Burden of potentially human papillomavirus-associated cancers of the oropharynx and oral cavity in the US, 1998–2003

Abstract

BACKGROUND.

As human papillomavirus (HPV) vaccination becomes widely available in the US for cervical cancer prevention, it may also affect the rates of other cancers potentially associated with HPV. The objective of the current study was to describe the incidence rates of oropharyngeal and oral cavity cancers in the US with a focus on anatomic sites potentially associated with HPV infection.

METHODS.

Incident cases diagnosed between 1998 and 2003 identified through 39 population-based registries that participate in the National Program of Cancer Registries and/or the Surveillance, Epidemiology, and End Results Program were examined. The incidence rates of potentially HPV-associated oropharyngeal and oral cavity cancers by various characteristics were estimated. The 1998 through 2003 trends in these rates were also compared with rates for sites not previously shown to be associated with HPV (comparison sites).

RESULTS.

In all, 44,160 cases of potentially HPV-associated cancers of the oropharynx and oral cavity were identified, including 19,239 (43.6%) tonsillar, 16,964 (38.4%) base of tongue, and 7957 (18.0%) other oropharyngeal cancers. The incidence rates for these sites were highest among blacks, and higher among non-Hispanics and men than among Hispanics and women. The annual incidence rates of potentially HPV-associated cancers of the tonsil and base of tongue both increased significantly from 1998 through 2003 (annual percentage change [APC], 3.0; P < .05 for both sites), whereas the incidence rates of cancer at the comparison sites generally decreased.

CONCLUSIONS.

The results of the current study provide baseline incidence rates of potentially HPV-associated cancers of the oropharynx and oral cavity that can be compared with rates after the widespread implementation of the HPV vaccination. Cancer 2008;113(10 suppl):2901–9. Published 2008 by the American Cancer Society.

The anatomy of the pharynx (oropharynx, hypopharynx, and nasopharynx) and oral cavity (lip, tongue, gum, floor of mouth, and palate) is a complex region. In a recent study, researchers estimated that 34,360 new cases of pharyngeal and oral cavity cancer would be diagnosed in 2007 in the US and that 7550 people will die from these cancers.1 The incidence rates for these cancers are higher among men and blacks than among women and whites.2 The mortality rate for these cancers is also significantly higher among black men (6.9 per 100,000) than among white men (3.7 per 100,000).2 Greater than 90% of the cancers in this anatomic region are squamous cell carcinomas.3, 4

The most frequently cited risk factors for cancers of the pharynx and oral cavity are tobacco and alcohol use, and the combined effect on cancer risk is believed to be multiplicative rather than additive.5–8 Although most of these cancers in developed countries have been attributed to these risk factors,3 some patients report having never smoked or drank heavily.9, 10 Recent epidemiologic and experimental data provide evidence for an association between human papillomavirus (HPV) infection and a subset of oropharyngeal and oral cavity cancers.7, 11–23 This evidence includes a positive association between risk for oropharyngeal cancer and lifetime number of vaginal and/or oral sex partners.23

There is considerable variation in the proportion of cancers that are HPV-positive among the different sites of the oropharynx and oral cavity.7, 11, 12, 14–19, 21, 22, 24, 25 In a review of 60 studies, Kreimer et al24 calculated the average HPV-DNA positivity rate to be 35.6% for oropharyngeal carcinomas and 23.5% for oral cavity carcinomas. Overall, approximately 25% of these cancers have been associated with high-risk HPV types, with HPV-16 accounting for 87% of HPV-infected oropharyngeal cancers and 68% of HPV-infected oral cavity carcinomas.24, 26–29 Among the cancers of the oropharynx and oral cavity, tonsillar squamous cell carcinoma has been most strongly and consistently associated with HPV-16 infection.16, 19, 30, 31

The recognition that HPV plays any role in oropharyngeal and oral cavity cancer has important implications for cancer prevention. As the vaccination for HPV becomes widely available in the US for cervical cancer prevention, the incidence rates of other cancers potentially associated with HPV, such as those of the oropharynx and oral cavity, may also decline. To provide baseline data against which to measure the effect of the HPV vaccine on rates of these cancers, we estimated the incidence rates of squamous cell carcinoma of the oropharynx and oral cavity in the US from 1998 through 2003, with a focus on anatomic sites in these areas previously found to be associated with HPV infection.

