Interaction between antioxidant vitamin supplementation and cigarette smoking during radiation therapy in relation to long-term effects on recurrence and mortality: A randomized trial among head and neck cancer patients
There has been concern that the efficacy of radiation therapy may be reduced when patients smoke or take antioxidant vitamins during treatment. Cancer prevention trials with beta carotene supplements documented adverse effects only among smokers. We conducted a randomized trial with alpha tocopherol (400 IU/day) and beta carotene (30 mg/day) supplements among 540 head and neck cancer (HNC) patients treated by radiation therapy. We examined whether smoking during radiation therapy modified the effects of the supplementation on HNC recurrence and on mortality. During the follow-up, 119 patients had a HNC recurrence and 179 died. Cox models were used to test the interaction between smoking and supplementation and to estimate the hazard ratios (HR) for HNC recurrence and death associated with the supplementation. Cigarette smoking either before or after radiation therapy did not modify the effects of the supplementation. In contrast, the interactions between supplementation and cigarette smoking during radiation therapy were statistically significant for HNC recurrence (p = 0.03), all-cause mortality (p = 0.02) and mortality from the initial HNC (p = 0.04). Among cigarette smokers, the HR were 2.41 (95% CI: 1.25–4.64) for recurrence, 2.26 (95% CI: 1.29–3.97) for all-cause mortality and 3.38 (95% CI: 1.11–10.34) for HNC mortality. All corresponding HR among nonsmokers were close to 1. These results could best be explained by the hypothesis that the combined exposures reduced the efficacy of radiation therapy. Particular attention should be devoted to prevent patients from both smoking and taking antioxidant supplements during radiation therapy. © 2007 Wiley-Liss, Inc.
Many patients take antioxidant supplements during cancer therapy to reduce adverse effects of treatments and to improve long-term outcomes.1, 2 Currently, the use of antioxidant supplements during radiation therapy is a controversial issue, since it is feared that they could interfere with treatment.3, 4 We conducted a double-blind, placebo-controlled, randomized trial among head and neck cancer patients treated by radiation therapy to assess the effects of a supplementation combining alpha tocopherol and beta carotene on outcomes of head and neck cancer patients. Contrary to expectation, the supplementation had adverse effects on the recurrence of the initial cancer, on the incidence of second primary cancers and on mortality.5–7
There is strong evidence from chemoprevention trials that beta carotene supplementation increases lung cancer incidence only among smokers.8–11 In head and neck cancer patients treated by radiation therapy, smoking during treatment was associated with a lower rate of complete response and a poorer 2-year survival.12 Therefore, in light of the results of these latter studies, we examined in our trial population whether the effects of the supplementation on head and neck cancer recurrence and on mortality were influenced by smoking during radiation therapy (i.e., whether smoking during radiation therapy modified the associations between the supplementation and the trial outcomes).
Study population and intervention
We conducted a multicenter, double-blind, placebo-controlled, randomized chemoprevention trial among 540 patients treated by radiation therapy for stage I or II head and neck cancer (HNC). The primary objective of the trial was to evaluate the efficacy of the supplementation in reducing the incidence of second primary cancers (SPC). The detailed description of the trial methods and the results concerning the main effects of the supplementation on SPC and cancer-free survival, on acute adverse effects of radiation therapy and recurrence of the initial cancer and on mortality have been presented.5–7 The institutional review board of each participating center approved the study protocol. All patients gave written informed consent prior to randomization. Eligible patients were aged 18 years and over, had received a first diagnosis of stage I or II, were histologically documented for squamous cell carcinoma of the head and neck and were scheduled to be treated by radiation therapy between October 1, 1994 and June 6, 2000 in 1 of 5 radiation therapy centers in the province of Quebec, Canada. A computer-generated randomization list was prepared in advance for each collaborating center using an allocation ratio of 1:1. Patients were randomly assigned to receive a daily supplementation combining alpha tocopherol (1 capsule of 400 IU dl-alpha-tocopherol) and beta carotene (1 capsule of 30 mg) or corresponding placebos during radiation therapy and for 3 years thereafter. All capsules, supplements and placebos, were supplied by Roche Vitamins (Parsippany, NJ). Both alpha tocopherol and beta carotene were of synthetic origin. The intervention began on the first day of radiation treatment since a beneficial effect of antioxidant supplementation during radiotherapy was anticipated.13 The supplementation with beta carotene was discontinued after 156 patients had been enrolled into the trial, because of ethical concerns following the release of the beta-carotene and retinol efficacy trial results in January 1996.9 The trial was continued with alpha-tocopherol alone. Throughout the trial, the patients, the treating physicians, the study personnel and the investigators were kept blind to the patients' intervention arm assignment.
