Trends in head and neck cancers in Peru between 1987 and 2008: Experience from a large public cancer hospital in Lima

Abstract

Background

Few studies have evaluated the trends in head and neck cancer in developing countries. The purpose of this study was to estimate trends in incidence of human papillomavirus–related (HPV-R) and HPV-unrelated (HPV-U) head and neck cancer in Lima, Peru, from 1987 to 2008.

Methods

Registry data from a single public cancer hospital were used to estimate age and sex-specific incidence rates. Annualized percent change was estimated using Poisson regression.

Results

The rate of total head and neck cancers, HPV-U, and HPV-R was 11.9, 10.9, and 0.8, respectively, per 100,000 person-years. Significant increases in HPV-U head and neck cancer were observed in men aged 30 to 44 (2.5%/year) and women 15 to 29 (4.2%/year), 30 to 44 (3.4%/ year), and 60 to 74 (2.0%/year). Significant increases in HPV-R head and neck cancer were observed only among men aged 45 to 59 (9.6%/year).

Conclusion

Although increased exposure to tobacco, occupational carcinogens, and changing sexual behaviors could be influencing these trends, additional analyses to assess generalizability of these findings to other regions of Peru are needed.

INTRODUCTION

Head and neck cancers are collectively the sixth most common cancer worldwide causing more than 500,000 incident cases each year.¹ Countries of the developing world bear a higher burden of incidence and mortality as compared to those of the developed world. Subsets of these cancers, namely the oropharynx and tonsils, are associated with infection of certain types of anogenital human papillomavirus (HPV), the causal agent of cervical carcinogenesis.^2,3

Over the last 20 years, the United States and Europe have observed a reduction in the overall incidence of head and neck cancer, specifically of HPV-unrelated (HPV-U) head and neck cancer. This can be attributed to a reduced consumption of tobacco and alcohol. However, there has been a 2% to 3% annual increase in the incidence of HPV-related (HPV-R) head and neck cancer subsets, particularly in men under the age of 60.^4–7 Hypotheses for this increase include increased risky sexual behavior leading to enhanced exposure to HPV.

Peru is a middle-income country that has reported low rates of oral cavity and pharyngeal cancers relative to the South American region and the world.^1,8 However, these cancers rates are derived using aggregated data from neighboring countries such as Colombia, Ecuador, and Brazil and are therefore not directly calculated using registry information from Peru itself.⁸ Furthermore, there are no studies that have estimated the changing burden of other head and neck cancer subsets, particularly those associated with HPV infection in Peru.

The purpose of this study is to estimate the change in HPV-U and HPV-R head and neck cancer incidence from 1987 to 2008 using registry data from a large public cancer hospital in Lima, Peru.

MATERIALS AND METHODS

Cancer case ascertainment at the Instituto Nacional de Enfermedades Neoplasticas

A cancer registry located at the Instituto Nacional de Enfermedades Neoplasticas (INEN) in Lima, Peru, were utilized to collect information on head and neck cancer during the years 1987 to 2008. INEN-Lima receives approximately 75% of the total cancer cases in Lima (a city with a population of 9 million, one-third the total population of Peru). All cases that are referred to INEN undergo histologic review to confirm pathologic diagnosis. International Classification of Disease-O-3 code information related to cancer topology and morphology as well as date of diagnosis, date of birth, age at diagnosis, and sex was abstracted from INEN database. The use of this registry data was approved by INEN and the Institutional Review Board of the Johns Hopkins Bloomberg School of Public Health.

Classification of disease

Head and neck cancer subsets were defined based on the American Joint Committee on Cancer classification. These subsets include cancers of the lip (C00.0–C00.9), oral cavity (C02.0–C02.3, C02.8, C0.30–C03.9, C04.0–C04.9, C05.0, C05.8, and C06.0–C06.9), salivary glands (C07.9–C08.9), paranasal sinuses and the nasal cavity (C30.0–C31.9), nasopharynx (C11.0–C11.9), hypopharynx (C13.0–C13.9), oropharynx including base of tongue, soft palate, and uvula (C01.9, C05.1–C05.2, and C10.0–C10.9), tonsils including the palatine tonsil, lingual tonsils, and Waldeyer Ring (C09.0–C09.9, C02.4, and C14.2), and larynx (C32). Cancers of the tongue-not other specified (NOS; C02.9), palate-NOS (C05.9), and pharynx-NOS (C14.0) were excluded from the analysis. Lymphomas and other hematologic cancers (9590–9999) were also excluded from the analysis.

