This cohort study assesses the association of smoking, drinking, and dietary habits with head and neck cancer risk among Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial patients.
Key Points
Question
How are smoking, drinking, and dietary habits associated with the risk of developing head and neck cancer (HNC)?
Findings
In this nested cohort study of 139 926 Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial participants, HNC risk associated with smoking increased with closer head and neck subsite proximity to the lungs. HNC risk increased with heavy drinking and decreased with healthy eating, as scored by Healthy Eating Index 2015.
Meaning
HNC risk associated with smoking varied significantly by subsite; heavy drinking was associated with greater HNC risk while healthy eating was associated with a modest reduction in HNC risk.
Abstract
Importance
There is a paucity of large-scale prospective studies evaluating the risk of developing head and neck cancer (HNC) associated with smoking, drinking, and dietary habits.
Objective
To determine the association of smoking, drinking, and dietary habits with the risk of developing HNC.
Design, Setting, and Participants
A nested cohort survival analysis of Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial participants was performed. Participants were between 55 and 74 years of age and recruited at 10 centers across the US from November 1993 to July 2001. Participants who developed HNC were matched with controls based on demographics and family history of HNC for analysis of smoking habits; for the analysis of drinking and dietary habits, matching was performed on smoking status and duration in addition to demographics and family history of HNC. Data analysis was performed from January to November 2023.
Exposures
Smoking, drinking, and dietary habits.
Main Outcome and Measure
Diagnosis of HNC.
Results
In total, 139 926 participants (51% female; mean [SD] age, 62.6 [5.4] years) were included in the analysis of smoking habits with a median (IQR) follow-up time of 12.1 (10.3-13.6) years, 571 of whom developed HNC. HNC risk associated with smoking increased the closer the proximity of the head and neck subsite to the lungs, with the greatest risk associated with smoking observed in laryngeal cancer (current smoker hazard ratio [HR], 9.36; 95% CI, 5.78-15.15 compared to a nonsmoker). For analysis of drinking and dietary habits, 94 466 participants were included in the analysis of smoking habits with a median (IQR) follow-up time of 12.2 (10.5-13.6) years, 264 of whom developed HNC. HNC risk increased with heavy drinking (HR, 1.85; 95% CI, 1.44-2.38) and decreased with consumption of whole grains (HR, 0.78; 95% CI, 0.64-0.94/oz per day), whole fruits (HR, 0.90; 95% CI, 0.82-0.98/cup per day), and overall healthy eating, as scored by Healthy Eating Index 2015 (HR, 0.87; 95% CI, 0.78-0.98/10 points).
Conclusions and Relevance
In this nested cohort study, the risk of HNC associated with smoking was higher for subsites that were closer to the lungs; heavy drinking was associated with greater HNC risk, while healthy eating was associated with a modest reduction in HNC risk.
Introduction
Head and neck cancer was the eighth most common cancer worldwide in 2020 (878 348 new cases and 444 347 deaths) and was estimated to account for 3.4% (66 920 new cases) of cancer diagnoses and 2.5% (15 400 deaths) of cancer deaths in the US in 2023.1,2 Although smoking and drinking are known risk factors for head and neck cancer, the degree to which they are associated with various subsites of the head and neck is not well established.3 Similarly, questions remain regarding the risk of head and neck cancer associated with drinking in moderation or drinking various types of alcohol; there is even less consensus regarding the effects of dietary habits on the development of head and neck cancer.
Prior studies evaluating the effects of tobacco, alcohol, and diet have used data from large-scale prospective trials, such as the European Prospective Investigation into Cancer and Nutrition (EPIC),4 and pooled analysis of respective studies, such as the INHANCE Consortium database.5,6,7 Large-scale prospective trials are ideal for identification of novel risk factors and quantification of the effect size of known risk factors due to their large sample sizes and ability to eliminate recall bias; however, few of these studies exist due to their prohibitive cost.
Analogous to the EPIC study in Europe, the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial in the US was a multisite, large-scale, prospective trial that collected detailed data on smoking, drinking, and diet via questionnaires and then followed up with patients prospectively to collect data on cancer diagnoses, including head and neck cancer.8 This study aimed to use data from the PLCO cohort to further elucidate the role of smoking, drinking, and diet in the development of head and neck cancer. In addition, we have included relevant, available human papillomavirus (HPV) data from prior studies using PLCO biospecimens as able.