MATERIALS AND METHODS

Cases included in this analysis were microscopically confirmed, newly diagnosed in situ or invasive squamous cell carcinomas (International Classification of Disease for Oncology, 3rd edition [ICD-O-3] morphology codes 8050-8084, 8120-8131) of the oropharynx and oral cavity identified from population-based cancer registries that participated in either the National Program of Cancer Registries (NPCR); the Surveillance, Epidemiology, and End Results (SEER) program; or both from 1998 through 2003.2, 32 NPCR and SEER data from 38 states and the District of Columbia met the data quality criteria (US Cancer Statistics Publication Standard) for all cancer sites during this period and were included in this analysis.

We focused our analysis on oropharynx and oral cavity carcinomas most strongly associated with HPV infection. Because the NPCR and the SEER Program do not collect information regarding the presence of HPV within tumors, we relied exclusively on anatomic sites' proclivity for HPV reported in previous studies.7, 11, 12, 14–19, 21, 22, 24, 25, 33–35 We divided potentially HPV-associated sites into 3 major anatomic areas: the tonsil, including the Waldeyer ring; the base of the tongue and the lingual tonsil; and other potentially HPV-associated sites within the oropharynx. Detailed descriptions and specific ICD-O-3 topography codes used to define these unique anatomic groupings are shown in Table 1.

Table 1. International Classification of Disease for Oncology 3^rd Edition (ICD-O-3) Topography Codes Used to Define Sites of Potentially HPV-associated Oropharyngeal and Oral Cavity Cancers and Comparison Sites

Sites of Potentially HPV-Associated Cancers
HPV indicates human papillomavirus; NOS, not otherwise specified.
Tonsil (including Waldeyer ring)
C09.0  Tonsillar fossa
C09.1  Tonsillar pillar
C09.8  Overlapping lesion of tonsil
C09.9  Tonsil, NOS
C14.2  Waldeyer ring
Base of tongue and lingual tonsil
C01.9  Base of tongue, NOS
C02.4  Lingual tonsil
Other oropharynx sites
C02.8  Overlapping lesion of tongue
C10.2  Lateral wall of oropharynx
C10.8  Overlapping lesion of oropharynx
C10.9  Oropharynx, NOS
C14.0  Pharynx, NOS
C14.8  Overlapping lesion of lip, oral cavity and pharynx
Comparison Sites
Oral tongue
C02.0  Dorsal surface of tongue, NOS
C02.1  Border of tongue
C02.2  Ventral surface of tongue, NOS
C02.3  Anterior 2/3 of tongue, NOS
C02.9  Tongue, NOS
Other oral cavity sites
C03.0  Upper gum
C03.1  Lower gum
C03.9  Gum, NOS
C04.0  Anterior floor of mouth
C04.1  Lateral floor of mouth
C04.8  Overlapping lesion of floor of mouth
C04.9  Floor of mouth, NOS
C05.0  Hard palate
C06.0  Cheek mucosa
C06.1  Vestibule of mouth
C06.2  Retromolar area
C06.8  Overlapping lesion of other and unspecified parts of mouth
C06.9  Mouth, NOS
Larynx
C32.0  Glottis
C32.1  Supraglottis
C32.2  Subglottis
C32.3  Laryngeal cartilage
C32.8  Overlapping lesion of larynx
C32.9  Larynx, NOS
Other oropharynx sites
C05.1  Soft palate, NOS
C05.2  Uvula
C05.8  Overlapping lesion of palate
C05.9  Palate, NOS
C10.0  Vallecula
C10.1  Anterior surface of epiglottis
C10.3  Posterior wall of oropharynx

