Outline of the study design. "Induction" refers to initial induction treatment. Patients were discontinued from the study for progression at any visit or stable disease only at the 6-month evaluation.
Top, A patient with lateral tongue leukoplakia. The arrowheads indicate the dense, white lesion. Bottom, Complete response at 3 months in the same patient. The lesion has completely cleared. The bright, white spots are unavoidable flash artifact.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Garewal HS, Katz RV, Meyskens F, et al. β-Carotene Produces Sustained Remissions in Patients With Oral Leukoplakia: Results of a Multicenter Prospective Trial. Arch Otolaryngol Head Neck Surg. 1999;125(12):1305–1310. doi:10.1001/archotol.125.12.1305
β-Carotene has been reported to produce regressions in patients with oral leukoplakia, a premalignant lesion. However, previous studies have all been of short duration, with clinical response as the end point.
To evaluate the duration of response and the need for maintenance therapy in subjects who respond to β-carotene.
In this multicenter, double-blind, placebo-controlled trial, subjects were given β-carotene, 60 mg/d, for 6 months. At 6 months, responders were randomized to continue β-carotene or placebo therapy for 12 additional months.
Fifty-four subjects were enrolled in the trial, with 50 being evaluable. At 6 months, 26 subjects (52%) had a clinical response. Twenty-three of the 26 responders completed the second, randomized phase. Only 2 (18%) of 11 in the β-carotene arm and 2 (17%) of 12 in the placebo arm relapsed. Baseline biopsies were performed in all patients, with dysplasia being present in 19 (38%) of the 50 evaluable patients. A second biopsy was obtained at 6 months in 23 subjects who consented to this procedure. There was improvement of at least 1 grade of dysplasia in 9 (39%), with no change in 14 (61%). Nutritional intake was assessed using food frequency questionnaires. There was no change in carotenoid intake during the trial. Responders had a lower intake of dietary fiber, fruits, folate, and vitamin E supplements than did nonresponders. β-Carotene levels were measured in plasma and oral cavity cells. Marked increases occurred during the 6-month induction. However, baseline levels were not restored in subjects taking placebo for 6 to 9 months after discontinuation of β-carotene therapy.
The activity of β-carotene in patients with oral leukoplakia was confirmed. The responses produced were durable for 1 year.
MOST HUMAN cancers are considered preventable, and epidemiological studies suggest a major role for diet. Macronutrients and micronutrients, and other potential substances, may be of importance. Chemoprevention is an approach whereby individual chemicals, either naturally occurring or synthetic, are studied for their effect on cancer prevention. There has been considerable recent interest in a role for naturally occurring compounds in the prevention of oral cavity cancer, in particular vitamin E, β-carotene, selenium, and vitamin A and its analogues.1-10
The reversal or suppression of premalignant lesions is an important therapeutic strategy for the prevention and control of cancer. Oral leukoplakia is an important premalignant lesion for oral cavity cancer that has been targeted in several trials. In trials of short durations, we and other researchers1,2,8-10 have shown that β-carotene can lead to clinical regressions in patients with oral leukoplakia without producing any major adverse effects.
Although remissions can be induced, it has been a frequent observation in previous, short-term, clinical intervention trials that lesions tend to recur soon after discontinuation of the intervention agent, often within 2 to 3 months. Most of these trials, however, have used short durations of intervention.
The present study had 2 major objectives: (1) to confirm the previously reported remission results with β-carotene therapy in patients with oral leukoplakia in a multicenter setting; and (2) to determine whether the responses would be maintained after discontinuation of β-carotene therapy in a placebo-controlled, 1-year follow-up period in which β-carotene continuation was randomized against placebo.
Adult patients with clinically measurable oral leukoplakia were eligible for enrollment in the trial. All patients had bidimensionally assessable lesions. The study protocol was approved by the applicable institutional review boards for human research, and written, informed consent was obtained from each participant.
The study design is outlined in Figure 1. After the initial biopsy, lesions were measured and photographed. Treatment began with β-carotene, 60 mg/d (30 mg twice daily) (Solatene; Hoffmann-LaRoche Inc, Nutley, NJ), and continued for 6 months. Patients were seen at 4- to 6-week intervals, with photography and documented assessment of response being performed every 3 months. At 6 months, responders were randomized to either continue β-carotene therapy, 60 mg/d, for another 12 months of treatment (total, 18 months) or receive an identical placebo capsule. Follow-up was continued as on the previous schedule. End points were either completion of 18 months of study or disease progression at any point in the trial. Three centers participated in the trial: University of Arizona, Tucson; University of Connecticut, Farmington; and University of California, Irvine.