Methods
PLCO Trial Background
The PLCO trial was a randomized clinical trial of screening examinations for prostate, lung, colorectal, and ovarian cancers.9 Participants were randomly assigned to either an intervention arm, where they received screening examinations for prostate, lung, colorectal, and ovarian cancers, or a control arm, where they received usual care. The trial enrolled approximately 155 000 participants at 10 centers across the US between November 1993 and July 2001 and collected data on all cancer diagnoses, including head and neck cancer, through 2009 (median follow-up time, 11.3 years). Participants were initially between 60 and 74 years of age, which was revised to 55 to 74 years of age in January 1996. Exclusion criteria included prior history of PLCO cancer, current treatment for any nonskin cancer, or any prior surgery, hormonal therapy, or screening testing that would render screening unnecessary or ineffective (see the National Cancer Institute’s Cancer Data Access System website for further information on the PLCO trial, including a complete list of exclusion criteria10 and a list of major publications from the PLCO trial11). This study was determined to be exempt from review by the Washington University School of Medicine Institutional Review Board because it did not contain identifiable data or specimens.
Study Data
Data were obtained from the PLCO head and neck, lung, and upper gastrointestinal datasets, which included data from the baseline questionnaire (BQ) completed by participants after enrollment in the trial. The BQ includes information on patient demographics, comorbidities, smoking, weight, height, and family history. Individuals’ race and ethnicity was provided by the participants in their responses to options defined by the PLCO investigators in the BQ. Additional data from the dietary history questionnaire (DHQ) were obtained for a subset of the study population and included extensive information on patient drinking and dietary habits. HPV data from patient serum samples from a recent study evaluating the timing of HPV16-E6 antibody seroconversion before oropharyngeal squamous cell carcinoma were also included in the analysis.12
Study Populations, Study Design, and Statistical Analysis
Patients who completed the BQ and had complete information on demographics, smoking status, duration, and frequency without prior history of head and neck cancer were included in the BQ study population (Figure 1). The majority of patients in the BQ study population also completed the DHQ and were included in the DHQ study population; patients with missing information regarding drinking or dietary habits were excluded from this study population. Drinking amount categories were defined using the National Health Interview Survey’s definitions from the Centers for Disease Control and Prevention: infrequent (≤1 drink/month), light (≤3 drinks/week), moderate (3-7 drinks/week for female individuals; 3-14 drinks/week for male individuals), and heavy (≥7 drinks/week for female individuals; ≥14 drinks/week for male individuals).13
Figure 1. Graphical Depiction of Matched Cohort Study Design.
The matched cohort study design for this study is displayed. Cases and controls were initially matched in the baseline questionnaire (BQ) study population based on select demographics (sex, age, race and ethnicity, education) and any risk factors for head and neck cancer (HNC) identified in preliminary multivariable analysis. Identical matching was performed in the dietary history questionnaire (DHQ) study population with the inclusion of smoking status and duration as matching criteria. PLCO indicates Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial.
Descriptive statistics were used to explore the distribution of patient demographics, comorbidities, height, weight, and family history, as well as patient smoking, drinking, and dietary habits. The association of patient characteristics and risk factors for development of head and neck cancer in the BQ study population was explored via Cox proportional hazards regression. Multivariable models with variables that were significant on univariable analysis and suspected to be of clinical importance were used to identify risk factors for head and neck cancer.
A matched cohort study design was used (Figure 1). Exact 2:1 matching was performed in the BQ study population on select demographics (sex, age, race and ethnicity, and education) and any other identified risk factors for head and neck cancer excluding smoking habits. Analysis of the risk associated with smoking status, duration, and frequency was then performed including analysis of cancer risk associated with smoking status by head and neck subsite. Data for lung and upper gastrointestinal cancer diagnoses were analyzed in identical fashion (ie, nested cohort study design) and included for reference. A nested matched cohort study was then performed. Exact 2:1 matching was then performed in the DHQ study population on select demographics (sex, age, race and ethnicity, and education) and identified risk factors including smoking habits, and analysis of cancer risk associated with drinking and dietary habits was performed in the DHQ study population. Prior to matching in the DHQ study population, multivariable models were run to assess for an interaction effect between smoking and drinking. Hazard ratios (HRs) and 95% CIs were reported for Cox proportional hazards models. All statistical analysis was performed in R, version 4.1.2 (R Foundation for Statistical Computing), using the dplyr, version 1.1, MatchIt, version 4.5, and RMS, version 6.4, packages.14,15,16,17 The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Results
Patient Demographics and Comorbidities (BQ)
In total, 139 926 of the 154 915 participants in the PLCO study met inclusion criteria and were included in the BQ study population (Table 1), which had a median (IQR) follow-up time of 12.1 (10.3-13.6) years. The study population had a slight female predominance (n = 71 801 [51%]) and was mostly married (n = 106 269 [76%]). The mean (SD) age was 62.6 (5.4) years. Most patients had attended college at least in some capacity (n = 80 324 [57%]) and were either currently working (n = 55 992 [40%]) or retired (n = 59 916 [43%]). The most commonly observed comorbidity was hypertension (n = 47 467 [34%]).
