Low-Dose Isotretinoin Versus  -Carotene to Prevent Oral Carcinogenesis: Long-term Follow-up

In 1993, we reported the results of the first randomized maintenance chemoprevention trial in oral premalignancy ( 1 ). In that study, 70 patients with advanced oral premalignant lesions underwent 3 months of induction therapy with high-dose isotretinoin (1.5 mg/kg per day), followed by 9 months of maintenance therapy with β-carotene or lowdose isotretinoin in 59 patients with responding or stable oral premalignant lesions. Thirty-three of 59 patients received β-carotene (30 mg/day) and 26 of 59 patients received low-dose isotretinoin (0.5 mg/kg per day). Lowdose isotretinoin was well tolerated and more effective than β-carotene: the 12- month oral premalignant lesion progression rate was significantly lower with isotretinoin than with β-carotene (8% versus 55%) ( 1 ). Our prior 3-month trial ( 2 ) proved that high-dose isotretinoin was effective in reversing oral premalignant lesions, but with substantial toxicity and high relapse rate upon discontinuation, providing the rationale for the maintenance study. With a median follow-up of 66 months, we now report long-term head and neck cancer rates from this maintenance oral premalignancy study ( 1 ). Overall, in situ or invasive carcinoma developed in 24% of the patients (17 of 70), an annual rate of 5.8%. Invasive squamous cell carcinoma developed in two of 11 patients who received induction therapy only. Among the 59 maintenance patients, six (23%) of 26 patients on the low-dose isotretinoin arm and nine (27%) of 33 on the β-carotene arm developed in situ or invasive carcinoma. There was no significant difference in the overall time to cancer development between the two maintenance groups ( P = .75; logrank test) ( Fig. 1 ). Median cancer-free survival time was not reached in either group. Among patients who developed cancer, the median time to cancer development was 37 months in the isotretinoin arm versus 29 months in the β-carotene arm. Currently, 85% (28 of 33) and 92% (24 of 26) of the patients are alive in the β-carotene and isotretinoin maintenance groups, respectively. Both patients who developed carcinoma during maintenance therapy were receiving β-carotene. The similar longer-term rates of cancer development in the two groups indicate that the retinoid becomes less effective over time after treatment. Multiple primary squamous cell carcinomas developed in four patients treated with β-carotene and in one patient treated with low-dose isotretinoin ( Table 1 ). In all, there were 20 invasive squamous cell carcinomas, mostly stage I and II: six cancers developed in four isotretinoin-treated patients and 14 cancers in seven β-carotene-treated patients. In situ carcinoma developed in two patients in each arm. The difference in oral premalignant lesion progression rates (during maintenance) for patients who developed cancer and for those who did not was statistically significant: oral premalignant lesion progression in 10 (67%) of 15 patients who developed cancer versus eight (18%) of 44 patients who did not ( P <.001). Malignant transformation of the original oral premalignancy site occurred in 21 cases, and that of a different head and neck site was observed in three. In this study, we report malignant transformation of advanced dysplastic oral premalignant lesions to invasive cancer (19% at 66 months follow-up) that is consistent with that reported by Silverman et al. ( 3 ) (36% at 96 months). We emphasize the high malignant transformation rate of advanced oral premalignant lesions, even after an active short-term chemopreventive intervention. The percentage of patients developing cancer within the isotretinoin arm was not significantly different from that in the β-carotene arm. With 80% power and a 5% two-sided type-I error rate, the small sample size in our study allowed us to detect only a greater than 35% difference in cancer-development rates. Even with this size limitation, however, the data suggest two positive effects of isotretinoin and a third result with extremely important implications for future trials. First, there may have been a short-term cancer-preventive effect of isotretinoin maintenance. This is indicated by the 8-month increase in median time to cancer development for cancer patients who had received isotretinoin, reflecting perhaps the 9-month maintenance. Longer maintenance may have further prolonged time to cancer development for isotretinoin patients. Second, there also may have been an overall positive effect of isotretinoin. This is indicated by the relatively smaller total number of cancers in the isotretinoin group (eight cancers in 26 patients) than in the β-carotene group (16 in 33) (numbers in parentheses reflect patients with multiple cancers). Third, our data suggest that oral premalignant lesion progression may be a valid intermediate end point for future head and neck cancer chemoprevention trials ( 4 , 5 ). This is indicated by the significant association between oral premalignant lesion progression during maintenance and subsequent cancer development ( 6 , 7 ). We currently are conducting a National Cancer Institute-sponsored study of low-dose isotretinoin versus retinyl palmitate with a longer retinoid intervention period and a larger sample size than were used in this study. This new trial may provide a more accurate estimate of the impact of retinoid intervention and determine the validity of oral premalignant lesions as intermediate trial end points ( 6 , 7 ). In this trial, we have also included cellular and molecular biomarkers ( 8 – 14 ) that will provide insight into the process of multistep head and neck carcinogenesis and its reversal. Kaplan-Meier estimate of in situ and invasive squamous cell carcinoma-free survival for each treatment group. The 95% confidence intervals for the 5-year cancer-free survival are .54 to .94 and .58 to .91 for the isotretinoin and β-carotene groups, respectively. 13cRA = isotretinoin.