High-Dose Fenretinide in Oral Leukoplakia Free

Fenretinide is a synthetic retinoid [N-(4-hydroxyphenyl)retinamide] with clinical activity in patients with the premalignant lesion oral leukoplakia (1, 2). It is postulated that fenretinide is active in oral premalignant lesions via modulating cell growth and/or differentiation or inducing apoptosis. A dual, dose-dependant mechanism of action mediates the antineoplastic effects of fenretinide. Fenretinide induced cell differentiation at a low concentration (1 μmol/L) and apoptosis at higher concentrations (3-10 μmol/L) in cancer cells in vitro (3–6). Of note, the induction of differentiation was not observed in mutant F9 murine embryonal carcinoma cells lacking expression of retinoic acid receptor γ and retinoid X receptor α, whereas apoptosis was shown following exposure to fenretinide in both wild-type and receptor-deleted cells (3). These collective results indicate that fenretinide-induced apoptosis is independent of retinoid receptors, whereas fenretinide-induced differentiation is not.

In a previous phase II trial, we found that low-dose fenretinide (200 mg/d) for 12 weeks produced a 34% response rate in patients with retinoid-resistant leukoplakia (never responded or responded and then relapsed during treatment; ref. 1). The responses, however, were partial in all cases, often short lived, and correlated with previous response to natural retinoids (1). Because fenretinide serum levels (mean, 0.230 μmol/L) were at least 10-fold lower than the concentrations required in vitro for induction of apoptosis (>3 μmol/L; refs. 6–8), we hypothesized that the limited clinical activity of low-dose fenretinide in this trial likely was due to retinoid receptor–mediated mechanisms and did not involve retinoid receptor–independent apoptotic effects.

A previous phase I trial showed that fenretinide at 900 mg/m² twice daily in 21-day cycles (7 days on drug followed by 14 days off) was tolerable and produced peak serum drug levels of 3 to 5 μmol/L (9). Based on the preclinical data (3–6), fenretinide at this dose theoretically would produce high serum levels capable of inducing retinoid receptor–independent apoptosis, eliminating premalignant clones, and thus producing sustained clinical responses in oral leukoplakia. We hypothesized that fenretinide would have this cytotoxic-like effect of a rapid burst of apoptotic activity and that the 14-day drug holiday would minimize potential adverse effects, paralleling the rationale behind cytotoxic drug regimens. Therefore, we assessed the effects of high-dose fenretinide on oral leukoplakia in a phase II trial and correlative in vitro studies.

This was an open-label, single-arm, phase II trial conducted at The University of Texas M. D. Anderson Cancer Center. The study was approved by the Institutional Review Board and was conducted in accordance with the provisions of the Declaration of Helsinki and Good Clinical Practice guidelines.

The main eligibility criterion was the presence of bidimensionally measurable or evaluable oral leukoplakia, which was required to have one or more of the following characteristics: presence of dysplasia (any grade), localization to the floor of the mouth, ventrolateral tongue, or soft palate complex, presence of symptoms, and/or extensive area. Other criteria included adequate liver, kidney, and bone marrow function; Zubrod performance status <2; no history of cancer within the preceding 6 mo (except for nonmelanoma skin cancer); and no exposure to retinoid or carotenoid supplements within the preceding 3 mo.

Before study entry, patients were required to sign an informed consent statement. Baseline evaluation included a complete history and physical examination, serum chemistry and hematologic tests, biopsy, and measurement of oral lesions. Patients were then treated with fenretinide (900 mg/m² twice daily) in four 21-d cycles (7 d on drug followed by 14 d off). Toxicity during treatment was regularly assessed using the National Cancer Institute Common Toxicity Criteria version 2.0 modified to include retinoid-specific toxicities (10). At the end of the treatment period, oral lesions were measured and biopsied again. Optional blood collection for pharmacokinetic analyses [i.e., measurements of serum levels of fenretinide and its major metabolite, N-(4-methoxyphenyl)retinamide (4-MPR)] was done at baseline (before treatment) and at 14 d after the last dose of fenretinide (week 12). Details of blood sample processing and analysis methodology were previously described (1).

The primary end point of the study was determination of clinical response rate (complete and partial response) at the end of the 3-mo treatment period. Response was evaluated as follows: complete response, gross inspection revealed no evidence of a lesion; partial response, the size of a lesion or of the sum of the measurements of all lesions decreased by at least 50%; stable disease, lesion sizes increased by <25% or decreased by <50%; disease progression, increase of at least 25% in the sum of the measurements of all lesions or appearance of any new lesion (10).

Simon's minimax two-stage design was applied, and we planned to accrue 16 patients in the first stage; if there were 4 or less responders, the study would be terminated and the treatment would be considered inefficacious. If there were 5 or more responders, additional patients would be added to the study in the second stage to achieve a total of 25 patients. At the end of the trial, the treatment was considered to be efficacious if there were 11 or more responders. The study had 90% power with a one-sided 10% type I error rate, assuming the null response rate would be 30%, and a response rate of 55% or higher would be of interest (25% improvement over previously reported of ∼30% with low-dose fenretinide; ref. 1). Additional end points included toxicity, pharmacokinetics assessments, and histologic response at month 3.