Table 1. Baseline Questionnaire Patient Demographic Characteristics.
| Characteristic | Patients, No. (%) | |
|---|---|---|
| HNC diagnosis (n = 571) | Overall (N = 139 926) | |
| Sex | ||
| Female | 133 (23) | 71 801 (51) |
| Male | 438 (77) | 68 125 (49) |
| Age, y | ||
| <60 | 156 (27) | 46 933 (34) |
| 60-64 | 188 (33) | 43 148 (31) |
| 65-69 | 151 (27) | 31 240 (22) |
| ≥70 | 76 (13) | 18 605 (13) |
| Race and ethnicity | ||
| American Indian | 3 (1) | 355 (1) |
| Asian or Pacific Islander | 22 (4) | 5954 (4) |
| Black | 26 (4) | 6967 (5) |
| Hispanic | 12 (2) | 2630 (2) |
| White | 508 (89) | 124 020 (88) |
| Marital status | ||
| Married | 402 (70) | 106 269 (76) |
| Divorced | 85 (15) | 15 266 (11) |
| Widowed | 61 (11) | 12 199 (9) |
| Never married | 18 (3) | 4809 (3) |
| Separated | 5 (1) | 1383 (1) |
| Education | ||
| <12 y | 63 (11) | 9947 (7) |
| High school | 117 (20) | 32 113 (23) |
| Post−high school training, not college | 86 (15) | 17 542 (13) |
| Some college | 119 (21) | 30 571 (22) |
| College graduate | 86 (15) | 23 903 (17) |
| Postgraduate | 100 (18) | 25 850 (18) |
| Occupation | ||
| Working | 214 (37) | 55 992 (40) |
| Retired | 279 (49) | 59 916 (43) |
| Unemployed or disabled | 45 (8) | 4789 (3) |
| Homemaker | 22 (4) | 16 168 (12) |
| Other | 11 (2) | 3061 (2) |
Abbreviation: HNC, head and neck cancer.
Incidence of Head and Neck Cancer Diagnosis (BQ)
In the BQ study population, 571 cases of head and neck cancer were diagnosed in 1 626 064 person-years, corresponding to a crude incidence rate of 35.12 cases per 100 000 person-years (95% CI, 32.3-38.12 cases/100 000 person-years). The mean (SD) age at head and neck cancer diagnosis was 69.23 (6.1) years. The most common head and neck subsite was the larynx (n = 203 [36%]) followed by the oral cavity (n = 131 [23%]) and the oropharynx (n = 107 [19%]).
Smoking Habits Associated With Head and Neck Cancer Risk (BQ)
In the BQ matched cohort, study participants who developed head and neck cancer were matched to 2 healthy controls based on sex, age, race and ethnicity, and education as well as identified risk factors for head and neck cancer (divorced, widowed, unemployed/disabled, family history of head and neck cancer); 5 cases were unable to be matched. After matching, both prior and current smokers were at significantly increased risk of developing head and neck cancer compared to nonsmokers with HRs of 1.77 (95% CI, 1.43-2.19) and 4.01 (95% CI, 3.18-5.04), respectively (Table 2). Prior smokers who quit within the last 10 years were at a higher risk of head and neck cancer (HR, 2.72; 95% CI, 2.06-3.58) when compared to nonsmokers than smokers who quit more than 20 years ago (HR, 1.29; 95% CI, 0.98-1.70). Head and neck cancer risk associated with smoking increased in a stepwise fashion with smoking duration, with the greatest risk observed in patients who had smoked for greater than 40 years (HR, 3.94; 95% CI, 3.14-4.94). A similar but less pronounced association was observed with smoking frequency and the composite metric of pack-years, with the highest risk observed in patients who typically smoked more than 30 cigarettes per day (HR, 2.86; 95% CI, 2.24-3.64) and patients with a smoking history in excess of 100 pack-years (HR, 3.46; 95% CI, 2.76-4.32), respectively.