Smoking and drinking habits were recorded at baseline and at 3-month intervals through the study. Compliance was monitored by a pill count at each 4- to 6-week visit. Serum samples were collected for measurement of β-carotene and vitamin E levels at baseline and 3-month intervals. Oral mucosal cell samples were also collected for level measurement, but this was limited to University of Arizona patients only. The methods used for these analyses have previously been described in detail.11
Dietary questionnaires were administered at baseline and at 12 and 18 months. The Arizona Food Frequency Questionnaire was used, which is a modification of the National Cancer Institute Health Habits and History Questionnaire food frequency component.12 The resulting instrument is a scannable food frequency questionnaire managed entirely by our own software but still basically using the portion size and food composition databases used by Block et al.12 The Arizona Food Frequency Questionnaire consists of a quantitative 113-item food frequency questionnaire that asks respondents to report how often they usually consumed each particular food during a specified period relative to times per day, week, or month and to identify their usual portion size of a food as small, medium, or large. Age-sex–specific portion size assumptions are used in the calculations. Output reports nutrient intake estimates including and excluding vitamin supplements as estimated by the questionnaire. The Arizona Food Frequency Questionnaires were reviewed by the nutrition staff (E.G.) for completeness and analyzed.13
Responses were defined in a standard manner. A complete response meant complete clinical disappearance of the lesion, as assessed visually, for at least 4 weeks. A partial response was defined as a 50% or greater reduction in the product of the 2 longest diameters. For patients with more than 1 lesion, target lesions suitable for assessment were identified at baseline. These were used to assess response as long as there was no progression anywhere else. Disease progression constituted an unequivocal increase in size of the lesion or appearance of new disease.
An attempt was made to repeat a biopsy at 6 months, especially in those patients with clinical responses. In those patients for whom the results of 2 biopsies, baseline and 6 months, were available, their histological characteristics were compared. A central pathological reading was obtained for each biopsy, which was evaluated by 2 experienced oral pathologists. In case of disagreement, both reviewed the histological characteristics together to arrive at a consensus. Criteria for dysplasia have been previously described.14
χ2 Tests and t tests were performed on several factors to determine what factors were related to response. The statistical significance of the nutritional information was tested using a 2-tailed test when there was no a priori hypothesis (eg, total energy intake bears no effect on responder status) and a 1-tailed test when there was an a priori hypothesis (eg, responders have poorer nutrient intake).
A total of 54 patients were enrolled in the study, with 50 being evaluable. Reasons for not being evaluable included poor compliance with failure to return for any follow-up visit (2 patients), lesion not consistent with leukoplakia on histological review (1 patient), and development of intercurrent illness (bleeding duodenal ulcer) (1 patient). Compliance assessed by pill count in the evaluable patients exceeded 80%.
Patient characteristics, risk factors, and histological features of the oral leukoplakia are shown in Table 1. Dysplasia was present in 19 patients, being mild to moderate in 16 and moderate to severe in 3. No patient discontinued tobacco and/or alcohol use during the study.
Response data are as follow. Overall, 26 of the 50 evaluable patients had a complete (2 [4%] of the patients) or partial (24 [48%] of the patients) clinical response at 6 months, giving a response rate of 52% (95% confidence interval, 38%-66%). Most responders (23 [88%]) showed a clinical response at 3 months, with only 3 patients (12%) responding between 3 and 6 months. Twenty (40%) of the 50 patients had stable disease. Four (8%) of the 50 patients showed disease progression, with oral squamous cell cancer being diagnosed in 1 shortly after conclusion of the induction phase.
Similar response rates were noted at the University of Arizona (9 [60%] of 15 patients) and the University of California (15 [60%] of 25 patients). Although not statistically significant (P = .08), the response rate was lower at the University of Connecticut (2 [20%] of 10 patients). The Connecticut study site joined the trial later, and it has been our clinical experience that the first few patients enrolling in such studies tend to have more clinically durable, longer-duration, less-responsive lesions than subsequent entrants. This is perhaps related to such patients being on the alert for new modalities for their very refractory lesions.
As in our previous study,1 the first change often noted was a thinning of a leukoplakic lesion with a decrease in the thickness and appearance of a more pink epithelium. This would subsequently become a decrease in the size of the lesion if the response continued. Figure 2 illustrates a complete response of a lateral tongue leukoplakic lesion observed at 3 months into the study.
