)
7718 3412 928 124 6
2395 1012 196 26 0
455 167 20 2 0
280 74 10 1 0
* Vaccination status attained at end of follow-up. † CIN2/3 = cervical intraepithelial neoplasia grade 2 or 3.
Table 2. Model 1: Risk for cervical lesions among human papillomavirus vaccinated women compared with nonvaccinated women in Danish birth cohorts 1989–1999, October 2006 to March 2012* Atypia or worse Birth cohort 1989–1990 1991–1992 1993–1994 1995–1996 1997–1999
CIN2/3
CIN3
HR (95% CI)
P†
HR (95% CI)
P†
HR (95% CI)
P†
0.75 (0.65 to 0.86) 0.46 (0.39 to 0.56) 0.40 (0.29 to 0.56) 0.43 (0.16 to 1.12) —
© The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected].
Article
Early Impact of Human Papillomavirus Vaccination on Cervical Neoplasia—Nationwide Follow-up of Young Danish Women Birgitte Baldur-Felskov, Christian Dehlendorff, Christian Munk, Susanne K. Kjaer Manuscript received July 26, 2013; revised November 27, 2013; accepted December 9, 2013. Correspondence to: Susanne Krüger Kjaer, MD, Virus, Lifestyle and Genes, Danish Cancer Society Research Center, DK-2100 Copenhagen, Denmark (e-mail: [email protected]).
In clinical trials, vaccines against human papillomavirus (HPV) have been highly effective against HPV16- or HPV18-associated cervical lesions. The quadrivalent HPV vaccine was licensed in 2006 and subsequently implemented in the Danish vaccination program. The study aim was to use individual information on HPV vaccination status to assess subsequent risk of cervical lesions.
Methods
Using a cohort study design, we identified all girls and women born in Denmark in the period from 1989 to 1999 and obtained information on individual HPV vaccination status in the period from 2006 to 2012 from nationwide registries. Incident cases of cervical lesions were identified by linkage to the nationwide Pathology Data Bank. We compared vaccinated and unvaccinated girls and women stratified by birth cohort in Cox proportional hazards models.
Results
Risk of atypia or worse (atypia+) and of cervical intraepithelial neoplasia grade 2 or 3 (CIN2/3) were statistically significantly reduced among vaccinated women in birth cohorts 1991 to 1994 (1991–1992atypia+: hazard ratio [HR] = 0.46, two-sided 95% confidence interval [CI] = 0.39 to 0.56; 1991–1992CIN2/3: HR = 0.56, 95% CI = 0.37 to 0.84; 1993–1994atypia+: HR = 0.40, 95% CI = 0.29 to 0.56; 1993–1994 CIN2/3: HR = 0.27, 95% CI = 0.10 to 0.67). The birth cohort 1989 to 1990 had a statistically significantly reduced risk of atypia+ (HR = 0.75; 95% CI = 0.65 to 0.86); the risk of CIN2/3 was also decreased but not statistically significant. No events occurred among girls in the birth cohort 1997 to 1999, whereas for the birth cohort 1995 to 1996 a hazard ratio could be calculated only for atypia+.
Conclusions
Six years after licensure of the quadrivalent HPV vaccine in Denmark, a reduced risk of cervical lesions is observed at the population level.
