Cannabis use and cancer of the head and neck: case-control study.

OBJECTIVE: To investigate whether cannabis smoking increases the risk of head and neck cancer.
DESIGN: Case-control study. SUBJECTS AND METHODS: Cases of head and neck cancer < or =55 years identified from hospital databases and the Cancer Registry, and controls randomly selected from the electoral roll completed interviewer-administered questionnaires. Logistic regression was used to estimate the relative risk of head and neck cancer.
RESULTS: There were 75 cases and 319 controls. An increased risk of cancer was found with increasing tobacco use, alcohol consumption, and decreased income but not increasing cannabis use. The highest tertile of cannabis use (>8.3 joint years) was associated with a nonsignificant increased risk of cancer (relative risk = 1.6, 95% confidence interval, 0.5-5.2) after adjustment for confounding variables.
CONCLUSIONS: Cannabis use did not increase the risk of head and neck cancer; however, because of the limited power and duration of use studied, a small or longer-term effect cannot be excluded.

Head and neck cancers represent a group of diverse cancers with varied etiology. The two main risk factors are tobacco smoking and alcohol, which have synergistic effects. Tobacco is a particularly strong risk factor for laryngeal, lip, and tongue cancers but also contributes to the risk at other upper-airway sites. Dietary factors modify risk with salted fish a risk factor for nasopharyngeal cancer, and diets rich in vitamin A and C are protective because of their antioxidative effects. A range of occupational risk factors has been identified such as the hardwood furniture industry and nasal fossa carcinoma. Viruses play an important role with DNA from Epstein-Barr virus found in nasopharyngeal cancers of all types and human papilloma virus being implicated in other cancers of the head and neck. At most sites, cancer occurs more commonly in men except postcricoid carcinoma, which predominates in women because of its association with Plummer-Vinson syndrome.

Cannabis smoking also has the potential to contribute to the risk of head and neck cancer. Cannabis smoke is qualitatively similar to tobacco smoke, although it contains up to twice the concentration of the carcinogenic polyaromatic hydrocarbons. Cannabis is less densely packed than tobacco cigarettes and tends to be smoked without filters to a smaller butt size, leading to higher concentrations of smoke inhaled. Several studies have shown precancerous histologic and genetic abnormalities in the respiratory tracts of cannabis smokers, and carcinogenic effects of cannabis smoke have been shown in vitro and in different in vivo animal models. An increased risk of lung cancer with cannabis smoking has been reported in most but not all case-control studies that have investigated the association.

Epidemiologic evidence for an association between cannabis and head and neck cancer is limited and conflicting. Case reports and case series have suggested a causative role for cannabis in cancers at different sites including lip, tongue, nasopharynx, pyriform fossa, tonsillar fossa, pharynx, and larynx. Some of the cases have been striking for their young age and lack of other risk factors, suggesting that cannabis may be an early initiator of head and neck cancers. There have been 3 case-control studies of cannabis and cancers of the oral cavity and 2 case-control studies of head and neck cancer. In only one of these five case-control studies was there a statistically significant association reported between cannabis use and cancer; however, the interpretation of these studies has been limited by the choice of controls, inability to quantify use, low response rates, and low power with the possibility of type II error.

To investigate the association between cannabis use and head and neck cancer, we undertook a case-control study in young adults in New Zealand.

Methods

Study Participants

Cases were patients with confirmed head and neck cancer aged 55 years and under at the time of diagnosis, diagnosed between January 2001 and July 2005, and identified from hospital databases and the New Zealand Cancer Registry. Cases were a mixture of prevalent and incident cases, in which the diagnosis was made within the previous five years. Subjects were excluded if they had metastasis from a distant primary other than head and neck or a histologic diagnosis of carcinoid, melanoma, or adenocystic carcinomas. Age at diagnosis, anatomic location of their malignancy, and histologic type were collected for cases. Controls (without respiratory tract cancer, head and neck cancer, or lung cancer) were randomly selected from the electoral roll and frequency matched in five-year age groups to the expected national incidence of head and neck cancer and district health boards to increase the study efficiency. The expected prevalence and regional variation of head and neck cancer were not available for this purpose. Subjects came from eight district health board regions, serving both urban and rural populations. The study was approved by the regional ethics committees, and each participant gave written informed consent.