Individual cancers were further categorized based on their association with HPV as defined in the literature and by the American Joint Committee on Cancer^9–11: (1) HPV-U head and neck cancers: oral cavity (excluding cancers of the base of the tongue, uvula, and soft palate), salivary glands, paranasal sinuses, nasal cavity, nasopharynx, hypopharynx, and larynx; (2) HPV-R head and neck cancers: oropharynx (including base of tongue, uvula, and soft palate cancers), and tonsil (including cancers of the Waldeyer Ring, palatine, and lingual tonsils).

Statistical analysis

Incidence rates were adjusted to estimate the burden of disease in Lima. The total number of cases reported at INEN-Lima was up-weighted to reflect the total number of cases occurring in Lima (case #/0.75). Population estimates were obtained from the Instituto Nacional de Estadistica e Informatico, the Peruvian National Census Institute, and down-weighted to estimate the number of individuals at risk in Lima (population at risk × 0.3). Incidence rates were further age-standardized to the 2000 global population provided by the World Health Organization. Age was categorized (15–29, 30–44, 45–59, 60–74, and 75+ years). Differences in the incidence of each cancer were compared across sexes using incidence rate ratios estimated using Poisson regression. Annualized percent change (APC; 95% confidence interval [CI]) in cancer incidence for each sex and age category was estimated using log-linear Poisson regression models. Differences in sex-specific APC values across age groups were assessed by inclusion of a product term in the Poisson model. Individuals <15 years of age were excluded from the analysis because of the low number of cancers reported. A p value of < .05 was considered statistically significant. Analyses were performed using STATA 11.0 (STATACORP, College Station, TX).

RESULTS

Total head and neck cancer

There were a total of 5745 documented incident cases of head and neck carcinomas in the INEN-Lima database between 1987 and 2008 (Table 1). The incidence rate of total head and neck cancers, HPV-U head and neck cancers, and HPV-R head and neck cancer cancers was, 11.9, 10.9, and 0.8 cases per 100,000 person-years, respectively. Men had a higher rate of total head and neck cancer (p < .01), HPV-U head and neck cancer (laryngeal, p < .01; nasopharynx, p < .01; and lip, p < .01); and HPV-R head and neck cancer (oropharynx, p < .01; and tonsils, p < .01). Conversely, men had a significantly lower incidence of oral cavity cancers as compared to women (p < .01). Higher rates of total head and neck cancer, HPV-U head and neck cancer, and HPV-R head and neck cancer were most strongly and consistently observed among men and women >45 years of age (Table 2). However, the magnitude of the male:female rate ratios differed significantly across age categories for total, HPV-U, and HPV-R head and neck cancer (p_interaction < .01).

TABLE 1.

Number of incident cases and cumulative incidence of head and neck squamous cell carcinoma by sex and site per 100,000 from 1987 to 2008 in Peru.

Cancer site	Total			Men			Women			Rate ratio (95% CI) Men: women†
	No. of cases	%	Rate*	No. of cases	%	Rate*	No. of cases	%	Rate*
Total	5745	100.0	11.9	3288	100.0	14.8	2457	100.0	9.5	1.42 (1.29–1.57)
HPV-R	5279	92.1	10.9	2982	90.9	13.5	2297	93.6	8.8	1.38 (1.25–1.53)
Oral cavity	1942	33.8	4.0	850	25.9	3.6	1092	44.4	4.4	0.80 (0.70–0.92)
Larynx	1001	17.4	2.0	871	26.5	4.0	130	5.3	0.4	8.26 (6.58–10.4)
Nasal cavity and sinuses	1057	18.4	1.8	547	16.6	2.0	510	20.8	1.6	1.15 (0.98–1.31)
Nasopharynx	269	4.7	0.3	187	5.7	0.5	82	3.3	0.2	2.54 (1.95–3.31)
Hypopharynx	289	5.0	0.5	147	4.5	0.7	142	5.8	0.4	1.08 (0.82–1.41)
Salivary gland	506	8.8	0.8	246	7.5	0.9	260	10.6	0.7	1.04 (0.85–1.28)
Lip	215	3.7	0.5	134	4.1	0.6	81	3.3	0.4	1.94 (1.40–2.67)
HPV-U	466	8.1	0.8	306	9.3	1.2	160	6.5	0.5	2.21 (1.75–2.78)
Oropharynx‡	308	5.4	0.5	193	5.9	0.8	115	4.7	0.3	1.88 (1.44–2.45)
Tonsils§	158	2.7	0.2	113	3.4	0.4	45	1.8	0.1	2.83 (1.87–3.84)

Abbreviations: CI, confidence interval; HPV-R, human papillomavirus related; HPV-U, human papillomavirus unrelated.