Table 2. Baseline Questionnaire Smoking Habits Analysis.
| Characteristic | HNC diagnosis (n = 571) | Overall (N = 139 926) | HR (95% CI) | Matched cohort HR (95% CI)a |
|---|---|---|---|---|
| Smoking status | ||||
| Nonsmoker | 122 (21) | 67 018 (48) | 1 [Reference] | 1 [Reference] |
| Current smoker | 190 (33) | 13 858 (10) | 8.32 (6.63-10.45) | 4.01 (3.18-5.04) |
| Prior smoker | 259 (45) | 59 050 (42) | 2.45 (1.98-3.04) | 1.77 (1.43-2.19) |
| Quit <10 y ago | 88 (15) | 12 819 (9) | 3.97 (3.02-5.23) | 2.72 (2.06-3.58) |
| Quit 10-20 y ago | 86 (15) | 17 354 (12) | 2.80 (2.12-3.68) | 1.80 (1.36-2.37) |
| Quit >20 y ago | 85 (15) | 28 877 (21) | 1.61 (1.22-2.13) | 1.29 (0.98-1.70) |
| Years smoked | ||||
| Nonsmoker | 122 (21) | 67 018 (48) | 1 [Reference] | 1 [Reference] |
| 1-15 | 46 (8) | 17 968 (13) | 1.41 (1.00-1.98) | 1.30 (0.93-1.83) |
| 16-30 | 77 (14) | 23 002 (16) | 1.85 (1.39-2.47) | 1.39 (1.05-1.85) |
| 31-40 | 125 (22) | 17 706 (13) | 4.07 (3.17-5.22) | 2.41 (1.88-3.09) |
| >40 | 201 (35) | 14 232 (10) | 8.57 (6.84-10.73) | 3.94 (3.14-4.94) |
| Cigarettes/d | ||||
| Nonsmoker | 122 (21) | 67 018 (48) | 1 [Reference] | 1 [Reference] |
| 1-10 (Light) | 51 (9) | 19 057 (14) | 1.48 (1.08-2.08) | 1.62 (1.16-2.25) |
| 11-30 (Moderate) | 257 (45) | 41 049 (29) | 3.55 (2.86-4.40) | 2.24 (1.81-2.79) |
| >30 (Heavy) | 141 (25) | 12 802 (9) | 6.41 (5.03-8.16) | 2.86 (2.24-3.64) |
| Pack-years | ||||
| Nonsmoker | 122 (21) | 67 018 (48) | 1 [Reference] | 1 [Reference] |
| <25 | 30 (9) | 16 433 (12) | 1.01 (0.67-1.50) | 1.17 (0.79-1.75) |
| 25-49 | 75 (45) | 16 881 (12) | 2.47 (1.86-3.30) | 1.79 (1.34-2.40) |
| 50-99 | 133 (25) | 22 858 (16) | 3.30 (2.58-4.22) | 2.00 (1.57-2.56) |
| ≥100 | 211 (25) | 16 736 (12) | 7.53 (6.02-9.41) | 3.46 (2.76-4.32) |
Abbreviations: HNC, head and neck cancer; HR, hazard ratio.
Exact 2:1 matching on sex, age, race and ethnicity, education, divorced, widowed, unemployment/disability, family history of HNC; 5 cases could not be matched.
Head and Neck Cancer Risk Associated With Smoking Status by Subsite (BQ)
The associations of smoking status with diagnosis of head and neck cancer by subsite for the BQ matched cohort are presented in Figure 2. The head and neck cancer risk associated with smoking increased with closer proximity of the head and neck subsite to the lungs, with the greatest risk associated with smoking observed in laryngeal cancer (current smoker HR, 9.36; 95% CI, 5.78-15.15 compared to a nonsmoker).
Figure 2. Risk Associated With Current and Prior Smoking by Cancer Subsite.
Figure 2 displays the hazard ratios (HRs) associated with cancer for current and prior smoking by cancer subsite, along with the 95% CIs. The HRs were obtained in the matched cohort after 2:1 exact matching on sex, age, race and ethnicity, education, divorced, widowed, unemployment/disability, and family history of cancer of interest. Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial data for lung and upper gastrointestinal cancer diagnoses were also available; they were analyzed in identical fashion (ie, nested cohort study design) and included for reference.
Oropharyngeal Cancer Risk Associated With HPV16-E6 Antibody Seroconversion (BQ)
Information regarding HPV16-E6 antibody seroconversion was available for 842 patients in the BQ study population, including 43 cases of oropharyngeal cancer. After matching on sex, age, race and ethnicity, smoking status, and years smoked, HPV16-E6 antibody seroconversion was associated with significantly increased odds of oropharyngeal cancer risk (HR, 5.84; 95% CI, 3.11-11.00).
Patient Demographics and Incidence of Head and Neck Cancer Diagnosis (DHQ)
In total, 94 466 of the 154 915 participants in the PLCO study met inclusion criteria and were included in the DHQ study population (Table 2), which had a median (IQR) follow-up time of 12.2 (10.5-13.6) years. Patient demographics and comorbidities were similar to those observed in the BQ study population. In the DHQ study population, 264 cases of head and neck cancer were diagnosed in 1 136 854 person-years, corresponding to a crude incidence rate of 23.22 cases per 100 000 person-years (95% CI, 20.50-26.20 cases/100 000 person-years).