Data Collection

Questionnaires were administered face to face by trained interviewers, usually at the home of the participant. Information on demographics (including ethnic group), smoking history, passive smoking exposure, recreational drug use, diet, occupation, income, education, alcohol consumption, and family history of malignancy was collected. A family history of upper respiratory tract cancer was defined as having a sibling or parent reported to have head and neck cancer. Cannabis smokers were asked about the amount, frequency, age of onset and duration of use, and the characteristics of their smoking. Subjects were asked to express their cannabis use in terms of frequency of joint use. If they smoked cannabis in another way (eg, pipes or bongs), they were asked to estimate the number of joints that quantity of cannabis would equate to. This conversion allowed cannabis use for all participants to be quantified in terms of the total number of joints smoked. The lifetime amount of cannabis use was expressed as joint years of use, with one joint year being equivalent to one joint per day for one year. Pack years of cigarette smoking were calculated, with one pack year equivalent to 20 cigarettes per day for one year. The questionnaire was piloted on reformed cannabis smokers before use. Patients with lung cancer were also interviewed, with the findings published separately.

Statistical Analysis

Standard methods for analysis of case-control studies were used. The mean delay from diagnosis to interview was subtracted from the date of interview to calculate a reference date for duration of exposure for each control. Relative risks were estimated by calculating odds ratios by logistic regression using SPSS version 11.0 for Mac OSX (SPSS Inc, Chicago, IL) and adjusted for confounding variables. Tests for trend in relative risks for ordered categories were also performed. Adjustment for age, joint years of cannabis smoking, and pack years of cigarette smoking was made by using them as continuous variables in the regression models. Level of income and a semiquantitative measure of total alcohol consumption were each categorized into four groups and also included in the regression models. The effects of pack years of cigarette smoking (quintiles of smoking for all subjects included) and joint years of cannabis smoking (tertiles of use for all subjects included) were assessed. The relative risks were also calculated based on cannabis use up to five years before diagnosis on the basis that exposure after that time was unlikely to have caused the malignancy. The age at which cannabis smoking started was categorized, and the relative risk associated with starting cannabis smoking under 16 years of age was compared with starting over 21 years of age.

Results

A total of 106 cases were contacted and invited for interview, of whom 81 (76%) agreed to participate. After exclusion of six in whom the diagnosis of cancer had been made more than five years before interview, there were 75 cases of head and neck cancer (Table 1). There were 493 controls contacted and invited for interview, of whom 324 (66%) agreed to participate. Five control subjects declined to provide their income and were excluded, leaving 319 control subjects for the analysis. None of these five controls reported using cannabis. The characteristics of the cases and controls are shown in Table 2.

Table 1

Anatomic site of malignancy by ICD code

Tumor site	ICD code	Number
Lip	C0-C00.9	11
Base of tongue	C01	1
Tongue	C02-C02.9	18
Floor of mouth	C04-C04.9	3
Palate	C05-C05.9	1
Mouth	C06-C06.9	2
Tonsil	C09-C09.9	12
Oropharynx	C10-C10.9	1
Nasopharynx	C11-C11.9	9
Hypopharynx	C13-C13.9	1
Pharynx	C14-C14.9	1
Nasal cavities	C30	2
Head and neck unspecified	—	3
Larynx	C32-C32.9	10

Table 2

The frequency distribution of cases and controls for selected variables

Variable	Category	Cases (n = 75)		Controls (n = 324)		Total (n = 399)
		No.	%	No.	%	No.	%
Age group	≤39	9	12.0	12	3.8	21	5.3
	40-44	12	16.0	28	8.8	40	10.2
	45-49	22	29.3	132	41.4	154	39.1
	50+	32	42.7	147	46.1	179	45.4
Sex	Male	56	74.7	144	45.1	200	50.8
	Female	19	25.3	175	54.9	194	49.2
Ethnic group	Maori	13	17.3	25	7.8	38	9.6
	Pacific Islander	3	4.0	3	0.9	6	1.5
	Other	59	78.7	291	91.2	350	88.8
Cigarettes	Never	18	24.0	169	53.0	187	47.5
	Ever	57	76.0	150	47.0	207	52.5
Pack years⁎	None	18	24.0	169	53.0	187	47.5
	1st quintile (<3.4)	5	6.7	37	11.6	42	10.7
	2nd quintile (3.4-11)	9	12.0	34	10.7	43	10.9
	3rd quintile (11.1-20)	6	8.0	34	10.7	40	10.2
	4th quintile (20.1-31.7)	17	22.7	24	7.5	41	10.4
	5th quintile (>31.7)	20	26.7	21	6.6	41	10.4
Cannabis	Never	59	78.7	280	87.8	339	86.0
	Ever	16	21.3	39	12.2	55	14.0
Joint years†	Never	59	78.7	280	87.8	339	86.0
	1st tertile (<1)	2	2.7	16	5.0	18	4.6
	2nd tertile (1-8.3)	4	5.3	15	4.7	19	4.8
	3rd tertile (>8.3)	10	13.3	8	2.5	18	4.6
Joint years of use up to 5 years before reference date	None	60	80.0	280	87.8	340	86.3
	<1	1	1.3	16	5.0	17	4.3
	1-8.3	4	5.3	15	4.7	19	4.8
	>8.3	10	13.3	8	2.5	18	4.6
Age at onset of cannabis use	None	59	78.7	280	87.8	339	86.0
	12-16 years	4	5.3	12	3.8	16	4.1
	17-20 years	7	9.3	18	5.6	25	6.3
	21+ years	5	6.7	9	2.8	14	3.6
Income $NZ	0-30,000	27	36.0	69	21.6	96	24.4
	30,001-50,000	24	32.0	75	23.5	99	25.1
	50,001-70,000	13	17.3	65	20.4	78	19.8
	>70,000	11	14.7	110	34.5	121	30.7
Education‡	Pre Sch. Cert	18	24.0	25	7.8	43	10.9
	School Cert	24	32.0	75	23.5	99	25.1
	Year 12,13	14	18.7	57	17.9	71	18.0
	Tertiary	19	25.3	162	50.8	181	45.9

