The study schema.
A severely dysplastic exophytic lesion in the vocal cord (A) shows partial resolution at 6 months (B) and complete resolution at 12 months (C).
Complete histologic reversal of moderate laryngeal dysplasia (A) at 6 months (B) and maintained at 12 months (C).
Papadimitrakopoulou VA, Clayman GL, Shin DM, Myers JN, Gillenwater AM, Goepfert H, El-Naggar AK, Lewin JS, Lippman SM, Hong WK. Biochemoprevention for Dysplastic Lesions of the Upper Aerodigestive Tract. Arch Otolaryngol Head Neck Surg. 1999;125(10):1083-1089. doi:10.1001/archotol.125.10.1083
To evaluate the efficacy and secondarily the toxic effects of biochemopreventive therapy (high-dose isotretinoin [13-cis-retinoic acid], α-tocopherol, and interferon alfa) in the reversal of advanced premalignant lesions of the upper aerodigestive tract and to correlate the therapeutic events with modulation of biomarkers.
Prospective, nonrandomized chemoprevention trial.
Tertiary cancer care referral center and ambulatory care.
Thirty-six patients with advanced premalignant lesions of the upper aerodigestive tract, without cancer during the 2 years before the intervention, with evaluable lesions, and without retinoid therapy for 3 months before the trial.
Administration of oral isotretinoin (100 mg/m2 per day), oral α-tocopherol (1200 IU/d), and subcutaneous interferon alfa (3 megaunits per square meter twice weekly) for 12 months, with serial biopsies and clinical examination at 0, 6, 12, and 18 months from study start.
Main Outcome Measures
Clinical and histologic responses to the intervention.
Of the 36 patients, evaluation was possible in 30 for response at 6 months and in 21 at 12 months. At 6 months, there were 10 pathologic complete responses and 7 partial responses; at 12 months, 7 complete and 3 partial responses. A striking difference in response was observed in favor of laryngeal lesions (9/19 [47%] complete response rate at 6 months and 7/14 [50%] at 12 months vs 1/11 [9%] and 0/7 [0%], respectively, for oral lesions). Toxic effects were acceptable and did not exceed grade 3.
Biochemoprevention is a promising biologic approach for laryngeal dysplasia and needs to be investigated further.
ADVANCED premalignant lesions of the upper aerodigestive tract carry a high risk for transformation to invasive cancer, which has been described in mainly retrospective series in the literature. They have an incidence of approximately 30% to 40%, depending on length of follow-up and the different study criteria.1- 3 Chemoprevention, defined as the administration of agents to block or reverse carcinogenesis, is a novel strategy for reducing the incidence and mortality of upper aerodigestive and lung cancers,4,5 which have decreased only marginally during the last 2 decades, despite advances in multiple-modality treatments.6- 8 The development of chemoprevention trials in the upper aerodigestive tract has been based on the concepts of field and multistep carcinogenesis, which describe the accumulation of events that lead to the emergence of malignant clones in a field diffusely affected by chronic carcinogen exposure.9,10 A series of clinical chemoprevention trials in the upper aerodigestive tract setting have been completed in the past 10 years, with several important findings.11- 13 Studies of oral leukoplakia (carrying histologic characteristics ranging from hyperplasia and hyperkeratosis to varying degrees of dysplasia) have shown that single-agent isotretinoin (13-cis-retinoic acid) treatment was ineffective at reversing advanced premalignant lesions (moderate to severe dysplasia), which carry an unfavorable prognosis.12 Investigations into the biological features of these lesions have shown that they demonstrate an increased degree of genomic instability, a high rate of p53 protein and gene alterations, and a lack of up-regulation of retinoic acid receptor-β (a signpost for binding of retinoids to the cell nuclear receptors).14 Based on these experiences, a combination biochemoprevention trial was designed using high-dose isotretinoin, interferon alfa, and α-tocopherol. The rationale for this combination was based on the previous experience with isotretinoin11,12; the knowledge that α-tocopherol is modestly active in reversing oral leukoplakia with virtually no toxic effects,15 has shown synergistic chemopreventive activity with retinoids in an animal model,16 and may abrogate the toxic effects of retinoids17; and the activity of interferon alfa as a single agent and its synergistic effects with retinoids in preclinical studies and clinical settings.18,19 We aimed to reverse advanced premalignant lesions of the oral cavity (moderate or severe dysplasia) and the larynx (mild to severe dysplasia). We now report the outcomes of this study, having accrued 36 patients and having encountered an encouraging response rate in the laryngeal premalignant setting.