Data collection and follow-up
Baseline data collection was completed before randomization. Follow-up information was obtained by the radiation oncologists and the study nurses at predetermined times; immediately and 1 month after radiation therapy, every 6 months during the 3 years following the end of radiation therapy and then once a year. A detailed questionnaire on past and present tobacco exposure was administered by the study nurse prior to randomization with separate questions on cigarette, cigar and pipe smoking as well as on quid. The same information was obtained at each visit for the time period elapsed since the preceding visit. For the present study, smoking status and number of cigarettes smoked were documented for 3 time periods: (i) for the year prior randomization and the beginning of radiation therapy, (ii) during radiation therapy (iii) for the year following the end of radiation therapy. In each period, participants were considered as smokers if they had reported to have smoked 1 or more cigarette, cigar, or pipe during the period. Only 30 of the 540 participants reported to have never smoked. None of the 540 patients smoked the pipe or used quid during the year before randomization and afterward. Three participants smoked both cigars and cigarettes during the year prior to randomization but stopped smoking cigars after randomization. One participant smoked only cigars before randomization and continued to do so during the follow-up. This participant was randomized to the supplement arm of the trial. In the analyses, he was assigned to the non smoking category for cigarette smoking. He died from a SPC 3 years after entering into the trial. Since, with the exception of this participant, smoking after randomization consisted exclusively of cigarette smoking, the analyses focused on cigarette smoking.
During each visit, the radiation oncologists assessed the recurrence of the initial tumor. Medical notes and hospital records were requested for all important health events and hospitalizations during follow-up. Copies of all relevant pathology reports and of all death certificates were obtained. For recurrence of the initial cancer, the follow-up ended on June 30, 2003, when the last patient enrolled had completed the supplementation period. In most instances, death was identified by the study staff in the collaborating hospitals. Furthermore, to ensure complete ascertainment, record linkage with the Quebec mortality files was performed using the unique Quebec health insurance identifier from January 1, 1994 until December 31, 2004. The cause of death was taken as indicated on the death certificate except when evidence from medical and hospital data clearly indicated another cause. In all cases, the cause of death was assigned without knowledge of the patient intervention arm.
The analyses were designed to address the following question: did the effects of the antioxidant vitamin supplementation on HNC recurrence and on mortality differ according to smoking status at 3 periods: (i) during the year prior to radiation therapy, (ii) during radiation therapy and (iii) during the year following the end of radiation therapy? Data sets used for these analyses differed in the number of eligible subjects depending on the smoking exposure and the outcome considered.