We examined incidence rates for in situ and invasive potentially HPV-associated oropharyngeal and oral cavity cancers, and examined invasive cases independently by cancer site and stage at diagnosis and by the age, sex, race, ethnicity, and geographic region of the study subjects. Because of potential racial misclassification, information regarding American Indian and Alaska Natives was not available for this analysis.36 Rates were calculated per 100,000 people and adjusted by direct standardization to the 2000 US population. SEER-Stat software (version 6.1.4) was used to calculate counts, age-adjusted rates, and 95% confidence intervals (95% CIs).37 The stages of cancer cases diagnosed from 1998 through 2000 were classified according to the 1977 SEER summary staging system38 criteria, whereas the stages of cases diagnosed from 2001 through 2003 were classified according to the 2000 SEER summary staging system39 criteria. Although changes between the 2 systems did result in differences in how regional and distant stage at some sites were classified, we accounted for these differences by defining SEER localized cancers as early-stage disease and combining SEER regional and distant cancers into a single category of late-stage disease.40, 41

Lastly, we included a subanalysis of incidence rates of cancers that have not been previously associated with HPV infection but are anatomically similar to the potentially HPV-associated cancers (hereafter referred to as comparison sites). We divided the comparison sites into 4 major anatomic areas (Table 1): the tongue, other oral cavity sites, other sites within the oropharynx, and a subset of sites within the larynx that are within close proximity to several potentially HPV-associated sites (eg, glottis, supraglottis, and subglottis). We examined trends in incidence rates for the comparison sites by year of diagnosis to compare trends among the potentially HPV-associated cancers. We calculated the annual percent change (APC) in incidence rates from 1998 through 2003 by fitting a least-squares regression line to the natural logarithm of the rates, using the calendar year as a regression variable. We rejected the hypothesis that the APC equaled 0 if the P was ≤0.05.

RESULTS

Incidence Rates of Potentially HPV-associated Invasive Oropharyngeal and Oral Cavity Cancers

The study data file for 1998 through 2003 contained 44,160 cases of potentially HPV-associated invasive oropharyngeal and oral cavity cancers, of which 19,239 (43.6%) were tonsillar, 16,964 (38.4%) were base of tongue, and 7957 (18.0%) were other potentially HPV-associated oropharyngeal cancers (Table 2). The age-adjusted incidence rates were at least 3 times higher among men than women for all site categories. The incidence rates were also higher among blacks than other racial groups, higher among non-Hispanics than among Hispanics, and generally higher among those who lived in the South. In addition, most of these cases were diagnosed at a late stage (78.1% for tonsil, 77.4% for base of tongue, and 68.7% for other oropharynx cancers). Compared with cancers of the tonsil and base of tongue and lingual tonsil, other oropharyngeal cancers were more likely to occur in older people, women, and blacks, and to have been diagnosed at an earlier stage.

Table 2. Age-adjusted Incidence Rates* of Potentially HPV-associated Invasive Squamous Cell Carcinomas of the Oropharynx and Oral Cavity,† by Selected Characteristics, US, 1998–2003

