The authors have declared that no competing interests exist.
Current address: Department of Health Education, Federal University of Sergipe, Lagarto, Sergipe, Brazil
Human papillomavirus (HPV) is responsible for the rise in the incidence of cancer in the oropharynx, tonsils, and base of the tongue (i.e., HPV-related subsites). HPV triggered the changes in the epidemiology of oropharyngeal and oral cavity cancer (OPC/OCC) in Asia, Europe, North America, and Oceania. Hence, the incidence of cancer in HPV-related subsites is augmenting, while that in other HPV-unrelated subsites is decreasing. In South America, although the incidence of HPV-positive tumors has gradually increased, there is an atypically low prevalence of HPV in people with OPC/OCC. To clarify whether this dramatic shift in incidence trends also occurred in this population, we estimated the burden of HPV on the incidence trends of OPCs/OCCs in São Paulo city in Brazil. In this population-based study, we categorized OPCs/OCCs by HPV-related and HPV-unrelated subsites. We used Poisson regression to assess the age-standardized incidence rates (ASRs) stratified by sex and age groups, as well as to examine the age-period-cohort effects. There were 15,391 cases of OPCs/OCCs diagnosed in HPV-related (n = 5,898; 38.3%) and HPV-unrelated (n = 9,493; 61.7%) subsites. Overall, the ASRs decreased for most subsites, for both sexes and for all age groups, except for HPV-related OPC/OCC in young males and females, which increased by 3.8% and 8.6% per year, respectively. In the birth-cohort-effect analysis, we identified an increasing risk for HPV-related OPC/OCC in both sexes in recent birth cohorts; however, this risk was sharply decreased in HPV-unrelated subsites. Our data demonstrate an emerging risk for HPV-related OPC/OCC in young people, which supports prophylactic HPV vaccination in this group.
Human papillomavirus (HPV) is an oncogenic virus that is sexually transmitted, and it is the cause of an estimated 630,000 new cancers each year worldwide [
In the mid-2000s, there was a striking rise, from 40.5% to 72.2%, in the overall HPV prevalence of oropharyngeal tumors [
To the best of our knowledge, no population-based study has focused on the burden of HPV associated with OPC/OCC incidence in Brazil. Prophylactic HPV vaccinations were introduced in 2014 in an effort to hasten a decrease in HPV prevalence, but its need in oral cavity and oropharyngeal subsites at the population level was unknown. For these reasons, we assessed the incidence trends of OPC/OCC in HPV-related and HPV-unrelated subsites over the 15 years prior to large-scale HPV immunization in São Paulo city.
The Research Ethics Committee (nº 83218318.8.0000.5421) and the Technical Advisory Committee of the Population-based Cancer Registry in the city of São Paulo (RCBP-SP) approved our investigation. We accessed data on the RCBP-SP repository in July 2018 [
This population-based study included new cases of OPC/OCC diagnosed between 1997 and 2013 in the RCBP-SP database. In São Paulo city, the RCBP-SP records cancer data of more than 12 million residents from 301 sources, including hospitals, clinics, and death investigation services; this database is one of the oldest and largest cancer registries in Latin America [
To investigate the potential role of HPV in the burden of OPC/OCC, we classified anatomical codes as a proxy for HPV exposure based on robust scientific evidence [
The incidence rates were further age-standardized per 100,000 persons using the direct method, the World Health Organization (WHO) standard population [