Leukoplakia cell lines MSK Leuk1 and MSK Leuk1s were obtained from Dr. P.G. Sacks (New York University College of Dentistry, New York, NY). The MSK Leuk1 cell line was established from a moderately dysplastic leukoplakia of the oral cavity. These cells are immortal but nontumorigenic (11, 12). MSK Leuk1 cells were exposed to high concentrations of calcium and serum (10% FCS), and the more-advanced cells that had lost sensitivity to the growth-inhibitory and differentiation-inducing effects of calcium and serum were selected to become the MSK Leuk1s subline (11, 12). Both cell lines were maintained routinely in serum-free keratinocyte growth medium (KGM; Lonza, Inc.). The malignant cell line Ca9-22 was established from a human oral squamous cell carcinoma (13). These cells are tumorigenic in immunocompromised nude mice.

The cells were maintained in DMEM/F12 plus 10% FCS but were switched to serum-free KGM 24 h before growth inhibition experiments. Fenretinide obtained from the National Cancer Institute was dissolved in DMSO. This stock solution was diluted into serum-free KGM medium at the beginning of each experiment. Cells were plated in triplicate wells of a 96-well plate and treated with various doses of fenretinide; controls were treated with the carrier (DMSO) alone for up to 3 d. Cell numbers were estimated by staining the monolayer cultures with sulforhodamine B (14), and growth inhibition was calculated from the equation (1 − At/Ac) × 100, where Ac and At represent the absorbencies at 510 nm of control and treated cultures, respectively.

Between February 2003 and April 2004, 21 patients were registered in this study. Six patients were found to be ineligible due to history of cancer within 6 months (3), nonmeasurable disease (2), or refusal of treatment (1). Therefore, 15 patients initiated and completed treatment with the study drug. Their baseline characteristics are described in Table 1. All 15 patients received the four planned cycles of fenretinide with no dose reductions. Self-reported compliance (confirmed by capsule count) was >90% in 13 of 15 patients.

Treatment was well tolerated, with only one patient experiencing grade 3 toxicity (diarrhea lasting <24 hours). No grade 4 toxicities were observed. The complete list of adverse events occurring in more than one patient (regardless of grade) is outlined in Table 2.

All 15 patients were evaluable for clinical response assessment at the end of 3 months. There were no complete responses. Three patients (20%) had a partial response, whereas stable disease was observed in 11 patients (73%). Only one patient (7%) progressed during treatment in terms of clinical response. Histologically, two patients (13%) had downgrading of the dysplasia, seven patients (47%) had no histologic change, and six patients (40%) had worsening of the degree of dysplasia. None of the patients developed invasive cancer within the 3-month follow-up period. After the initial 15 patients had been evaluated, we recognized that the prespecified minimum clinical response rate during the first stage of the trial (i.e., 5 of 16 patients) could not be achieved. This prompted early termination of the study due to lack of efficacy of the investigational agent.

Fourteen patients had both baseline and posttreatment blood samples collected for analysis of drug serum levels. As expected, fenretinide was undetectable in all 14 samples at baseline. The mean (±SD) serum level of fenretinide increased to 0.122 ± 0.093 μmol/L after treatment (i.e., 14 days after the last dose of fenretinide). Similarly, 4-MPR (a major fenretinide metabolite) was undetectable in all 14 samples at baseline and increased to 0.643 ± 0.398 μmol/L (mean ± SD) after treatment. We extrapolated the mean trough level of 0.122 μmol/L at day 21 (2 weeks after last dose) to time of maximum concentration; this calculation showed that fenretinide levels during the initial days of the cycle (which were not assessed directly) were comparable with the high concentration range reported previously for high-dose fenretinide (based on the 14- to 25-hour termination half-life of fenretinide and linear pharmacokinetics; data not shown; refs. 9, 15, 16).

The immortalized less-progressed (MSK Leuk1) and more-progressed (MSK Leuk1s) leukoplakia cell lines and the malignant cell line (Ca9-22) were all sensitive to the growth-inhibitory effects of fenretinide. Growth inhibition was dose and time dependent (Fig. 1). The least-progressed cell line (MSK Leuk1) was more sensitive to lower concentrations of fenretinide than were MSK Leuk1s and Ca9-22 cells and showed inhibition effects below 1 μmol/L fenretinide. In contrast, MSK Leuk1s and Ca9-22 cells were more sensitive to the highest fenretinide concentration (versus MSK Leuk1; Fig. 1).

Although well tolerated, high-dose fenretinide (900 mg/m² twice daily) produced objective responses in only 3 of the first 15 patients (20%). The trial was terminated early (after these 15 patients) by early stopping rule of the protocol (≤4 responses in the initial 16 patients). Considering the hypothesis that high-dose fenretinide would induce apoptosis in oral leukoplakia tissue and thus eliminate premalignant clones, why did the response rate of the present trial not exceed that of the low-dose fenretinide trial we used for a historical control? Two major considerations involved with this question are dose and schedule of fenretinide and mechanisms of fenretinide actions in carcinogenesis.