Drinking Habits Associated With Head and Neck Cancer Risk (DHQ)
In the DHQ matched cohort, study participants who developed head and neck cancer were again matched to 2 healthy controls based on the same criteria used for the BQ matched cohort in addition to smoking status and duration; 15 cases were unable to be matched. After matching, the estimates of head and neck cancer risk associated with prior and current drinking status were imprecise (Table 3). Current heavy drinking was associated with head and neck cancer risk (HR, 1.63; 95% CI, 1.16-2.29), while current light and moderate drinking were not. Accounting for prior drinking history strengthened the association between any heavy drinking and head and neck cancer diagnosis (HR, 1.85; 95% CI, 1.44-2.38). Head and neck cancer risk was more strongly associated with a history of heavy consumption of beer or liquor rather than a history of heavy wine consumption (Table 3).
Table 3. Dietary Health Questionnaire Dietary and Drinking Habits Analysis.
| Characteristic | Patients, No. (%) | HR (95% CI) | Matched cohort HR (95% CI)a | |
|---|---|---|---|---|
| HNC diagnosis (n = 264) | Overall (n = 94 466) | |||
| Drinking status | ||||
| Never | 9 (3) | 9918 (10) | 1 [Reference] | 1 [Reference] |
| Prior | 45 (17) | 14 046 (15) | 3.25 (1.67-6.33) | 1.55 (0.76-3.18) |
| Current | 210 (80) | 70 766 (75) | 3.61 (1.77-7.39) | 1.43 (0.73-2.79) |
| Current drinking amountb | ||||
| Abstainer/infrequent | 63 (24) | 29 948 (31) | 1 [Reference] | 1 [Reference] |
| Light | 72 (27) | 34 809 (37) | 0.97 (0.69-1.37) | 0.76 (0.54-1.08) |
| Moderate | 52 (20) | 17 908 (19) | 1.33 (0.92-1.93) | 0.90 (0.62-1.30) |
| Heavy | 77 (29) | 12 065 (13) | 2.98 (2.14-4.16) | 1.63 (1.16-2.29) |
| Ever heavy drinker | ||||
| No | 112 (42) | 69 261 (73) | 1 [Reference] | 1 [Reference] |
| Yes | 152 (58) | 25 469 (27) | 3.72 (2.92-4.75) | 1.85 (1.44-2.38) |
| Heavy beer drinker | ||||
| No | 164 (62) | 82 566 (87) | 1 [Reference] | 1 [Reference] |
| Yes | 100 (38) | 12 164 (13) | 4.27 (3.33-5.47) | 1.69 (1.31-2.19) |
| Heavy liquor drinker | ||||
| No | 185 (70) | 85 707 (90) | 1 [Reference] | 1 [Reference] |
| Yes | 79 (30) | 9023 (10) | 4.08 (3.13-5.31) | 1.63 (1.24-2.15) |
| Heavy wine drinker | ||||
| No | 247 (94) | 89 766 (95) | 1 [Reference] | 1 [Reference] |
| Yes | 17 (6) | 4964 (5) | 1.25 (0.76-2.04) | 1.26 (0.74-2.12) |
| HEI-2015 total score, mean (SD) | 62.6 (10.4) | 66.6 (9.66) | 0.65 (0.58-0.73)/10 Points | 0.87 (0.78-0.98)/10 Points |
| Dark green/orange vegetables, mean (SD) | 0.24 (0.24) | 0.32 (0.35) | 0.36 (0.21-0.61)/Cup per day | 0.66 (0.41-1.06)/Cup per day |
| Whole fruit, mean (SD) | 1.59 (1.45) | 2.22 (1.87) | 0.74 (0.67-0.82)/Cup per day | 0.90 (0.82-0.98)/Cup per day |
| Whole grains, mean (SD) | 0.82 (0.71) | 0.98 (0.75) | 0.69 (0.57-0.85)/oz per Day | 0.78 (0.64-0.94)/oz per Day |
| Nonwhole grains, mean (SD) | 3.80 (1.79) | 3.63 (1.96) | 1.04 (0.98-1.10)/oz per Day | 0.96 (0.91-1.02)/oz per Day |
| Potatoes, mean (SD) | 0.45 (0.32) | 0.39 (0.32) | 1.66 (1.20-2.29)/Cup per day | 1.14 (0.79-1.65)/Cup per day |
Abbreviations: HEI-2015, Healthy Eating Index 2015; HNC, head and neck cancer; HR, hazard ratio.
Exact 2:1 matching on sex, age, race and ethnicity, education, divorced, widowed, unemployment/disability, smoking status, years smoked, and family history of cancer of interest; 15 cases could not be matched.