JNCI J Natl Cancer Inst (2014) 106(3): djt460 doi:10.1093/jnci/djt460
Two vaccines are currently available against human papillomavirus (HPV) (1,2), the main etiological factor for cervical cancer (3). The bivalent vaccine contains viruslike particles of HPV types 16 and 18 (1), whereas the quadrivalent vaccine contains viruslike particles of HPV types 6, 11, 16, and 18 (2). Both vaccines have been shown to be highly effective against HPV16/18–associated cervical intraepithelial neoplasia grades 2 and 3 (CIN2/3) and adenocarcinoma in situ, endpoints accepted in trials for vaccine efficacy against cervical cancer (4–8). The quadrivalent HPV vaccine was licensed in Denmark in 2006. Subsequently, the vaccine was chosen by tender and included in the free-of-charge general childhood vaccination program for girls aged 12 years on January 1, 2009 (birth cohort 1996–1997). A free-of-charge catch-up vaccination program for girls aged 13 to 15-years was added in fall 2008 (birth cohorts 1993–1995) (9). Older girls and women and boys and men are not covered by the HPV vaccination program but can pay to receive the vaccine. HPV types are categorized into high risk and low risk based on their biological properties and oncogenic potential (10). HPV16 1 of 7 Article | JNCI
and HPV18 are the most common high-risk types, accounting for about 70% of all cervical cancer cases and about 50% of all CIN3 cases (10,11), whereas HPV6 and HPV11 are low-risk types associated with nearly 90% of all genital warts cases (12). Thus, it is estimated that HPV vaccination could prevent up to 70% of cervical cancer cases (13,14). Because progression from infection with a high-risk HPV to invasive cervical cancer can take decades (10), it may be many years before the effect of HPV vaccination on the incidence of cervical cancer can be assessed. Nevertheless, recent ecological studies have shown a decrease in the incidence of genital warts among young women since introduction of the vaccine (15–18). In another ecological study, Brotherton et al. (19) found a statistically significant decrease in the incidence of high-grade cervical lesions among girls aged less than 18 years in Australia, which began shortly after introduction of a nationally funded program for vaccination with the quadrivalent vaccine. An important limitation of these studies is that the data were not analyzed by individual vaccination status. To our knowledge, data on individual vaccination status was used in only four studies (20–23) besides the studies Vol. 106, Issue 3 | djt460 | March 12, 2014
Downloaded from http://jnci.oxfordjournals.org/ at Purdue University Libraries ADMN on June 13, 2015
Background
based on the vaccination trials. All showed a protective effect of vaccination. However, so far no nationwide population-based study has been published on the occurrence of cervical abnormalities in vaccinated and nonvaccinated women based on information about vaccination status on the individual level. The aim of the study reported herein was to determine the association between individual HPV vaccination status and subsequent risk of cervical lesions in a nationwide study.
Methods
Collection of Socioeconomic Data Lastly, we investigated socioeconomic factors for vaccinees and nonvaccinees. To obtain this information, our cohort was linked with data from the population-based Statistics Denmark. Socioeconomic position variables comprised mothers’ highest attained education at the beginning of follow-up (basic school/ high school, vocational education, higher education, unknown) and disposable income (categorized in tertiles). Statistical Analysis A Cox proportional hazard regression model stratified by five birth cohorts (1989–1990, 1991–1992, 1993–1994, 1995–1996, and 1997– 1999) was used to estimate hazard ratios (HRs) and corresponding
Figure 1. Study design. HPV, human papillomavirus. jnci.oxfordjournals.org
JNCI | Article 2 of 7
Downloaded from http://jnci.oxfordjournals.org/ at Purdue University Libraries ADMN on June 13, 2015
Cohort Data Collection The study design is shown in Figure 1. In Denmark, all citizens are assigned a unique personal identification number, which is used throughout society, including in all health registries, and is registered in the computerized Civil Registration System, which was established in 1968 (24). From this system, we identified all women born in the period from 1961 to 1999 and living in Denmark and obtained information on their date of emigration, death, or disappearance. To obtain information on HPV vaccination status, we linked this cohort with the National Health Service Registry and the Danish National Prescription Registry. The National Health Service Registry holds data since 1990 on the activities of primary health professionals and includes information about citizens, providers, and health services (25). From this registry, we identified girls who had received HPV vaccination within the general childhood vaccination or catch-up program since it started in 2008 up to July 1, 2011. Data on all girls and women who purchased an HPV vaccine on prescription between October 1, 2006, and March 1, 2012 were obtained from the Danish National Prescription Registry, which holds information on all prescription drugs sold in Danish community pharmacies since 1994 (26) and information on more than 90% of HPV vaccinations given outside the program. Hence, we were able to identify almost all women in each birth cohort who had been vaccinated and those who had not and the date of vaccination for each vaccinated woman. We included only birth cohorts with a vaccination coverage rate greater than 10%, corresponding to birth cohorts 1989 to 1999. Because we examined only the effect of the quadrivalent HPV vaccine used in the Danish program, we excluded women who had received the bivalent vaccine (n = 148).