About dose and schedule, prior phase I studies of high-dose fenretinide, including the same dose and schedule we used, produced high mean peak serum levels of fenretinide of 3 to 10 μmol/L in cancer patients (9, 15, 16). It is possible, however, that drug levels in leukoplakia tissue were inadequate for response, although there are data showing that high tissue levels occur in association with high serum fenretinide levels in other systems. A phase II study (17) of the same (as our) high dose and schedule in renal cancer patients showed that fenretinide concentrated at 3 to 10 μmol/L in serum and at 3 to 8 μmol/L in kidney tumors. High ratios of tissue-to-serum fenretinide levels (associated with lower doses of fenretinide) also have been documented in the bladder (18) and breast (19). Therefore, we do not believe that lack of efficacy in the present trial was due to low levels of fenretinide in target oral tissue.

We further examined the dose issue in correlative in vitro studies in a cell model of human multistage oral carcinogenesis (11): immortalized (MSK Leuk1; established from a clinical leukoplakia sample), progressed (MSK Leuk1s), and malignant cells (Ca9-22). We found that sensitivity to lower-concentration fenretinide was higher in premalignant than malignant cells and that sensitivity to higher-concentration fenretinide was higher in malignant than premalignant cells. This high-dose in vitro result is consistent with our negative high-dose clinical finding. The results of several other in vitro studies also bear on the question of whether premalignant cells are less sensitive to fenretinide than are malignant cells, where the vast majority of data on receptor-independent apoptosis have been generated (20, 21). Low-concentration fenretinide was less inhibitory of colony formation by premalignant SV40 T–immortalized lung cells than by malignant lung cells (22). Sensitivity to fenretinide at low or high doses was similar, however, in human papillomavirus–immortalized premalignant skin cells and malignant skin cells of a human skin carcinogenesis model (23). Different fenretinide effects in the various in vitro systems could relate to differences between organ systems and methods of immortalization [e.g., viral versus spontaneous (e.g., MSK Leuk1)].

What mechanism potentially underlies the ineffectiveness of high-dose fenretinide in both our in vitro and clinical studies? High-dose fenretinide can induce apoptosis through generation of reactive oxygen species (24, 25). Normal and premalignant cells have more endogenous antioxidant capacity than do cancer cells; therefore, normal and premalignant oral cells may tolerate reactive oxygen species and thus counteract apoptosis induction better than do cancer cells.

Our original hypothesis was that high-dose fenretinide would be highly active in less-advanced leukoplakia if low-dose fenretinide (as shown in our previous trial; ref. 1) was moderately active in more-advanced leukoplakia. This hypothesis is consistent with general prevention drug development (much of it in cancer), which hypothesizes that agents active in cancer should be more active in premalignancy. Our in vitro findings contradicted this general hypothesis. Some angiogenesis inhibitors also reflect this contradiction, better preventing the angiogenic switch in more-advanced than less-advanced lesions (26). These findings highlight the importance of conducting preclinical studies of agents in advanced versus nonadvanced carcinogenesis before launching clinical trials. Although limited in the past, preclinical technology and models for testing agents in premalignancy are rapidly emerging.

Fenretinide at the dose and schedule of the present study was not an effective chemoprevention strategy in patients with oral premalignant lesions. Randomized studies of this agent for chemoprevention of breast and bladder cancer also have failed to show efficacy (18, 27). It is important to note, however, that the value of oral leukoplakia as a surrogate end point for cancer development is not certain, notwithstanding substantial data indicating that it may be (28). A very recently reported chemoprevention trial found that oral leukoplakia response did not correlate with oral cancer development (29), and so it cannot be ruled out categorically that high-dose fenretinide would reduce oral cancer risk despite its limited activity in oral premalignancy. Leukoplakia in this prior and the present trial was not selected based on advanced stage and so likely had a relatively moderate overall risk. Our current in vitro results showing that neoplastic oral cells progressed in association with higher sensitivity to high-dose (retinoid receptor independent) fenretinide suggest that high-dose fenretinide may be more clinically active in advanced head and neck premalignancies such as that with severe dysplasia, cyclin D1 aberrations (30, 31), or certain loss of heterozygosity profiles (32, 33). A new formulation of fenretinide incorporated into an organized lipid matrix called LYM-X-SORB may improve the oral bioavailability, increase blood and tissue levels to high micromolar levels, and thus improve clinical outcomes in advanced oral leukoplakia patients; several phase I clinical trials of this formulation are under way (34).

Given the limitations of retinoid-based therapy for oral cancer chemoprevention illustrated by this and previous trials (1, 10, 29, 35–37), our group is now focusing on the development of molecular-targeted agents for patients with oral premalignant lesions. As an example, we are currently conducting a placebo-controlled randomized study of the epidermal growth factor receptor tyrosine kinase inhibitor erlotinib, with cancer development as the primary end point. Molecular-targeted agents may improve therapeutic index and the chances for establishing standard chemoprevention of oral cancer in the future (38).

No potential conflicts of interest were disclosed.