^*Per 100,000 persons; rates age-standardized to 2000 global population.

^†Age-adjusted.

^‡Including cancers of the base of tongue, soft palate, and uvula.

^§Including cancers of the palatine, lingual tonsil, and Waldeyer’s ring.

TABLE 2.

Number of incident cases and incidence (per 100,000) of head and neck cancers by age groups for men and women from 1987 to 2008 in Peru.

Age group	Total cancers					HPV-U cancers					HPV-R cancers
	Men		Women			Men		Women			Men		Women
	No. of cases	Rate*	No. of cases	Rate*	Rate ratio (95% CI)	No. of cases	Rate*	No. of cases	Rate*	Rate ratio (95% CI)	No. of cases	Rate*	No. of cases	Rate*	Rate ratio (95% CI)
15–29	117	0.15	114	0.15	1.01 (0.74–1.38)	113	0.14	108	0.14	1.04 (0.78–1.38)	4	0.01	6	0.01	0.66 (0.19–2.33)
30–44	347	0.69	376	0.71	0.94 (0.79–1.12)	315	0.63	349	0.66	0.95 (0.80–1.12)	32	0.06	27	0.05	1.24 (0.75–2.06)
45–59	935	3.2	673	2.2	1.47 (1.32–1.65)	844	2.9	633	2.1	1.40 (1.25–1.57)	91	0.32	40	0.13	2.39 (1.61–3.56)
60–74	1292	9.1	807	5.1	1.78 (1.62–1.95)	1166	8.2	751	4.8	1.72 (1.56–1.90)	126	0.89	56	0.36	2.49 (1.81–3.43)
75+	597	16.5	487	10.2	1.62 (1.42–1.85)	544	15.0	456	9.6	1.58 (1.38–1.81)	53	1.5	31	0.65	2.25 (1.43–3.54)

Abbreviations: CI, confidence interval; HPV-R, human papillomavirus related; HPV-U, human papillomavirus unrelated.

^*Per 100,000 person-years; reference group for rate ratio 5 female.

In 1987, the rate of total head and neck cancer was 9.0 cases per 100,000 person-years, whereas in 2008, the rate was 11.0 cases per 100,000 person-years. The overall APC for all head and neck cancers was 1.8% per year. Significant annual increases were found in cancers of the oral cavity (2.6% per year; p < .01), salivary gland (1.9% per year; p < .01), tonsils (5.4% per year; p < .01), and oropharynx (4.2% per year; p < .01; Table 3). Similar APC estimates by cancer subsite were observed between men and women.

TABLE 3.

Annualized percent change (95% CI) of individual cancers by site (age-adjusted).

Cancer type	Total	Men	Women
Total head and neck cancer	1.8 (0.9–2.6)	1.3 (0.1–2.4)	2.5 (1.4–3.6)
HPV-U	1.6 (0.8–2.4)	0.9 (−0.2 to 2.2)	2.4 (1.3–3.6)
Oral cavity	2.6 (1.5–3.7)	2.3 (0.6–3.9)	2.9 (1.4–4.4)
Salivary glands	1.9 (0.2–3.7)	1.8 (−0.7 to 4.4)	2.1 (−0.2 to 4.5)
Nasopharynx	−0.2 (−2.1 to 1.9)	0.6 (−1.8 to 3.1)	−2.0 (−5.5 to 1.6)
Hypopharynx	2.0 (−0.2 to 4.3)	−1.7 (−4.9 to 1.3)	6.1 (2.9–9.6)
Nasal cavity/sinus	0.9 (−0.2 to 2.2)	1.6 (−0.1 to 3.3)	0.3 (−1.5 to 2.0)
Lip	−0.4 (−2.9 to 2.1)	−0.8 (−3.9 to 2.5)	0.2 (−3.9 to 4.3)
Larynx	−0.9 (−2.3 to 0.6)	−0.5 (−2.2 to 1.1)	−3.0 (−5.9 to −0.1)
HPV-R	4.9 (3.0–6.7)	5.4 (3.0–7.8)	3.8 (0.9–6.7)
Tonsils*	5.4 (2.2–8.7)	6.3 (2.4–10.3)	3.3 (−2.1 to 8.6)
Oropharynx†	4.2 (2.2–6.1)	4.4 (1.9–6.9)	3.7 (0.6–6.8)

Abbreviations: CI, confidence interval; HPV-U, human papillomavirus unrelated; HPV-R, human papillomavirus related.

^*Includes palatine, lingual tonsils, and Waldeyer’s ring.

^†Includes base of tongue, soft palate, and uvula.