We always ensured that each smoking exposure period preceded outcome occurrence. When we examined the effects of smoking in the year prior to radiation therapy, the follow-up time started at the time of randomization and there was no exclusion. When we examined the effects of smoking during radiation therapy, the follow-up time started on the day following the end of radiation therapy, and 6 patients (1 in the supplement arm and 5 in the placebo arm) without information on smoking during radiation therapy were excluded. When we examined the effects of smoking during the year following the end of radiation therapy, the follow-up time started 1 year after the end of radiation therapy, and 30 patients (16 in the supplement arm and 14 in the placebo arm) without information on smoking during the year following the end of radiation therapy were excluded. In addition, for this latter analysis, 65 patients who had a cancer recurrence during the year following the end of radiation therapy were excluded when assessing recurrence, and 29 who died during the year following the end of radiation therapy were excluded when assessing mortality. For cancer recurrence, the follow-up time was counted until the date of the first of the following events: HNC recurrence, death or last visit. For mortality outcomes (death from any cause, death from the initial HNC), the follow-up time was counted until the date of last visit (for 10 participants who had not consented to record linkage), the date of death or December 31, 2004. The information presented above is summarized in Table I which shows separately for the 3 smoking periods considered the population studied, the follow-up period, the exclusions and the number of events, either recurrence or death.
Table I. Population Studied, Follow-Up Period, Exclusions and Number of HNC Recurrence and Death According to the Cigarette Smoking Exposure Considered
|Smoking exposure and follow-up period|
| Exposure||Cigarette smoking during the year prior to radiation therapy||Cigarette smoking during radiation therapy||Cigarette smoking during the year following the end of radiation therapy|
| Beginning of follow-up||On the first day of radiation therapy||At the end of radiation therapy||1 year after the end of radiation therapy|
| End of follow-up||June 30, 2003 for recurrence, December 31, 2004 for mortality|
| Exclusion for lack of data on smoking exposure (w)||n = 0||n = 6||n = 36|
|Outcome: HNC recurrence|
| Exclusion for earlier recurrence (x)||n = 0||n = 0||n = 65|
| Exclusion for earlier death without HNC recurrence (y)||n = 0||n = 0||n = 4|
| Patients eligible for follow-up [540 − (w + x + y)]||N = 540||N = 534||N = 435|
| Patients with HNC recurrence||n = 119||n = 119||n = 50|
| Exclusion for earlier death (z)||n = 0||n = 0||n = 29|
| Patients eligible for follow-up [540 − (w + z)]||N = 540||N = 534||N = 475|
| Death of any cause||n = 179||n = 175||n = 128|
| Death from first cancer||n = 54||n = 53||n = 28|
Cox proportional hazards models were used to compare the rates of recurrence of the initial HNC and the mortality rates in the 2 treatment arms of the trial.14 The Cox models were used to test the interaction between smoking and supplementation by including a cross-product term in the model. The analyses were also performed separately among smokers and among non smokers for each of the 3 successive exposure periods. The Cox models were used to adjust for covariates. In particular, we examined the potential confounding effect of radiation therapy doses, interruption of radiation treatment and alcohol, but none of these variables were confounders. For each hazard ratio (HR), the associated 95% confidence interval (CI) was calculated. The proportionality assumption of the models was assessed visually, by checking the parallelism of the log cumulative hazard function plotted against the log of follow-up time, and tested by the weighted Schoenfeld residual score test.14, 15 Since the 156 first patients enrolled in the trial had received beta carotene in addition to alpha tocopherol or their placebos, the analyses were also conducted separately for the first 156 participants and for the 384 patients subsequently enrolled. All analyses were performed using SAS, version 9.1 (SAS Institute, Cary, NC). All statistical tests were 2-sided.
The baseline characteristics of the 273 and 267 patients randomized to the supplement and placebo arms were well balanced (Table II). The participants' follow-up is summarized in Table I. The median duration of radiation therapy was 43 days and the median duration of the supplementation was 37 months. During the supplementation period, 36 participants in the supplement arm and 34 in the placebo arm stopped taking the capsules. Compliance, as assessed by capsule count, was identical in the 2 arms: 90% during radiation therapy and 87% during the first year after the end of radiation therapy.