A second biopsy was obtained after 6 months of therapy in 23 patients. There was no change in the degree of dysplasia in 14, with improvement of at least 1 grade in 9 (39% response rate).
Twenty-six subjects (responders) were eligible for randomization at 6 months. Three refused to be randomized, opting to continue β-carotene treatment in view of their response. Therefore, 23 were randomized, 11 to the β-carotene arm and 12 to the placebo arm. Nineteen (83%) of the 23 subjects completed 18 months of treatment without progression. Four patients (17%) progressed, 2 (18%) in the β-carotene arm and 2 (17%) in the placebo arm. All 4 cases of progression were noted at 3 to 4 months after randomization.
Baseline and 3- and 6-months levels of β-carotene and vitamin E in plasma and buccal cells are shown in Table 2. As expected, there was a significant increase in β-carotene levels with supplementation. There was no statistically significant effect on vitamin E levels, either in plasma or in the buccal cells. There was also no statistically significant difference in plasma β-carotene or vitamin E levels between the 3 treatment sites. There were no correlations between baseline levels in either plasma or buccal cells vs response status. Similarly, there was no correlation between subsequent, postsupplement levels and response. Finally, analyzing levels in plasma and buccal cells in the patients receiving placebo after discontinuation of active treatment at 6 months revealed that plasma β-carotene levels returned to baseline at 15 months, ie, 9 months after discontinuation, while buccal levels returned to baseline at 12 months, ie, 6 months after discontinuation.
Table 3 shows the nutrient and food group intakes of nonresponders and responders at baseline. There were significant differences in dietary patterns between the 2 groups. Energy intake was similar. The responders reported higher alcohol intake than the nonresponders. For micronutrients, the nonresponders showed a trend toward better micronutrient intake than the responders, but the differences are significant only for folate (borderline for dietary fiber). Vitamin E supplementation use approached statistical significance (P = .06). Follow-up food frequency questionnaires completed at 1 year and later showed no significant differences between intakes of the carotenoids at baseline and later. In fact, the average change for each carotenoid, although nonsignificant, was negative, meaning that, if anything, they decreased their intake during the year. Thus, the continued maintenance of high plasma levels after placebo changeover cannot be accounted for by dietary alterations.
Several studies1-10 have shown the ability of chemopreventive agents to clinically produce improvement in patients with oral leukoplakia. Responses have been noted with vitamin A analogues, such as 13-cis-retinoic acid, but the toxicity of such agents precludes their clinical use for disease prevention.3,4 Amongst those with little or no toxicity, attention has focused on those that are easily amenable to supplementation. In this context, clinical trials have been done with vitamin A supplementation, especially in Asia, which have resulted in some of the highest reported response rates.5,6 However, the population in Asia is likely to be quite different nutritionally, and this compound needs to be studied in a western, well-nourished population group. Other agents that have shown activity include β-carotene, vitamin E, and selenium.1-10
The present trial extends our previous observation of a significant response rate of patients with oral leukoplakia to β-carotene supplementation,1 with the primary goal being to determine the durability of response after an intensive initial treatment phase. The trial design involved initial treatment of all eligible patients with β-carotene followed by randomization of responders only. In this context, the study differs notably from a recent study15 using 13-cis-retinoic acid for induction followed by randomization to β-carotene or 13-cis-retinoic acid. This trial found that β-carotene was less effective in maintaining 13-cis-retinoic acid–induced responses than 13-cis-retinoic acid itself. In that study, responders and stable patients were eligible to continue, thereby permitting much larger numbers of initial subjects to continue into the maintenance phase. Furthermore, all subjects received 13-cis-retinoic acid up to the point of randomization and were then switched to β-carotene or continued to receive 13-cis-retinoic acid. Thus, patients in the β-carotene arm would have a decline in 13-cis-retinoic acid tissue levels soon after randomization, while β-carotene levels would increase gradually. Conversely, 13-cis-retinoic acid levels would remain elevated in those randomized to continue treatment with it. In the present trial, all patients would have elevated β-carotene levels at randomization, with no switch to another agent.