Pack year: 1 pack of 20 tobacco cigarettes smoked per day for 1 year.

Joint year: 1 cannabis joint smoked per day for 1 year.

School Cert: New Zealand Secondary School Certificate Examination (age 15-16 years), year 12, 13 (aged 17-18 years).

The risk of head and neck cancer did not vary with age because of the controls being frequency matched on the age of cases in five-year age groups to improve the efficiency of the study. A family history of upper respiratory tract cancer was not significantly associated with an increased risk of head and neck cancer (relative ratio [RR] = 0.6; 95% confidence interval [CI], 0.2-1.9) (Table 3). Males had a significantly increased risk of cancers of the head and neck compared with females (RR = 3.4, 95% CI, 1.7-6.7). In the age group studied, the increased relative risk of head and neck cancer for Maori and Pacific Island people compared with all other ethnicities was not statistically significant when adjusted for age, sex, pack years of tobacco smoking, joint years of marijuana use, alcohol consumption, and income level (RR = 1.9; 95% CI, 0.8-4.4) (data not shown). When ethnic groups were further divided into New Zealand European, Chinese or Indian, Maori, and other ethnic groups, the relative risks for Maori and Chinese or Indian ethnic groups compared with New Zealand Europeans were 2.2 (95% CI, 1.0-5.2) and 3.9 (95% CI, 0.6-24.8), respectively (Table 3).

Table 3

Tobacco use, cannabis use, and alcohol consumption and risk of head and neck cancer

Characteristic	Number of cases	Number of controls	RR of head and neck cancer	95% CI
Cigarette smoking⁎
Never	18	169	1.0
Ever	57	150	2.1	(1.1-4.1)
Pack years
None	18	169	1.0
1st quintile	4	37	0.5	(0.1-2.0)
2nd quintile	10	34	2.2	(0.8-5.7)
3rd quintile	6	34	0.8	(0.3-2.4)
4th quintile	17	24	3.7	(1.5-9.0)
5th quintile	20	21	4.9	(1.9-12.4)
Cannabis use†			trend: P < 0.001
Never	59	280	1.0
Ever	16	39	1.0	(0.5-2.3)
Joint-years
None	59	280	1.0
1st tertile	2	16	0.4	(0.1-2.2)
2nd tertile	4	15	1.2	(0.3-4.2)
3rd tertile	10	8	1.6	(0.5-5.2)
Alcohol consumption‡			trend: P = 0.57
None	5	22	1.0
Light	26	125	1.2	(0.3-4.0)
Moderate	31	157	1.9	(0.5-6.7)
Heavy	13	15	5.7	(1.2-25.9)
Ethnic group			trend: P = 0.008
NZ European	55	265	1.0
Maori/PI	16	27	2.2	(1.0-5.2)
Chinese/Indian	2	5	3.9	(0.6-24.8)
All other	2	22	0.4	(0.1-2.2)
Income $NZ
0-30,000	27	69	1.0
30,001-50,000	24	75	0.8	(0.4-1.8)
50,001-70,000	13	65	0.4	(0.2-1.0)
>70,000	11	110	0.2	(0.1-0.4)
Sex			trend: P < 0.001
Female	19	175	1.0
Male	56	144	3.4	(1.7-6.7)
Family history of respiratory tract cancer
No	69	290	1.0
Yes	6	29	0.6	(0.2-1.9)

Tobacco use (pack years) quintiles were <3.4, 3.4-11, 11.1-20, 20.1-31.7, and >31.7.