Table II. Baseline Characteristics of Trial Participants by Treatment Arm
|Total number of patients randomized||273|
|Patients randomized to α-tocopherol and β-carotene or placebos||79|
|Patients randomized to α-tocopherol alone or placebo||194|
|Age (years), mean (SD)||63 (10)|
|Male sex, no. (%)||223 (82)|
|Cigarette smokers in preceding year, no. (%)||178 (65)|
|Laryngeal cancer, no. (%)||225 (82)|
|Stage II, no. (%)||101 (37)|
|Total dose of external beam radiation therapy (grays), mean (SD)||61 (7)|
|Number of radiation therapy fractions, mean (SD)||29 (6)|
|Plasma α-tocopherol (μmol/L), mean (SD)||33 (10)|
|Plasma β-carotene (μmol/L), mean (SD)||0.23 (0.18)|
During the year preceding randomization, 343 patients were cigarette smokers (63%). The percentage of cigarette smokers decreased to 30% during radiation therapy and remained at this level (33%) during the year following the end of radiation therapy. The median number of cigarettes smoked daily as reported by smokers was, respectively, 10 in the year preceding randomization, 5 during radiation therapy and 10 in the year following the end of radiation therapy. On the basis of the data in the 3 periods, it was possible to identify different profiles of smoking. More than 90% of the trial participants followed 1 of 3 cigarette smoking profiles over the 3 time periods; 196 participants were non smokers throughout, 150 were consistently smokers and 141 patients had smoked in the year preceding radiation therapy but did not smoke during and after radiation therapy.
During the follow-up, 119 patients had a recurrence of the initial head and neck cancer and 179 died. The median duration of follow-up was 51 months for recurrence and 78 months for mortality. Table III presents the number of eligible patients, the number of events for each study outcome (recurrence, death from any cause, death form HNC) by cigarette smoking status and by randomization arm. There was no statistically significant interaction (p-values for interaction ranging from 0.18 to 0.68) between supplementation and cigarette smoking in the year prior to radiation therapy (Table III, section A) nor in the year following the end of radiation therapy (Table III, section C) for any outcome. In contrast, smoking cigarettes during radiation therapy modified the effect of the supplementation (Table III, section B). This was observed for all outcomes: recurrence of the initial head and neck cancer (p value for interaction =0.03), all-cause mortality (p = 0.02) and mortality from the initial cancer (p = 0.04). In fact for these outcomes, the excess incidence associated with the supplementation was only observed among cigarette smokers: HR = 2.41 (95% CI: 1.25–4.64) for recurrence, HR = 2.26 (95% CI: 1.29–3.97) for all-cause mortality and HR = 3.38 (95% CI: 1.11–10.34) for deaths due to the initial HNC. All corresponding HR among nonsmokers were close to 1.
Table III. Crude and Adjusted Hazard Ratios (HR) for HNC Recurrence and Death with 95% Confidence Intervals (CI) Associated with the Supplementation by Cigarette Smoking Status and p-Value for Interaction Between Cigarette Smoking and Supplementation for Three Consecutive Periods: (A) During the Year Prior to Radiation Therapy, (B) During Radiation Therapy, and (C) During the Year Following the End of Radiation Therapy
|A: Cigarette smoking during the year prior to radiation therapy|
| HNC recurrence||Smokers||52/178||34/165||1.49 (0.97–2.30)||1.56 (1.01–2.40)||0.54|
|Non smokers||17/95||16/102||1.16 (0.58–2.29)||1.08 (0.54–2.15)|
| Death of any cause||Smokers||74/178||49/165||1.53 (1.06–2.19)||1.47 (1.02–2.12)||0.25|
|Non smokers||28/95||28/102||1.05 (0.62–1.77)||1.12 (0.66–1.89)|
| Death from first cancer||Smokers||25/178||17/165||1.42 (0.77–2.63)||1.47 (0.79–2.74)||0.68|
|Non smokers||6/95||6/102||1.07 (0.34–3.32)||0.98 (0.30–3.17)|