Drinking quantity definitions (as defined by Centers for Disease Control and Prevention): infrequent (≤1 drink/mo), light (≤3 drinks/week), moderate (3-7 drinks/week for female individuals; 3-14 drinks/week for male individuals), and heavy (≥7 drinks/week for female individuals; ≥14 drinks/week for male individuals).
Dietary Habits Associated With Head and Neck Cancer Risk (DHQ)
The association of dietary habits with head and neck cancer diagnosis was also examined in the DHQ matched cohort (Table 3). Overall healthy eating, as scored by the Healthy Eating Index 2015 Total Score (HEI-2015),18 was associated with reduced risk of head and neck cancer (HR, 0.87; 95% CI, 0.78-0.98/10 points). A similar association was observed for increased rates of consumption of whole grains (HR, 0.78; 95% CI, 0.64-0.94/oz per day) and whole fruits (HR, 0.90; 95% CI, 0.82-1.00/cup per day). Although estimates were imprecise, the data suggest a potentially clinically important reduction in head and neck cancer risk with increased consumption of dark green and orange vegetables (HR, 0.66; 95% CI, 0.41-1.06/cup per day).
Smoking, Drinking, and Dietary Habits Associated With Risk of Head and Neck Squamous Cell Carcinoma
An additional sensitivity analysis, which only included cases of head and neck squamous cell carcinoma and excluded other histologic types, was conducted and is available in eTables 1 and 2 and the eFigure in Supplement 1. Overall, limiting the analysis to cases of head and neck squamous cell carcinoma resulted in similar trends and strengthened associations with smoking, drinking, and dietary risk factors.
Discussion
In this study, we observed that the cancer risk associated with smoking increased with closer proximity of the head and neck subsite to the lungs. The head and neck subsite with the highest cancer risk associated with smoking was the larynx, followed by the oropharynx, then the oral cavity. These findings may be secondary to airflow patterns during cigarette smoking, as the smoke is funneled from the oral cavity to the oropharynx and into the larynx, or varying levels of tissue susceptibility to cigarette smoke. Although not explicitly stated in this manner, these findings are consistent with prior large-scale studies on head and neck cancer risk associated with smoking.4,6,19 When comparing the HRs from our study directly to those from the Agudo et al4 analysis of the EPIC trial, we observed lower HRs for the risk of laryngeal cancer associated with current and prior smoking and higher HRs for the risk of cancer associated with prior smoking at all other sites. All of the HRs for cancer risk associated with current and former smoking from our analysis of the PLCO trial are within the 95% CIs for the HRs reported in the EPIC study.4
We also observed statistically significant associations between head and neck cancer risk and smoking status, duration, frequency, and the composite metric pack-years. A greater stepwise association was observed between head and neck cancer risk and smoking duration than was observed with smoking frequency. A possible explanation for this finding is that smoking frequency, reported as cigarettes per day, has been shown to inaccurately reflect carcinogen exposure.20
Our study observed an increased risk of head and neck cancer with heavy drinking, predominantly driven by heavy consumption of beer and liquor rather than heavy consumption of wine. Interestingly, increased risk of head and neck cancer was not observed with light or moderate drinking, and an interaction effect was not observed between smoking and drinking. This is contrary to prior studies, which suggested a dose-response trend with an increasing risk of head and neck cancer as alcohol consumption increased.7,21,22 However, it is worth noting that at least 1 of these studies from the INHANCE consortium used a modeling approach that, in effect, assumed a dose-response trend.7 A pooled analysis of the INHANCE cohort also showed that drinkers who consumed of beer and liquor had a higher risk of head and neck cancer when compared to drinkers who consumed wine with exception of drinkers who consumed excessive amounts of wine (>30 drinks/week). Further studies are needed to determine whether wine truly behaves differently than beer and liquor or whether it is simply a strong proxy for socioeconomic status, as well as if there is a safe level of alcohol consumption with regard to risk of developing head and neck cancer.