Cannabis use (joint years) tertiles were <1, 1-8.3, and >8.3.

Adjusted for age, sex, ethnicity, alcohol consumption, income, and joint years of cannabis use.

Adjusted for age, sex, ethnicity alcohol consumption, income, and pack years of cigarette smoking.

Adjusted for age, sex, ethnicity, income, pack-years of cigarette smoking, and joint years of cannabis use. Alcohol consumption was measured by using a semiquantitative scale.

A low level of income was strongly associated with an increased risk of cancer of the head and neck, after adjustment for potential confounders, with the risk for those earning more than $70,000 per annum about one fifth of those earning less than or equal to $25,000 per annum (RR = 0.2; 95% CI, 0.1-0.4).

The relative risk from ever smoking cigarettes was statistically significant (RR = 2.1; 95% CI, 1.1-4.1) (Table 3). The relative risk of head and neck cancer increased with successive increases in cigarette smoking, with the highest quintile of pack years of cigarette smoking having a relative risk of 4.9 (95% CI, 1.9-12.4) after adjusting for confounding variables. This increased risk was 4% for each pack year of exposure (95% CI, 2%-6%) after adjustment for confounding variables (Table 4). A significant increase in the relative risk with increasing alcohol consumption was observed (test for trend, P < 0.01), and the relative risk of head and neck cancer from heavy alcohol consumption compared with nondrinkers was 5.7 (95% CI, 1.2-25.9).

Table 4

Cannabis use and tobacco use as continuous variables and relative risk of head and neck cancer

Characteristic	Relative risk of head and neck cancer	95% CI
Pack year cigarette smoking⁎	1.04	(1.02-1.06)
Joint year cannabis smoking†	1.04	(0.97-1.11)
Joint years to 5 years before diagnosis‡	1.08	(0.77-1.53)

Adjusted for age, sex, ethnicity, alcohol consumption, income, and joint years of cannabis use.

Adjusted for age, sex, ethnicity, alcohol consumption, income, and pack years of tobacco use.

Adjusted for age, sex, ethnicity, alcohol consumption, income, and pack years of tobacco use.

The median duration of cannabis use was 10.5 years among controls (range, 0.25 to 29 years) and 25 years among cases (range, two to 32 years). Ever use of cannabis was not associated with a significantly increased risk of head and neck cancer (Table 3). The risk associated with the highest tertile of cannabis use (>8.3 joint-years of exposure) was not statistically significant, RR=1.6 (95% CI, 0.5 to 5.2) after adjustment for confounding variables including tobacco smoking, alcohol consumption, and level of income. When cannabis use was fitted as a continuous variable, the estimated 4% increase in the risk of head and neck cancer for each joint year of exposure (RR = 1.04; 95% CI, 0.97-1.11) was not statistically significant, after adjustment for confounding variables (Table 4). When cannabis use in the five years before diagnosis or reference date was excluded, the risk from each joint year of exposure increased but again was not statistically significant (RR = 1.08; 95% CI, 0.77-1.53). Adjustment for whether cannabis use was solely from joints or combined with other methods of use did not appreciably alter these results. The age at which subjects started smoking cannabis was not associated with the risk of head and neck cancer (data not shown). Adjustment for occupational risk of respiratory tract cancer or the consumption of different food groups did not appreciably alter the estimate obtained for the relative risk of head and neck cancer from cannabis use.

Discussion

This population-based study did not find a statistically significant increase in the risk of head and neck cancer in young adults from cannabis use. However, the median duration of cannabis use in cases and controls may have been too short for longer-term risks to be observed.

The major limitation of this study was the small sample size, with 75 cases of head and neck cancer and 319 controls being included in the analysis. For example, it was not possible to assess the RR associated with the different types of cancer that make up the group of head and neck cancers. Although tobacco smoking dominates the etiology of head and neck cancers as a group, other diverse factors, including viral infection, alcohol, diet, and occupation, contribute to specific cancers within the group, and there may be a differential risk associated with long-term cannabis use for the different cancers within this group. The strong association of low income with increased risk suggests that unadjusted residual confounding may have influenced the associations observed. Based on the frequency of cannabis use of 12% in our population, it is estimated that about 150 subjects with cancer at a specific site would be required to provide 80% statistical power to identify a two-fold increase in risk (alpha = 0.05) using a 2:1 ratio of controls to cases.