Human papillomavirus–unrelated head and neck cancer

There were a total of 5279 incident cases of HPV-U head and neck cancer in Peru between the years 1987 and 2008. The rate for HPV-U head and neck cancer in 1987 was 8.5 cases per 100,000 person-years, whereas in 2008, the number increased to 9.9 cases per 100,000 person-years (Figure 1). The overall APC was 1.6% per year (p < .01; Table 3). There was a significant increase in HPV-U head and neck cancer in men aged 30 to 44 years (2.1% per year; p = .03) and in women aged 15 to 29 years (4.4% per year; p = .01); 30 to 44 years (3.1% per year; p < .01), and 60 to 74 years (1.9% per year; p < .01; Figure 2).

FIGURE 1.

Age-standardized incidence rates for human papillomavirus (HPV)-unrelated (HPV-U) and HPV-related (HPV-R) head and neck cancers per 100,000 people from 1987 to 2008 in Lima, Peru (reference population = 2000 global population). APC, annualized percent change.

FIGURE 2.

Annual percentage change (APC) of total, human papillomavirus (HPV)-related (HPV-R), and HPV-unrelated (HPV-U) head and neck cancer cases by sex. APC values and 95% confidence intervals (CIs) for HPV-R head and neck cancer (■), HPV-U head and neck cancer (•) are presented as coefficients generated using a Poisson regression with robust variance. Individuals <14 years of age were dropped because of the limited number of cases.

Human papillomavirus–related head and neck cancers

There were a total of 466 incident cases of HPV-R head and neck cancers in Peru between the years 1987 and 2008. In 1987, the rate for HPV-R head and neck cancer was 0.4 cases per 100,000 person-years, whereas in 2008, the rate was 1.0 case per 100,000 person-years (Figure 1). The overall APC was 4.9% per year (p < .01; Table 3). There was a significant increase in HPV-R head and neck cancer only among men aged 45 to 59 years (8.7% per year; p < .01; Figure 2).

DISCUSSION

Using registry data from a large public cancer hospital in Lima, Peru, we observed significant increases in the incidence of both HPV-R and HPV-U head and neck cancer subsets from 1987 to 2008. The magnitude of these increases differed by subset with larger per year increases observed among HPV-R as compared to HPV-U cancers. Last, these increases, particularly among HPV-U cancers, were observed among men and women of younger age groups.

The combined rate of oral cavity, lip, laryngeal, and pharyngeal cancers in our study (7.5 cases per 100,000) is higher than what has been estimated for Peru by the World Health Organization/Institut Catal a d’Oncologia⁸ (4.3 cases per 100,000) but lower than rates reported in the entire South American continent (11.9 cases per 100,000).⁶ The burden of head and neck cancer in South America among countries with definitive cancer registries is heterogeneous with rates as high as 16.3 cases per 100,000 in Brazil and as low as 4.4 and 4.9 cases per 100,000 in Chile and Ecuador. Head and neck cancer rates in Peru are currently estimated using incidence-mortality ratios generated from cancer registry data aggregated across neighboring Colombia, Brazil, and Ecuador. These ratios are then applied to mortality data from Peru. Differences in the ascertainment of cause-specific mortality between Peru and other countries may therefore influence the application of this methodology. Therefore, empirical estimates such as ours may be helpful to better inform these modeled estimates.

The incidence rates of HPV-U head and neck cancer observed in our study was significantly higher as compared to rates reported in North America across similar time periods.¹² Furthermore, unlike in the United States, which had declining rates of HPV-U head and neck cancer, our study demonstrated significant yearly increases in incidence, particularly among men and women in younger age groups. Since the late 1970s, Peru has been experiencing rapid industrialization and modernization potentially including uptake of alcohol and tobacco use. However, rates of smoking and alcohol consumption in Peru measured from 1955 to 1989 have not significantly changed and the current rate of tobacco use is considered low relative to other developing countries.¹³ Alternatively, the change from an agrarian to an industry-based economy causing large urban migrations, particularly in Lima, may have resulted in changes in lifestyle factors such as diet, physical activity, and occupation, thereby enhancing exposure to airborne particulates and other carcinogens.