Characteristic	Tonsil				Base of Tongue and Lingual Tonsil				Other Oropharynx
	No.	%	Rate	(95% CI)	No.	%	Rate	(95% CI)	No.	%	Rate	(95% CI)
HPV indicates human papillomavirus; 95% CI, 95% confidence interval. Data are from 39 population-based registries that participate in the National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and meet high-quality data criteria.36 These registries cover approximately 83% of the US population. * Rates are per 100,000 population and are age-adjusted to the 2000 US standard population (19 age groups: Census P25-1130). † Potentially HPV-associated cancers were defined on the basis of International Classification of Disease for Oncology 3^rd edition (ICD-O-3) topography codes (as shown in Table 1). ‡ Hispanic origin is not mutually exclusive from race categories (White, Black, Asian/Pacific Islander). § For 1998-2000, SEER Summary Stage 1997 was used; for 2001-2003, SEER Summary Stage 2000 was used. Data for the 2 staging systems are combined into “early” (includes localized) and “late” (includes regional and distant) to account for minor differences observed in comparative analyses.40, 41
Total	19,239	100	1.4	(1.3-1.4)	16,964	100	1.2	(1.2-1.2)	7957	100	0.6	(0.6-0.6)
Age group (median)	(57)				(60)				(63)
0-29 y	47	0.2	0.0	(0.0-0.0)	43	0.3	0.0	(0.0-0.0)	23	0.3	0.0	(0.0-0.0)
30-39 y	541	2.8	0.3	(0.2-0.3)	300	1.8	0.1	(0.1-0.2)	121	1.5	0.1	(0.0-0.1)
40-49 y	4159	21.6	1.9	(1.9-2.0)	2414	14.2	1.1	(1.1-1.2)	898	11.3	0.4	(0.4-0.4)
50-59 y	6433	33.4	4.1	(4.0-4.2)	5294	31.2	3.3	(3.3-3.4)	2046	25.7	1.3	(1.2-1.3)
60-69 y	4442	23.1	4.3	(4.2-4.4)	4584	27.0	4.4	(4.3-4.6)	2322	29.2	2.3	(2.2-2.3)
70-79 y	2778	14.4	3.4	(3.3-3.5)	3227	19.0	3.9	(3.8-4.1)	1853	23.3	2.3	(2.2-2.4)
80+ y	839	4.4	1.7	(1.6-1.9)	1102	6.5	2.3	(2.1-2.4)	694	8.7	1.4	(1.3-1.5)
Sex
Male	14,976	77.8	2.3	(2.2-2.3)	13,234	78.0	2.0	(2.0-2.1)	5736	72.1	0.9	(0.9-0.9)
Female	4263	22.2	0.6	(0.5-0.6)	3730	22.0	0.5	(0.5-0.5)	2221	27.9	0.3	(0.3-0.3)
Race
White	16,787	87.3	1.4	(1.4-1.4)	14,894	87.8	1.2	(1.2-1.2)	6578	82.7	0.5	(0.5-0.5)
Black	1936	10.1	1.5	(1.4-1.6)	1709	10.1	1.3	(1.3-1.4)	1207	15.2	1.0	(0.9-1.0)
Asian or Pacific Islander	259	1.3	0.5	(0.4-0.6)	186	1.1	0.4	(0.3-0.4)	77	1.0	0.2	(0.1-0.2)
Ethnicity‡
Non-Hispanic	18,229	94.8	1.4	(1.4-1.4)	16,217	95.6	1.2	(1.2-1.3)	7536	94.7	0.6	(0.6-0.6)
Hispanic	1010	5.2	0.9	(0.8-1.0)	747	4.4	0.7	(0.7-0.8)	421	5.3	0.4	(0.4-0.5)
Region
Northeast	3965	20.6	1.2	(1.2-1.2)	3961	23.3	1.2	(1.2-1.2)	1973	24.8	0.6	(0.6-0.6)
Midwest	5108	26.6	1.3	(1.3-1.4)	4317	25.4	1.1	(1.1-1.2)	2106	26.5	0.6	(0.5-0.6)
South	6052	31.5	1.6	(1.5-1.6)	5118	30.2	1.3	(1.3-1.4)	2529	31.8	0.6	(0.6-0.7)
West	4114	21.4	1.3	(1.3-1.4)	3568	21.0	1.2	(1.1-1.2)	1349	17.0	0.4	(0.4-0.5)
Cancer grade
I	1104	5.7	0.1	(0.1-0.1)	1171	6.9	0.1	(0.1-0.1)	766	9.6	0.1	(0.1-0.1)
II	7506	39.0	0.5	(0.5-0.5)	6490	38.3	0.5	(0.4-0.5)	3364	42.3	0.2	(0.2-0.2)
III	7278	37.8	0.5	(0.5-0.5)	5977	35.2	0.4	(0.4-0.4)	2073	26.1	0.1	(0.1-0.2)
IV	304	1.6	0.0	(0.0-0.0)	219	1.3	0.0	(0.0-0.0)	75	0.9	0.0	(0.0-0.0)
Unknown	3047	15.8	0.2	(0.2-0.2)	3107	18.3	0.2	(0.2-0.2)	1679	21.1	0.1	(0.1-0.1)
Cancer stage§
Early	3148	16.4	0.2	(0.2-0.2)	2830	16.7	0.2	(0.2-0.2)	1506	18.9	0.1	(0.1-0.1)
Late	15,035	78.1	1.1	(1.1-1.1)	13,122	77.4	0.9	(0.9-0.9)	5465	68.7	0.4	(0.4-0.4)
Unstaged	1056	5.5	0.1	(0.1-0.1)	1011	6.0	0.1	(0.1-0.1)	986	12.4	0.1	(0.1-0.1)

Incidence Rates of Potentially HPV-associated Invasive Tonsil Cancers by Sex and Race/Ethnicity

Incidence rates of tonsil cancer among men stratified by age and race/ethnicity were highest among blacks ages 60 to 69 years (9.0; 95% CI, 8.1-10.0), and lowest among Asian or Pacific Islander (API) men of all ages (Fig. 1). The rates of tonsillar cancer among women were substantially lower than those among men and were highest among women ages 60 to 69 years (2.0%; 95% CI, 1.9-2.1%) and 70 to 79 years (2.0%; 95% CI, 1.9-2.2).