This study observed a modest reduction in head and neck cancer risk associated with overall healthy eating, as indicated by higher total scores on the HEI-2015, and with increased consumption of whole grains, whole fruits, and dark green and orange vegetables. No association with incidence of head and neck cancer was observed with increased consumption of nonwhole grains or potatoes, which are often included in composite vegetable indices but are frequently eaten fried and prepared in ways that are unlikely to provide health benefits. While prior studies from the INHANCE consortium and the EPIC trial have shown reduced head and neck cancer risk associated with fruit and vegetable intake,23,24,25 we are not aware of prior large-scale studies that have shown a reduced head and neck cancer risk associated with HEI-2015 scores or whole grains. There has also been suggestion of increased risk of head and neck cancer with increased consumption of red meat and processed meat,5,26 which was not observed in our study after controlling for other risk factors for head and neck cancer. It is worth noting that the analysis of dietary habits is difficult and has many potential confounders; this is reflected in the difference between the unadjusted hazard ratios and hazard ratios in the matched cohort in Table 3, and many prior studies have likely overestimated the effects of healthy eating or specific dietary modifications.5,27
Limitations
Limitations of this study included selection bias in the DHQ study population as reflected in the difference between the crude incidence rate of head and neck cancer in the BQ and DHQ study populations. This selection bias influenced our decision to perform 2:1 exact matching for analysis and prevented a meaningful assessment of head and neck cancer risk associated with drinking and dietary habits by subsite. In addition, we were not able to control for drinking habits in the BQ smoking analysis and had low sample sizes for the nasopharynx and hypopharynx cancer cases, precluding their inclusion in Figure 2. Other limitations included lack of HPV data for most of the study population; the lack of racial and ethnic diversity in the PLCO trial cohort, which reduces the overall generalizability of the study; and the inability to control for levels of physical activity in the DHQ analysis.
Conclusions
This cohort study demonstrated that the risk of head and neck cancer associated with smoking was higher for head and neck subsites that had closer proximity to the lungs. It also observed that heavy drinking, especially heavy consumption of beer and liquor, was associated with greater head and neck cancer risk, while overall healthy eating, including the consumption of whole fruits and whole grains, was associated with a modest reduction in head and neck cancer risk. These findings emphasize the importance of smoking cessation, drinking in moderation, and healthy eating for the prevention of head and neck cancer. They also demonstrate the need for further large-scale studies to evaluate the risk of head and neck cancer associated with smoking, drinking, HPV infection, and dietary habits, ideally with the goal of creating a robust risk assessment tool for head and neck cancer.
eTable 1. Baseline Questionnaire (BQ) Smoking Sensitivity Analysis, Squamous Cell Carcinoma (SCC) Only
eTable 2. Dietary Health Questionnaire (DHQ) Dietary & Drinking Sensitivity Analysis, Squamous Cell Carcinoma (SCC) Only
eFigure. Risk associated with Current and Prior Smoking by Subsite for Squamous Cell Carcinoma
Data Sharing Statement
References
- 1.Siegel RL, Miller KD, Wagle NS, Jemal A. Cancer statistics, 2023. CA Cancer J Clin. 2023;73(1):17-48. doi: 10.3322/caac.21763 [DOI] [PubMed] [Google Scholar]
- 2.Sung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209-249. doi: 10.3322/caac.21660 [DOI] [PubMed] [Google Scholar]
- 3.Rettig EM, D’Souza G. Epidemiology of head and neck cancer. Surg Oncol Clin N Am. 2015;24(3):379-396. doi: 10.1016/j.soc.2015.03.001 [DOI] [PubMed] [Google Scholar]
- 4.Agudo A, Bonet C, Travier N, et al. Impact of cigarette smoking on cancer risk in the European prospective investigation into cancer and nutrition study. J Clin Oncol. 2012;30(36):4550-4557. doi: 10.1200/JCO.2011.41.0183 [DOI] [PubMed] [Google Scholar]
- 5.Chuang SC, Jenab M, Heck JE, et al. Diet and the risk of head and neck cancer: a pooled analysis in the INHANCE consortium. Cancer Causes Control. 2012;23(1):69-88. doi: 10.1007/s10552-011-9857-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Di Credico G, Edefonti V, Polesel J, et al. Joint effects of intensity and duration of cigarette smoking on the risk of head and neck cancer: a bivariate spline model approach. Oral Oncol. 2019;94:47-57. doi: 10.1016/j.oraloncology.2019.05.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Di Credico G, Polesel J, Dal Maso L, et al. Alcohol drinking and head and neck cancer risk: the joint effect of intensity and duration. Br J Cancer. 2020;123(9):1456-1463. doi: 10.1038/s41416-020-01031-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kramer BS, Gohagan J, Prorok PC, Smart C. A National Cancer Institute sponsored screening trial for prostatic, lung, colorectal, and ovarian cancers. Cancer. 1993;71(2)(suppl):589-593. doi: 10.1002/cncr.2820710215 [DOI] [PubMed] [Google Scholar]