The large yearly increases in the incidence of tonsil and oropharyngeal cancers observed in our study are similar in magnitude to the increases observed in North America and Europe. Infection with HPV is attributed to approximately 30% to 80% of these cancers and the prevalence of HPV detected in oropharyngeal cancers has increased more than 200% from 1984 to 2004.¹⁴ Anogenital types of HPV that are detected in oropharyngeal tumors are primarily transmitted through sexual activity such as oral sex.¹⁵ Therefore, the increases in the incidence of oropharyngeal cancers in North America and Europe has been associated with increased sexual activity among younger birth cohorts resulting in increased exposure to oral HPV.^4,12 In Peru, similar changes in sexual risk behavior have been observed among adolescents, including younger age of sexual debut and increased number of sexual partners.¹⁶ However, despite these large annual increases in the incidence, the underlying rates in our study are significantly lower than in North America and Europe. This suggests that the influence of oral HPV infection of these cancers is a relatively recent phenomenon. Currently, we are measuring the prevalence of HPV DNA in oropharyngeal and tonsillar tumor tissue taken from the same cancer cases utilized in this study in order to estimate the changing fraction of oropharyngeal cancers that can be attributed to HPV.

The male predominance in the incidence of most head and neck cancer subsets in this study has been observed previously.¹⁷ This disparity is hypothesized to be driven by behavioral differences between men and women that increase the likelihood and frequency of carcinogenic exposures, such as alcohol use, smoking, and HPV infection. Additionally, studies looking at the effects of smoking and alcohol use on head and neck cancer risk have demonstrated differential risk across sexes at similar levels of carcinogenic exposure.^18,19 Last, a pooled analysis assessing sex differences in lung cancer risk among never smokers showed a higher risk among women, suggesting an independent biological effect of sex on cancer risk, irrespective of exposure to environmental carcinogens.²⁰ However, given the lack of individual-level information on such risk factors as HPV exposure, smoking, and alcohol use, it is difficult to determine which hypothesis is most relevant to our population-level findings. Most likely, these 3 potential mechanisms are operating in a nonmutually exclusive fashion to induce differences in head and neck cancer risk by sex.

Contrary to the male predominance of most of the head and neck cancer subsets in this study, oral cavity cancer incidence was higher among women as compared to men in this study. In addition, the increase in oral cavity cancer incidence in our study was similar in magnitude to the HPV-R head and neck cancer subsets. Cancers of the oral cavity have shown similar significant increases in incidence over a 20-year period in the United States Surveillance, Epidemiology, and End Results study, particularly among women <40 years of age.²¹ In this study, we did observe a strong and significant increase in the incidence of total oral cavity cancers among women 15 to 29 years of age (APC, 6.9; p = .03). The reasons behind these increases are not fully understood as only a very small fraction of oral cavity cancers can be attributed to HPV infections and are more closely linked to smoking and alcohol use.^22,23 One possibility is the potential misclassification of base of tongue cancers, which are strongly associated with HPV. However, this is unlikely given that these patterns have been observed across multiple cancer registries in the United States Surveillance, Epidemiology, and End Results. More recently, marijuana use has been implicated as a potential risk factor in oral cavity cancer as increases in drug use mirror increases in cancer incidence.²⁴ However, case-control studies that have set out to explore this relationship have been inconsistent with some studies showing a protective effect of marijuana use on oral cavity cancer.^25,26 Additional work is therefore needed to better understand the factors driving the increases in oral cavity cancers, particularly among women.

This study has some limitations. First, the use of cancer incidence data from a single public hospital in Lima makes generalizations regarding the changing epidemiology of head and neck cancer across Peru difficult. A recent study looking at cervical cancer rates across various regions of Peru demonstrated significantly lower rates in Lima as compared to rural, highland regions.²⁷ Therefore, the rates reported in this study may not be wholly reflective of the true head and neck cancer burden throughout Peru. Furthermore, the distribution of head and neck cancer subsets, particularly HPV-R oropharyngeal and tonsillar cancer is slightly different than in other studies from the United States and Europe, which may be a result of using cancer registry information from a single cancer hospital as opposed to a nationally representative cancer registry.^4,12,28 As a result, follow-up studies using registry data from other INEN hospitals in other regions of Peru should be used to assess the generalizability of these findings. Third, we were unable to obtain Lima city-specific census figures for a more direct estimate of our population at risk and therefore applied a weighting scheme to the countrywide census estimates under the assumption that the age and sex-specific population distribution as reported in the countrywide census is similar to that of Lima. This weighting scheme may be biased if certain age and sex groups are differentially represented in Lima as compared to the entirety of Peru.

In one of the first studies to assess the changing burden of head and neck cancer in Peru, increases in incidence over the 22-year observation period may reflect the negative consequences of modernization and the subsequent shift in cultural and social values. Although the overall rate of HPV-R head and neck cancer subset is generally low in this population, the relatively large year-over-year increases creates a cause for concern. These estimates could be utilized by local public health officials to help guide research efforts as well as develop and apply effective prevention programs.