For the incidence trend analysis, we adjusted Poisson regression models by sex. The dependent variable was the annual incidence, and the explanatory variables were the HPV group (HPV-related or HPV-unrelated subsites), age group (≤39, 40–59, and 60+ years), and time (years). We used as offset the population size according to the age group and year. With the models, we estimated the incidence rates and the annual percent change (APC; 95% confidence interval [CI]) for each segment (i.e., tumor classification and age group). To assess whether the trends were similar in different segments, we assessed contrasts under the general linear hypotheses [
Additionally, we investigated the effects of age, calendar year (period), and birth year (cohort) on incidence rates with Poisson regression to measure the relative risk [
Over the entire study period, there were 15,391 new cases of OPC/OCC in São Paulo city. Of these cases, 5,898 occurred in HPV-related subsites (38.3%), and 9,493 occurred in HPV-unrelated subsites (61.7%). There was a marked male dominance in both HPV-related and HPV-unrelated subsites and all anatomical site groups (
TOTAL | SEX | |||
---|---|---|---|---|
Male | Female | Ratio (M:F) | ||
15,391 (100.0%) | 11,538 | 3,853 | 3.0 | |
5,898 (38.3%) | 4,801 | 1,097 | 4.4 | |
Base of the tongue | 1,287 (8.4%) | 1,062 | 225 | 4.7 |
Oropharynx | 2,275 (14.8%) | 1,915 | 360 | 5.3 |
Tonsils | 1,545 (10.0%) | 1,189 | 356 | 3.3 |
Waldeyer’s ring, soft palate and uvula | 791 (5.1%) | 635 | 156 | 4.1 |
9,493 (61.7%) | 6,737 | 2,756 | 2.4 | |
Gums | 625 (4.1%) | 372 | 253 | 1.5 |
Hard palate | 836 (5.4%) | 486 | 350 | 1.4 |
Mouth | 3,847 (25.0%) | 2,810 | 1,037 | 2.7 |
Other parts of the tongue | 4,185 (27.2%) | 3,069 | 1,116 | 2.8 |
M:F: male:female ratio.
In the temporal analysis, incidence trends decreased or remained stable for all anatomical sites, sexes, and age groups (
When we compared the inclines (β1) with the contrasts under the general linear hypotheses, we found statistically significant differences among HPV-related and HPV-unrelated subsites for young females (p≈ 0.010) and males (p = 0.033). Additionally, the incidence trends were different between individuals aged ≤39 years and older age groups in both sexes for HPV-related subsites, which indicates a distinct pattern for HPV-related and HPV-unrelated subsites in the young population.
In the age-period-cohort analysis, the incidence rates of HPV-related OPC/OCC increased in the youngest birth cohorts in both sexes, while it decreased in HPV-unrelated subsites (
Model | Female | Male | ||||||
---|---|---|---|---|---|---|---|---|
Resid. Df | Resid. Dev | Deviance | p-value | Resid. Df | Resid. Dev | Deviance | p-value | |
Age | 26 | 43.477 | 26 | 88.034 | ||||
Age-Drift | 25 | 42.438 | 1.038 | 0.308 | 25 | 52.329 | 35.705 | <0.001 |
Age-Cohort | 12 | 10.832 | 31.607 | 12 | 15.429 | 36.900 | ||
Age-Period-Cohort | 11 | 10.776 | 0.055 | 0.814 | 11 | 12.023 | 3.406 | 0.065 |
Age-Period | 24 | 42.374 | -31.598 | 0.003 |
24 | 48.994 | -36.971 | <0.001 |
Age-Drift | 25 | 42.438 | -0.064 | 0.800 | 25 | 52.329 | -3.335 | 0.068 |
Age | 26 | 60.580 | 26 | 346.440 | ||||
Age-Drift | 25 | 30.405 | 30.175 | <0.001 |
25 | 61.740 | 284.696 | <0.001 |
Age-Cohort | 12 | 22.483 | 7.923 | 0.849 | 12 | 18.470 | 43.270 | <0.001 |
Age-Period-Cohort | 11 | 13.165 | 9.318 | 11 | 8.800 | 9.676 | ||
Age-Period | 24 | 21.008 | -7.844 | 0.854 | 24 | 52.130 | -43.329 | <0.001 |
Age-Drift | 25 | 30.405 | -9.397 | 0.002 |
25 | 61.740 | -9.618 | 0.002 |
Df: degree of freedom; Dev: deviance
*: statistically significant.
Understanding the HPV pattern over time is critical to support prophylactic interventions focused on diminishing its harmful impacts on populations. For this reason, we performed this study to determine the burden of HPV in OPC/OCC in São Paulo city, which is the largest city in Latin America. Overall, the incidence rates decreased for most OPCs/OCCs in both sexes in all age groups, and particularly in those with HPV-unrelated subsites. However, we identified a dramatically increasing incidence trend for HPV-related OPC/OCC in young females, emphasizing the highest risk in recent birth cohorts.