Figure 1. Invasive tonsil* squamous cell carcinoma age-specific incidence rates† by sex and race/ethnicity, US, 1998 through 2003. Data are from 39 population-based registries that participate in the National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and meet high-quality data criteria.36 These registries cover approximately 83% of the US population. *Potentially human papillomavirus-associated tonsil cancers were defined on the basis of the International Classification of Diseases-Oncology (ICD-O)-3 topography codes (as shown in Table 1). †Rates are per 100,000 population and are age-adjusted to the 2000 US standard population (19 age groups: Census P25-1130). Rates based on <6 cases are not displayed. Rates marked with a solid square are based on at least 6 but <16 cases and should be interpreted with caution. ‡Hispanic origin is not mutually exclusive from race categories (White, Black, Asian/Pacific Islander [API]).

Incidence Rates of Potentially HPV-associated Invasive Base of Tongue Cancers by Sex and Race/Ethnicity

Incidence rates of base of tongue carcinomas among men stratified by age and race/ethnicity were also highest among blacks ages 60 to 69 years (9.5; 95% CI, 8.6-10.5), whereas API men had the lowest rate among men in this age group (2.1; 95% CI, 1.5-2.9) (Fig. 2). Overall, among men ages 40 to 79 years, the rates were highest among blacks, whereas among men aged ≥80 years, the rates were highest among whites and Hispanics. Women again had substantially lower rates than men, with the incidence rate highest among those ages 70 to 79 years. Although rates among women differed less by race/ethnicity than rates among men, black women ages 40 to 59 years did have a higher rate of invasive base of tongue carcinomas than white or Hispanic women in the same age group.

Figure 2. Invasive base of tongue* squamous cell carcinoma age-specific incidence rates† by sex and race/ethnicity, US, 1998 through 2003. Data are from 39 population-based registries that participate in the National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and meet high-quality data criteria.36 These registries cover approximately 83% of the US population. *Potentially human papillomavirus-associated base of tongue cancers were defined on the basis of the International Classification of Diseases-Oncology (ICD-O)-3 topography codes (as shown in Table 1). †Rates are per 100,000 population and are age-adjusted to the 2000 US standard population (19 age groups: Census P25-1130). Rates based on <6 cases are not displayed. Rates marked with a solid square are based on at least 6 but <16 cases and should be interpreted with caution. ‡Hispanic origin is not mutually exclusive from race categories (White, Black, Asian/Pacific Islander [API]).

Incidence Rates of Other Potentially HPV-associated Oropharyngeal Cancers by Sex and Race/Ethnicity

The incidence rates of potentially HPV-associated carcinomas at other oropharyngeal sites were lower than those observed for tonsil or base of tongue cancers among both men and women (Fig. 3). Similar to the other cancers, however, the incidence rates for other oropharyngeal cancers were highest among men ages 60 to 79 years, with the highest rate noted among black men ages 60 to 69 years (7.7; 95% CI, 6.9-8.6). Among women, the rate was highest among those ages 70 to 79 years (1.4; 95% CI, 1.3-1.5), and rates differed little by race or ethnicity.

Figure 3. Invasive other potentially human papillomavirus (HPV)-associated oropharyngeal* squamous cell carcinoma age-specific incidence rates† by sex and race/ethnicity, US, 1998 through 2003. Data are from 39 population-based registries that participate in the National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and meet high-quality data criteria.36 These registries cover approximately 83% of the US population. *Other potentially HPV-associated oropharyngeal cancers were defined on the basis of International Classification of Diseases-Oncology (ICD-O)-3 topography codes (as shown in Table 1). †Rates are per 100,000 population and are age-adjusted to the 2000 US standard population (19 age groups: Census P25-1130). Rates based on <6 cases are not displayed. Rates marked with a solid square are based on at least 6 but <16 cases and should be interpreted with caution. ‡Hispanic origin is not mutually exclusive from race categories (White, Black, Asian/Pacific Islander [API]).