Our results confirm the ability of β-carotene to produce responses in patients with oral leukoplakia. A reasonable criticism of this study is the absence of a blinded, placebo-controlled arm in the initial, induction phase of the trial, especially since leukoplakia is a lesion that can remit spontaneously. However, in the planning process, this was given serious consideration and not included for 2 main reasons. First, any attempt at blinding would be purely delusional since yellowing of the skin, and perhaps other bodily material such as feces, at this dose of β-carotene would result in easy recognition of the subjects receiving the active agent. Second, the sample size requirements would become prohibitive. The main goal of the trial was assessment of the need for prolonged maintenance therapy and, therefore, the final design used was selected. It should be emphasized that the spontaneous remission rate of leukoplakic lesions in other published trials6 of similar duration has been less than 10%.
Interestingly, the responses produced were highly durable, an observation somewhat different from previous clinical experience. In the past, based on studies using lower doses in the induction phase, shorter induction phases, or both, the lesions tend to recur in most patients within 2 to 3 months of discontinuation of the agent being tested. One possible explanation of the durability of responses in our study relates to the pharmacological features of clearance of β-carotene after the intensive treatment phase. High concentrations of β-carotene were achieved, probably resulting in markedly elevated stores that did not reach baseline levels for up to 6 to 9 months after randomization to placebo. From a practical standpoint, this is a favorable observation in that use of this agent for chemoprevention would not be affected by occasional missed doses as these would have little or no effect on levels after a steady state is reached, presumably in several weeks.
In this trial, we attempted to use photography and clinical measurements to assess response. However, as we have discussed previously, assessment of response is more problematic in lesions such as oral leukoplakia, which are visible to the naked eye, than in the standard oncologic radiographic lesion.16 Consequently, we required 2 independent observers to agree before assigning a response category and used photographs primarily for guidance rather than as a means for measurement of lesion size. We have also discussed the issue of histological response in the past, pointing out that this category is extremely problematic not only in oral leukoplakia but in other similar lesions, such as Barrett esophagus, where site-to-site variability within a lesion, variability between lesions, and the impact of a previous biopsy on histological characteristics make a simple pretreatment and posttreatment assessment of unclear significance.16 Furthermore, although participants were initially agreeable, numerous patients declined a second biopsy at the 6-month point.
Another intermediate end point biomarker in oral cancer chemoprevention trials has been micronucleus frequency in exfoliated oral cells.17 We have reported our experience with this assay previously.18 In our experience, micronucleus frequency at baseline was simply too low to allow its measurement in a meaningful manner. Thus, although initially planned, it was abandoned as a useful potential biomarker.
Analysis of nutritional intake confirmed that there was no significant change in carotenoid intake during the trial. Although the detailed analyses must be considered preliminary, responders had a lower intake of dietary fiber, fruits, folate, and vitamin E supplements than nonresponders. Conversely, they consumed more red meat and alcohol. One possible explanation for these unexpected findings is that micronutrient supplementation is most effective in those with "poor" nutritional habits.
No significant change was noted in vitamin E levels, neither in serum nor in buccal cells, as a result of β-carotene supplementation. These levels were measured because of the possibility of an effect on vitamin E absorption produced by β-carotene. For example, one study19 had reported lowering of vitamin E blood levels by β-carotene. However, subsequent findings of several trials,20-23 including randomized studies, have largely failed to confirm this finding. The latter is consistent with our results. In addition to serum levels, to our knowledge, our study is the first to demonstrate a lack of effect of oral β-carotene on buccal cell vitamin E levels also.
In summary, the results of chemopreventive intervention studies with nutritional agents in patients with oral leukoplakia continue to be encouraging. However, the selection of specific agents and durations of intervention in future trials need to be based on the population being targeted. For example, the results of recent studies24,25 with β-carotene in smokers, targeting lung cancer, have been negative and have even suggested the possibility of harm. Consequently, especially in western populations, studies with vitamin E, selenium, or both may be more rewarding in that these agents have shown benefit in other western diseases, such as cardiovascular disease26 and cancer.27 However, in Asian countries, where oral cavity cancer is endemic, vitamin A and carotenoids need to be pursued further. Finally, a strategy needs to be devised to integrate these findings in patients with oral leukoplakia into the actual prevention, either primary or secondary, of oral cavity cancer using chemoprevention.
Accepted for publication August 13, 1999.
Supported in part by grant CA-P01-27502 from the National Institutes of Health, Bethesda, Md.
Reprints: Harinder S. Garewal, MD, PhD, Section of Hematology-Oncology (1-111D), Southern Arizona Veterans' Affairs Healthcare System, 3601 S Sixth Ave, Tucson, AZ 85723 (e-mail: email@example.com).