Cytology and Assessment of Outcomes In Denmark, cervical cytology testing is recommended every third year for women aged 23 to 49 years and every fifth year for women aged 50 to 64 years (9), but opportunistic screening is also common. To identify cases of cervical lesions, the cohort was linked to the nationwide Danish Pathology Data Bank, established in 1999, which holds information on all cervical cytology (organized and opportunistic) and all cervical biopsies and cones from all Danish pathology departments (27). Abnormal cervical diagnoses are usually reported as atypia, mild dysplasia, moderate dysplasia, severe dysplasia, carcinoma in situ, or cancer. The Bethesda nomenclature (28) is increasingly being used for cytological diagnoses, whereas the CIN nomenclature is used for histological diagnoses, with moderate dysplasia categorized as CIN2 and severe dysplasia and carcinoma in situ as CIN3 (29). We assessed three outcomes. The first, atypia or worse, comprised cytological diagnoses of atypia or atypical squamous cells of undetermined significance or worse. The second outcome, CIN2/3, comprised histological diagnoses of CIN2, CIN3, carcinoma in situ (CIS), and adenocarcinoma in situ (AIS). The third outcome, CIN3, consisted of diagnoses of CIN3, CIS, and AIS. No cervical cancers were detected during follow-up. To identify all cervical lesions in the study period and before, we collected information on all cervical examinations (cytology, biopsy, cone) undergone by the women in our cohort registered in the Pathology Data Bank and excluded women who had any cervical abnormalities before study start (n = 72).
Results From the Civil Registration System, we identified 399 244 Danish women in the birth cohorts 1989 to 1999. Characteristics of this population are shown in Table 1. A total of 247 313 (62%) women
were vaccinated during the period. The vaccination coverage was high (>85%) in the youngest birth cohorts of 1993 to 1999, which were included in the childhood and catch-up vaccination program, and much lower in the birth cohorts not covered by the program (1989–1990: 14%; 1991–1992: 27%). The total number of events was 3629 for atypia or worse, 708 for CIN2/3, and 365 for CIN3. The women constituting the study population contributed more than 2 million person-years of follow-up in the analysis for each outcome. Overall, the event rate for atypia or worse was 2.4 times higher among unvaccinated women (0.22%) than among vaccinated women (0.09%) and that for CIN2/3 and CIN3 were both increased twice (nonvaccinees: 0.04%; vaccinees: 0.02%; nonvaccinees: 0.02%; vaccinees: 0.01%, respectively). Screenings rates for nonvaccinees and vaccinees were estimated separately using the respective number of screenings records and person-years. Subsequently we calculated screenings rate ratios (RRs) by dividing the screenings rates of vaccinees by that of the nonvaccinees and found that in all birth cohorts, vaccinees had higher screenings rates than nonvaccinees (1989–1990: RR = 1.55, 95% CI = 1.48 to 1.63, P < .001; 1991–1992: RR = 1.22, 95% CI = 1.14 to 1.30, P < .001; 1993–1994: RR = 3.75, 95% CI = 3.25 to 4.33, P < .001; 1995– 1996: RR = 8.87, 95% CI = 5.19 to 15.13, P < .001; 1997–1999: RR = 17.03, 95% CI = 1.99 to 143.76, P = .005). Table 2 shows the hazard ratios for cervical lesions among vaccinated women compared with unvaccinated women in the birth cohorts in the two models. There were too few events in birth cohort 1995 to 1996 to estimate hazard ratios for CIN2/3 or for CIN3, and hazard ratios could not be calculated for any outcome for the birth cohorts 1997 to 1999 because there were no events. In model 1 (Table 2), the risk of atypia or worse was statistically significantly lower among vaccinated women in the birth cohorts 1989 to 1994 (1989–1990: HR = 0.75, 95% CI = 0.65 to 0.86, P < .001; 1991–1992:
Table 1. Characteristics and number of events in the Danish female birth cohorts 1989–1999, October 2006 to March 2012 Birth cohort
Total population
1989–1990 1991–1992 1993–1994 1995–1996 1997–1999
78 448 74 323 72 544 71 214 102 715
Number of vaccinees, (%)*
Screened population
Atypia or worse
CIN2/3†
CIN3
11 243 (14.3) 19 867 (26.7) 63 922 (88.1) 63 912 (89.8) 88 369(86.0)
7718 3412 928 124 6
2395 1012 196 26 0
455 167 20 2 0
280 74 10 1 0
* Vaccination status attained at end of follow-up. † CIN2/3 = cervical intraepithelial neoplasia grade 2 or 3.
Table 2. Model 1: Risk for cervical lesions among human papillomavirus vaccinated women compared with nonvaccinated women in Danish birth cohorts 1989–1999, October 2006 to March 2012* Atypia or worse Birth cohort 1989–1990 1991–1992 1993–1994 1995–1996 1997–1999
CIN2/3
CIN3
HR (95% CI)
P†
HR (95% CI)
P†
HR (95% CI)
P†
0.75 (0.65 to 0.86) 0.46 (0.39 to 0.56) 0.40 (0.29 to 0.56) 0.43 (0.16 to 1.12) —