Our study revealed a striking reduction in the overall incidence trends of OPC/OCC and an age-period-cohort effect in HPV-unrelated subsites. Since the 1990s, the Brazilian tobacco control policy has led to a decrease in smoking [
The lung cancer incidence is decreasing worldwide, while the OPC incidence is increasing, which suggests that HPV infection is a reason for this growth [
A possible cause for the increased incidence of cancer at HPV-related subsites is the changing sexual behaviors [
Another potential reason for our findings is the prevalence of alcohol and tobacco use by the individuals with cancer in HPV-related subsites. Previous studies have reported an HPV infection rate of 6.2% in a healthy Brazilian population [
Although our results are not sufficiently robust to indicate a viral epidemic in São Paulo city, they highlight the need to monitor the burden of these cancers in young populations. Considering that cancer is a time-dependent disease, we expected stable trends for individuals aged ≤ 39 years, but we found an upward trend in HPV-related OPC/OCC in young individuals. In addition, the remarkable decrease in smoking may be masking the increasing incidence in the HPV-related subsites, as this tumor classification occasionally includes tobacco-related cancers. For this reason, our data suggest that a dramatic increase in HPV-related OPC/OCC in its initial phase is occurring in the city of São Paulo. Indeed, this is a public health concern due to high survival rates and its potential magnitude, which impairs patients’ quality of life and the healthcare and social security systems, particularly in developing countries, such as Brazil.
Our study had limitations that are important to its interpretation. As in other countries [
This population-based study had strengths including its generalizability and limited selection bias, which allowed us to demonstrate the emerging risk of HPV-related OPC/OCC in young males and females, whereas the risk in HPV-unrelated subsites sharply decreased. For these reasons, our data support the HPV vaccination program and the continuity of the Brazilian tobacco control policy, which has had positive effects on OPC/OCC incidence in São Paulo city. Furthermore, it is crucial to continue to specifically monitor the incidence of HPV-related OPC/OCC and to investigate vaccine effectiveness in the oropharyngeal and oral cavity subsites in the long term. Moreover, cancer control programs should broaden screening coverage in young subjects to prevent sexually transmitted infections and HPV-related cancers to minimize the increasing risk.
[A]: Base of the tongue; [B]: oropharynx; [C]: tonsils; [D]: Waldeyer’s ring, soft palate and uvula1; [E]: gums; [F]: hard palate; [G]: mouth; and [H]: other parts of the tongue.a,b APC: annual percent change; *: statistically significant APC (95% CI). a For better graph visualization, we applied the simple moving average of 5 years. b We analyzed these data with joinpoint regression models. 1 As there were 11 cases of cancer in Waldeyer's ring, they were combined with the cases of cancer in the soft palate and uvula.
(TIF)
[A]: Base of the tongue by age groups; [B]: oropharynx by age groups; [C]: tonsils by age groups; [D]: Waldeyer’s ring, soft palate and uvula1 by age groups; [E]: gums by age groups; [F]: hard palate by age groups; [G]: mouth by age groups; and [H]: other parts of the tongue by age groups.a,b APC: annual percent change; *: statistically significant APC (95% CI). a For better graph visualization, we applied the simple moving average of 5 years. b We analyzed these data with joinpoint regression models. 1 As there were 11 cases of cancer in Waldeyer's ring, these cases were combined with the cases of cancer in the soft palate and uvula.
(TIF)
[A]: Incidence trends for OCC/OPC according to HPV groups; [B]: incidence trends for OCC/OPC by sex and HPV groups; [C]: incidence trends for HPV-related OCC/OPC by age groups; and [D]: incidence trends for HPV-unrelated OCC/OPC by age groups.a,b APC: annual percent change; *: statistically significant APC (95% CI). a For better graph visualization, we applied a simple moving average of 5 years. b We analyzed these data with joinpoint regression models.
(TIF)
The authors are grateful to the Federal University of Sergipe for supporting FSM in his doctoral degree. Also, we thank the Population-based Cancer Registry of São Paulo for providing the cancer database, and the SEADE Foundation for demographic information on the city of São Paulo.