| B: Cigarette smoking status during radiation therapy|
| HNC recurrence||Smokers||30/85||13/79||2.45 (1.28–4.69)||2.41 (1.25–4.64)||0.03|
|Non smokers||39/187||37/183||1.03 (0.65–1.61)||1.07 (0.68–1.68)|
| Death of any cause||Smokers||39/85||18/79||2.38 (1.36–4.50)||2.26 (1.29–3.97)||0.02|
|Non smokers||62/187||56/183||1.09 (0.76–1.56)||1.14 (0.79–1.64)|
| Death from first cancer||Smokers||14/85||4/79||3.60 (1.18–10.95)||3.38 (1.11–10.34)||0.04|
|Non smokers||17/187||18/183||0.93 (0.48–1.81)||1.06 (0.54–2.07)|
| C: Cigarette smoking status during the year following the end of radiation therapy|
| HNC recurrence||Smokers||13/67||10/82||1.63 (0.71–3.72)||1.53 (0.67–3.53)||0.53|
|Non smokers||15/149||12/137||1.14 (0.53–2.44)||1.21 (0.56–2.59)|
| Death of any cause||Smokers||27/73||19/83||1.76 (0.98–3.19)||1.70 (0.93–3.08)||0.22|
|Non smokers||45/166||37/153||1.13 (0.73–1.75)||1.24 (0.80–1.93)|
| Death from first cancer||Smokers||7/73||3/83||2.89 (0.74–11.19)||2.99 (0.77–11.65)||0.18|
|Non smokers||9/166||9/153||0.94 (0.37–2.37)||1.20 (0.47–3.08)|
Similar patterns were observed among the first 156 participants who, for some time at the beginning of the trial, had received both alpha tocopherol and beta carotene supplements, or their placebos, as well as for the 384 subsequent participants who received only alpha tocopherol or its placebo throughout the supplementation period. For example, the adjusted HR for all-cause mortality was 2.12 (95% CI: 1.04–4.30) among smokers and 1.16 (95% CI: 0.73–1.85) among non smokers for participants exposed to alpha tocopherol supplement or its placebo only. The corresponding figures for participants exposed to both alpha-tocopherol and beta carotene supplements, or their placebos, were 2.34 (95% CI: 0.91–6.06) among cigarette smokers and 1.06 (95% CI: 0.58–1.95) among non smokers.
Our results suggest that the adverse effects of the antioxidant vitamin supplementation in the trial on HNC recurrence, on mortality due to HNC and on mortality from any cause were mainly attributable to patients who smoked cigarettes during radiation therapy.
Our trial has several strengths, including good compliance with the intervention, complete follow-up and prospective measurement of smoking with standardized questionnaires during the follow-up. A major protocol change occurred early in the trial with the cessation of the beta carotene supplementation. This could be seen as a limitation. However, our results indicate that supplementation with alpha tocopherol alone and supplementation combining alpha tocopherol and beta carotene similarly interact with smoking cigarettes during radiation therapy.
Our randomized trial was designed to assess the efficacy of a supplementation with antioxidant vitamins on the occurrence of second primary cancers and other long-term outcomes in patients with HNC. Since smoking is a well recognized determinant of both HNC and the most frequent types of SPC, we carefully documented participants' smoking during the trial. We relied on patients' response to a structured smoking questionnaire administered by the study nurse during a confidential interview. Since all patients had been encouraged to discontinue smoking after HNC diagnosis and during radiation therapy, they may have hidden or minimized their smoking habits. There is, however, evidence from validation studies with biological indicators that self-report of smoking is accurate both by healthy adults16 and by patients treated for head and neck cancer.17–19 Furthermore, in our trial, smoking assessment was always conducted prospectively before the occurrence of any outcome and blindly with respect to the intervention arm assignment. It is, therefore, unlikely that errors affecting smoking assessment might have seriously affected the results.