Average Annual Number of Potentially HPV-associated Oropharyngeal and Oral Cavity Cancers and Comparison Sites

An average of 7360 cases of potentially HPV-associated invasive oropharyngeal and oral cavity cancers were diagnosed annually from 1998 through 2003 (Fig. 4). Overall, only 1.1% of the potentially HPV-associated cases were diagnosed in situ, whereas 5.4% of the comparison cases were.

Figure 4. Average annual number of squamous cell carcinomas of the oropharynx and oral cavity* potentially associated with the human papillomavirus (HPV) and the average annual number not associated with HPV, by site, US, 1998 through 2003. Data are from 39 population-based registries that participate in the National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and meet high-quality data criteria.36 These registries cover approximately 83% of the US population. *Potentially HPV-associated oropharngeal and oral cavity cancers and comparison cancers were defined on the basis of the International Classification of Diseases-Oncology (ICD-O)-3 topography codes (as shown in Table 1).

Trends in Incidence Rates of Potentially HPV-associated Invasive Oropharyngeal and Oral Cavity Cancers and Comparison Sites

The age-adjusted cancer incidence rates of tonsillar and base of tongue cancers increased significantly for men and women combined from 1998 through 2003 (APC, 3.0; P < .05 for both), whereas rates declined significantly during this period at almost all the comparison sites (Fig. 5). Laryngeal cancer incidence rates declined most sharply from 4.5 (95% CI, 4.4-4.6) in 1998 to 3.7 (95% CI, 3.7-3.8) in 2003, an APC of −3.5. Among men (Fig. 6), the rates of tonsil and base of tongue cancers increased significantly (P < .05), whereas the rate of laryngeal cancers declined (APC, −3.6; P < .05). Among women (Fig. 7), the incidence rates remained fairly stable at potentially HPV-associated sites but increased marginally at most of the comparison sites.

Figure 5. Trends in age-adjusted incidence rates* of potentially human papillomavirus (HPV)-associated and comparison sites of the oropharynx and oral cavity, US, 1998 through 2003 for men and women. Data are from 39 population-based registries that participate in the National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and meet high-quality data criteria.36 These registries cover approximately 83% of the US population. *Rates are per 100,000 population and are age-adjusted to the 2000 US standard population (19 age groups: Census P25-1130). †The annual percent change (APC) is significantly different from 0 (P < .05).

Figure 6. Trends in age-adjusted incidence rates* of potentially human papillomavirus (HPV)-associated and comparison sites of the oropharynx and oral cavity, US, 1998 through 2003 in men. Data are from 39 population-based registries that participate in the National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and meet high-quality data criteria.36 These registries cover approximately 83% of the US population. *Rates are per 100,000 population and are age-adjusted to the 2000 US standard population (19 age groups: Census P25-1130). †The annual percentage change (APC) is significantly different from 0 (P < .05).

Figure 7. Trends in age-adjusted incidence rates* of potentially human papillomavirus (HPV)-associated and comparison sites of the oropharynx and oral cavity, US, 1998 through 2003 in women. Data are from 39 population-based registries that participate in the National Program of Cancer Registries (NPCR) and/or the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program and meet high-quality data criteria.36 These registries cover approximately 83% of the US population. *Rates are per 100,000 population and are age-adjusted to the 2000 US standard population (19 age groups: Census P25-1130). †The annual percentage change (APC) is significantly different from 0 (P < .05).

DISCUSSION

The results of the current study support prior findings demonstrating that cancer incidence rates at sites within the pharynx and oral cavity were higher among blacks than other racial/ethnic groups and higher among men than women.1 These disparities in incidence rates could be due, at least in part, to differences in rates of high-risk HPV infection or to differences in the prevalence and intensity of exposure to other risk factors such as smoking or alcohol consumption. Historical data from the 1994-1995 National Health Interview Survey showed that the age-adjusted prevalence of current smoking was 18.0% among Hispanics, 26.4% among whites, and 26.5% among blacks, and that in all these racial and ethnic groups men had significantly higher rates of current smoking than woman.42 Comparable alcohol consumption prevalence data were unavailable.