Cases were identified from both the National Cancer Registry and from hospital outpatient and discharge databases to ensure case ascertainment was as complete as possible. We used a population-based control group randomly selected from the electoral roll rather than hospital-based controls because the latter is susceptible to significant bias because of the many medical conditions associated with cannabis use. In New Zealand, about 93% of adults are listed on the electoral roll, and this approach was used to avoid the potential sources of bias inherent in the selection of the control groups in previous case-control studies of head and neck cancer. The study of Zhang et al selected a control group from healthy blood donors who were less likely to have a history of substance abuse, resulting in a prevalence of cannabis use considerably less that that reported from a comparative United States population. It has been suggested that these characteristics of the control group may have contributed to the increased risk of head and neck cancer with cannabis smoking reported in that study. The two British case-control studies of oral cancer and cannabis use recruited controls through the case’s general practitioner, although the method of selection was not stated and on occasions other local general practitioners had to be used. The most recent case-control study from the United States recruited controls by canvassing the neighborhood of each enrolled case. This approach may have made detection of an association less likely as cannabis use is likely to be similar within distinct neighborhoods. This may explain the higher rate of cannabis use among the control group than would have been expected from the United States population data, limiting the potential to find an association.

In our study, no mention was made of the primary risk factor of interest to avoid recruitment bias with either the cases or the controls. To minimize response bias, the interviewer did not state the specific research hypothesis and took a detailed history of all well known risk factors. Recall of the amount of cannabis smoked over a long period of time may have been difficult for some subjects; however, this is likely to have been similar for both cases and controls. The exposure assessed was joint years of cannabis use, which combines both the intensity (amount and frequency) and duration of use. This approach follows the current convention for quantifying lifelong cannabis consumption and recognizes the evidence that the risk of lung cancer with cigarette smoking is related to both intensity and duration of use. The administration of the questionnaire by trained interviewers in this way overcame difficulties of incomplete data from self-completed questionnaires, which has occurred previously, and allowed quantification of cannabis use, which may have varied considerably during the smoker’s lifetime.

The study was limited to subjects 55 years and under because of case series that have suggested that the role of cannabis smoking in head and neck cancer may be particularly strong in this age group. Furthermore, time trends indicate a progressively increased use of cannabis in New Zealand since the 1950s, resulting in this age group being predominantly exposed to cannabis compared with older adults.

This study was undertaken in parallel with a case-control study of the role of cannabis and lung cancer, and it is informative to contrast and compare the findings relating to cancers of the upper and lower respiratory tract. The relative risks of cannabis use for head and neck cancer were lower than those found for lung cancer and did not reach statistical significance. For example, when the tertiles of cannabis use in the analysis of lung cancer were used, the relative risk for lung cancer in those with the highest tertile of cannabis use compared with nonusers was 5.7 (95% CI, 1.5-21.6) but only 1.6 (95% CI, 0.5-7.0) for head and neck cancer. When joint years of use were fitted as a continuous variable, the risk for head and neck cancer was not significant and was lower than the significant risk for lung cancer (4% vs 8% per joint year, respectively). When the relationship was assessed for cannabis use up to 5 years before diagnosis, we found the magnitude of the risk was increased for both head and neck cancers and lung cancer to 8% and 10%, respectively, but with the former estimate not being statistically significant. This observation suggests a biological plausibility in that it is the exposure several years before the diagnosis of the malignancy that is relevant to its causation. The risk of head and neck cancer did not increase with early onset of cannabis use or whether joints alone or mixed methods of cannabis use were included.

Conclusion

No significantly increased risk of head and neck cancer from cannabis use was found in young adults; however, because of the limited power and duration of use studied, a small or longer-term effect cannot be excluded. Likewise, because head and neck cancer is a heterogeneous group, the negative overall finding does not preclude an effect of cannabis on specific histological subtypes. Cancers of the head and neck are the fourth most common cause of cancer worldwide, with approximately 500,000 new cases diagnosed each year. Larger studies are required to assess the risk of cancers of the head and neck associated with cannabis use for the different cancer sites of the head and neck.

Author Contributions

Sarah Aldington, study design, data collection, analysis, writing; Matire Harwood, data collection; Brian Cox, study design, analysis, writing; Mark Weatherall, study design, writing; Lutz Beckert, data collection; Anna Hansell, analysis, writing; Alison Pritchard, database development; Geoffrey Robinson, study design, writing; Richard Beasley, study design, writing.

Financial Disclosure

Supported by The New Zealand Ministry of Health, The Hawke’s Bay Medical Research Foundation, and GlaxoSmithKline (UK). Associate Professor Brian Cox was funded by the Director’s Cancer Research Trust. Dr Anna Hansell is a Wellcome Trust Intermediate Clinical Fellow supported by Grant Number 075883.