There is evidence that patients with head and neck cancer who continue to smoke after diagnosis have poorer outcomes than those who do not smoke during radiation treatment and afterward. Patients who continue to smoke have been reported to be at increased risk for treatment-related adverse effects,20 for HNC recurrence,21 for SPC incidence,17, 22 and for mortality.21, 22 Few studies have focussed specifically on the effect of smoking during the period of active treatment. Browman et al. observed that HNC patients who continued to smoke during radiation therapy had a lower rate of complete response and a poorer 2-year survival than those who did not smoke during treatment.12 A subsequent investigation, however, comparing recent quitters to persistent smokers during radiation therapy did not find a difference in the rate of complete response nor in overall survival.18 A major difficulty in assessing the effect of smoking specifically during radiation therapy is to take into account the effect of smoking during the preceding and subsequent time periods.23 Although it seems reasonable to expect no benefit from smoking during therapy for head and neck cancer, no study to date has documented convincingly that smoking during radiation therapy affects disease long-term outcomes beyond what could be ascribed to smoking in other time periods.
An interaction between antioxidant supplementation and smoking could have been anticipated given the results of the beta carotene cancer chemoprevention trials.8–11 No significant adverse effects were associated with beta carotene supplementation in healthy predominantly non smoking men10 and women.11 On the other hand, lung cancer incidence and mortality were increased by beta-carotene supplementation in smokers in the ATBC trial8 and by a supplementation combining beta carotene and retinol among heavy former or current smokers and asbestos-exposed participants in the CARET trial.9 In the ATBC and CARET trials, participants' smoking habits were relatively stable before the enrolment and during the follow-up period. In contrast, in our study, a large proportion of participants were former smokers who had quit shortly before the beginning of the supplementation. This could explain why the adverse effects of the supplementation observed in our trial were not modified by smoking status in the year prior to enrolment.
Our data suggest that the efficacy of radiation therapy was hampered by the combined exposure to high levels of alpha tocopherol and to cigarette smoke during the treatment. Antioxidants, such as alpha tocopherol, are known for their ability to scavenge free radicals, potentially reducing the damage caused by ionizing radiations.24 The radioprotective effect of vitamin E has been documented but varies according to forms of vitamin E, doses and timing.25 Alpha tocopherol reduced the radiosentivity of erythrocyte membranes when given during radiation but not thereafter.26 There is also indirect evidence that smoking during radiation therapy could reduce its efficacy. Smoking increases blood carboxyhemoglobin and decreases the oxygen unloading capacity, thus producing tissue hypoxia.27 This impairs the oxygen-dependent effects of radiation therapy.24 In an experimental model, increased levels of carboxyhemoglobin reduced the effectiveness of radiation therapy by increasing tumor hypoxia.28
Cancer patients and their treating physicians are faced with the difficult challenge of deciding whether or not to use antioxidant supplements during radiation therapy, since the relevant scientific evidence is scarce. In most randomized trials of antioxidant supplementation among head and neck cancer patients, the supplementation was started only after successful completion of conventional therapies.29–31 Therefore, these studies cannot provide information on the potential interference of supplementation with treatment efficacy. Proponents of antioxidant supplementation support their hypothesis by theoretical considerations and experimental data on cell lines or animal models. As illustrated in a recent discussion,32, 33 “new considerations, including unpredicted interactions, arise when science move from the test tube to humans.” To our knowledge, our study is the only large-size randomized trial, which directly addresses this question. It is possible that vitamins of natural rather than synthetic origin, that different dosages, or different combinations of antioxidants would not yield results similar to those observed in our trial. There is, however, to date no human data to support such claims.
The adverse long-term effects observed in our trial among patients randomized to the supplementation arm, who smoked during radiation therapy could best be explained by the hypothesis that this combined exposure reduced the efficacy of radiation therapy. Most patients undergoing radiation therapy for head and neck cancer are encouraged to stop smoking and are often warned against the potential adverse effects of antioxidant supplementation. Particular attention should be devoted to prevent patients from both smoking and taking antioxidant supplements during radiation therapy.
Dr. Bairati was the recipient of a senior scientist award from the Fonds de Recherche en Santé du Québec. The investigators are grateful to the study participants, members of the Ethical and Safety Monitoring Committee and research staff.