Differences in rates of early diagnosis and treatment of precursor lesions among various subpopulations may also be a factor in the observed differences, although to our knowledge data to support this hypothesis are limited. In addition, in situ disease for all these cancers is rare because the premalignant lesions in these anatomic areas are largely asymptomatic and are difficult to detect visually. However, several studies have shown that people with lower income or educational attainment have an increased risk for cancers of the pharynx and oral cavity.43, 44

We were particularly concerned about the apparent increase in rates of cancers of the tonsil and base of tongue in a relatively short time period. In a recent analysis of SEER data from 1973 through 2004, Chaturvedi et al45 reported an increase in incidence rates for HPV-associated oral squamous cell carcinomas. Similar trends in incidence rates, particularly for tonsil and tongue carcinomas, have been observed previously.46, 47 Also recently, Sturgis and Cinciripini48 suggested that increases in rates of cancer in certain sites of the oropharynx, despite the dramatic reduction in tobacco exposure among Americans over the past 40 years, could be attributable to HPV infections.

This study was limited by several factors. The first involved the difficulty of determining which cancers were associated with HPV given that 1) cancer of the oropharynx and oral cavity can involve multiple adjacent sites without clear indications of where the primary tumor may have originated, 2) previously reported associations between these cancers and HPV varied substantially by subsite, and 3) these associations were generally much weaker than in other regions, such as the cervix. Because we were unable to classify the HPV status of a tumor on the basis of HPV detection, we relied on anatomic associations presented in the literature. Although multiple lines of molecular and epidemiologic evidence suggest that HPV may have an etiologic role in the development of oropharyngeal and oral cavity cancer, much of the data were based on tumor HPV detection alone, and the proportion of these cancers found to be associated with HPV has varied. This variation has been attributed in part to differences in sampling techniques and HPV detection methods.21 However, in a recent study based on data from the International Agency for Research on Cancer (IARC) multinational study, Parkin and Bray49 estimated HPV-attributable fractions of 12% for oropharyngeal cancer and 3% for oral cancer, which were substantially lower than the percentage of tumors with detectable HPV DNA.11, 14–16, 18, 22, 25, 49 Because of these factors, we limited our analysis to sites such as the tonsil and base of tongue, in which HPV has been found to be associated with all spectrums of disease.33–35 Viral copy numbers for some tumors at these sites have been comparable to those reported in other HPV-associated malignancies, such as cervical cancer.50 However, despite limiting our analysis in this way, some of the cancer cases in our analyses may have been misclassified because of the complexity of determining primary cancer sites in this anatomic region.

A second limitation to the current study was that because central cancer registries do not collect data on tobacco or alcohol use among cancer cases, we were unable to assess the extent to which these risk factors may have affected the association between HPV infection and cancer incidence rates. To establish a baseline measurement in this dataset against which to measure the future impact of the HPV vaccine on potentially HPV-associated cancers of this type, we conducted subanalyses of cancer incidence rates at anatomically similar sites within the oropharynx, oral cavity, and larynx that to our knowledge have not been previously associated with HPV infection. The results of these subanalyses may help future researchers determine the potential effectiveness of HPV vaccination in preventing oropharyngeal and oral cavity cancers over time. However, it is important to note that none of the vaccine trials have assessed the efficacy of the HPV vaccine in reducing the incidence rate of oropharyngeal and oral cavity HPV infection; therefore, more studies are needed to determine the effect that the vaccine will have on incidence rates of these cancers.

Cancers of the oral tongue, oral cavity, and larynx that are unrelated to HPV and likely attributable primarily to tobacco and alcohol use are more than twice as frequent as cancers of the oropharynx and oral cavity potentially associated with HPV, and, as noted above, the HPV vaccination has not been shown to prevent oropharyngeal and oral cavity cancers. Nevertheless, with an average of 7360 cases of potentially HPV-associated cancers occurring annually in the US, the morbidity and death associated with these cancers is still substantial. Thus, measures to reduce the risk of infection with high-risk HPV strains (including HPV vaccination and sexual risk reduction), combined with avoidance of smoking and excessive alcohol use, could potentially prevent thousands of cases of cancer of the oropharynx and oral cavity in the US each year.