- 9.Gohagan JK, Prorok PC, Hayes RB, Kramer BS; Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial Project Team . The prostate, lung, colorectal and ovarian (PLCO) cancer screening trial of the National Cancer Institute: history, organization, and status. Control Clin Trials. 2000;21(6)(suppl):251S-272S. doi: 10.1016/S0197-2456(00)00097-0 [DOI] [PubMed] [Google Scholar]
- 10.National Cancer Institute Cancer Data Access System . Trial summary. Accessed January 2, 2024. https://cdas.cancer.gov/learn/plco/trial-summary/
- 11.National Cancer Institute Cancer Data Access System . Main findings. Accessed January 2, 2024. https://cdas.cancer.gov/learn/plco/main-findings/
- 12.Kreimer AR, Ferreiro-Iglesias A, Nygard M, et al. Timing of HPV16-E6 antibody seroconversion before OPSCC: findings from the HPVC3 consortium. Ann Oncol. 2019;30(8):1335-1343. doi: 10.1093/annonc/mdz138 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.National Center for Health Statistics . Glossary - alcohol. National Health Interview Survey. Accessed November 23, 2023. https://www.cdc.gov/nchs/nhis/alcohol/alcohol_glossary.htm
- 14.Harrell FE Jr. Regression modeling strategies. Biostatistics for Biomedical Research [course material]. Accessed January 9, 2024. https://hbiostat.org/doc/rms/book/
- 15.R: A Language and Environment for Statistical Computing. Accessed January 2, 2024. https://www.R-project.org/
- 16.Ho D, Imai K, King G, Stuart EA. MatchIt: nonparametric preprocessing for parametric causal inference. J Stat Softw. 2011;42(8):1-28. doi: 10.18637/jss.v042.i08 [DOI] [Google Scholar]
- 17.Wickham H, François R, Henry L, Müller K, Vaughan D. Dplyr: a grammar of data manipulation. Accessed January 2, 2024. https://dplyr.tidyverse.org/
- 18.Krebs-Smith SM, Pannucci TE, Subar AF, et al. Update of the Healthy Eating Index: HEI-2015. J Acad Nutr Diet. 2018;118(9):1591-1602. doi: 10.1016/j.jand.2018.05.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lee YA, Li S, Chen Y, et al. Tobacco smoking, alcohol drinking, betel quid chewing, and the risk of head and neck cancer in an East Asian population. Head Neck. 2019;41(1):92-102. doi: 10.1002/hed.25383 [DOI] [PubMed] [Google Scholar]
- 20.Khariwala SS, Carmella SG, Stepanov I, et al. Self-reported tobacco use does not correlate with carcinogen exposure in smokers with head and neck cancer. Laryngoscope. 2015;125(8):1844-1848. doi: 10.1002/lary.25290 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Bagnardi V, Rota M, Botteri E, et al. Alcohol consumption and site-specific cancer risk: a comprehensive dose-response meta-analysis. Br J Cancer. 2015;112(3):580-593. doi: 10.1038/bjc.2014.579 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Marziliano A, Teckie S, Diefenbach MA. Alcohol-related head and neck cancer: summary of the literature. Head Neck. 2020;42(4):732-738. doi: 10.1002/hed.26023 [DOI] [PubMed] [Google Scholar]
- 23.Boeing H, Dietrich T, Hoffmann K, et al. Intake of fruits and vegetables and risk of cancer of the upper aero-digestive tract: the prospective EPIC-study. Cancer Causes Control. 2006;17(7):957-969. doi: 10.1007/s10552-006-0036-4 [DOI] [PubMed] [Google Scholar]
- 24.Freedman ND, Park Y, Subar AF, et al. Fruit and vegetable intake and head and neck cancer risk in a large United States prospective cohort study. Int J Cancer. 2008;122(10):2330-2336. doi: 10.1002/ijc.23319 [DOI] [PubMed] [Google Scholar]
- 25.Steck SE, Murphy EA. Dietary patterns and cancer risk. Nat Rev Cancer. 2020;20(2):125-138. doi: 10.1038/s41568-019-0227-4 [DOI] [PubMed] [Google Scholar]
- 26.Bradshaw PT, Siega-Riz AM, Campbell M, Weissler MC, Funkhouser WK, Olshan AF. Associations between dietary patterns and head and neck cancer: the Carolina head and neck cancer epidemiology study. Am J Epidemiol. 2012;175(12):1225-1233. doi: 10.1093/aje/kwr468 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Grosso G, Bella F, Godos J, et al. Possible role of diet in cancer: systematic review and multiple meta-analyses of dietary patterns, lifestyle factors, and cancer risk. Nutr Rev. 2017;75(6):405-419. doi: 10.1093/nutrit/nux012 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Baseline Questionnaire (BQ) Smoking Sensitivity Analysis, Squamous Cell Carcinoma (SCC) Only
eTable 2. Dietary Health Questionnaire (DHQ) Dietary & Drinking Sensitivity Analysis, Squamous Cell Carcinoma (SCC) Only
eFigure. Risk associated with Current and Prior Smoking by Subsite for Squamous Cell Carcinoma
Data Sharing Statement