Treatment followed a protocol that was approved by the Institutional Review Board of the University of Texas M. D. Anderson Cancer Center, Houston. Patients were eligible if they had histologic evidence of moderate or severe dysplasia (oral cavity, oropharynx); mild, moderate, or severe dysplasia of the larynx; or measurable, evaluable disease from which biopsy specimens could be obtained. Participants should have discontinued previous retinoid therapy at least 3 months before study entry; have received no other chemotherapy, immunotherapy, or hormonal therapy within 3 weeks of entry into the study and during the study; be older than 16 years; have adequate bone marrow, liver, and kidney function and a baseline fasting triglyceride level no more than 2.5 times the upper limits of normal; and have no other serious intercurrent illness and no cancer history for at least a year before study entry. All patients signed an informed consent indicating that they were aware of the investigational nature of this study and consented to periodic photography and biopsies to document clinical and histologic responses. Patients of child-bearing potential were required to practice adequate contraception. Life expectancy of at least 12 weeks and a Zubrod performance status20 of no more than 2 was required.
Thirty-six individuals with biopsy-confirmed advanced premalignant lesions of the upper aerodigestive tract, 27 men and 9 women with a median age of 58 years (range, 33-79 years), were enrolled in this study (Table 1). These individuals were self-referred or more commonly referred by an ear, nose, and throat specialist or a head and neck surgeon because of recurrent dysplastic lesions in the upper aerodigestive tract that were excised or underwent biopsy repeatedly. The sites of dysplasia were laryngeal in 23 patients and oropharyngeal in 13.
After review of previous biopsy material to verify results of past histologic studies, a new biopsy specimen was obtained in most cases to document the degree of dysplasia and to rule out the presence of invasive cancer. Biopsy specimens obtained from outside institutions were acceptable as baseline specimens only if they were obtained within 2 months from patient accrual to the study and only if the patient's clinical presentation did not suggest progression during this time. The specimens were small but of sufficient depth to allow for complete histologic evaluation of the degree of dysplasia, and took into account that a measurable lesion would have to remain after the biopsy specimen was taken to allow for evaluation of response. In most cases, a single biopsy specimen of a solitary lesion was obtained. In cases where the lesions were diffuse or multiple, 2 or 3 specimens were taken, and the lesion with the highest degree of dysplasia was defined as the index lesion for subsequent response evaluation. Baseline physical examination through oral cavity inspection or indirect laryngoscopy followed by videostroboscopy, which also recorded any abnormalities in vibratory function, was performed in all cases, and photographic documentation for future comparison was obtained. Pretreatment evaluation also included a complete history, physical examination, and laboratory analysis (complete blood cell count with differential; liver function tests; levels of serum urea nitrogen, creatinine, electrolytes, fasting serum triglyceride and cholesterol, 13-cis-retinoic acid, and cotinine; and serum pregnancy test). Monitoring included monthly clinic visits with repeated laboratory analysis, head and neck surgery visit with complete physical examination, inspection of the lesion, photographic documentation, and, in cases of laryngeal lesions, complete videostroboscopic examination with photographic documentation of the response to treatment every 3 months. Patients were taking isotretinoin at a dose of 100 mg/m2 per day; α-tocopherol, 1200 IU/d; and subcutaneous interferon alfa, 3 megaunits per square meter, twice weekly (Figure 1). Patients were instructed on aseptic interferon alfa admixture and administration techniques at our center so that they could self-administer the medication at home. There are 4 preestablished dose levels for the medications allowing for dose tapering according to the adverse effects encountered (Table 2). Starting dose level was level −1, and dose was escalated to level 0 if no greater than grade 2 adverse effects were encountered. One course of treatment was defined as 1 month. Patients were instructed to keep a diary to monitor potential adverse effects, and pill counts were performed by the research nurse during the monthly clinic visits. Biopsies and clinical response evaluations were performed again at 6 and 12 months after initiation of treatment. In addition, a biopsy was performed at 6 months after study completion for histologic monitoring. After response evaluation at 6 months, the patients who had treatment responses or stable disease received 6 additional months of treatment.
All patients underwent assessment for toxic effects at each visit according to the National Cancer Institute Common Toxicity Criteria21 and following a special scale for retinoid-induced toxic effects (Table 3). These specific criteria were used in addition to the National Cancer Institute–World Health Organization–M. D. Anderson Cancer Center Toxicity Criteria and superseded them.
The recommended dose modifications for hematologic toxic effects were as follows: no change for lowest granulocyte level of greater than 1000 and/or lowest platelet level of greater than 100×109/L, decrease by 1 level for lowest granulocyte level of 500 to 1000 or lowest platelet level of 50×109/L to 100×109/L, decrease by 1 additional level for lowest granulocyte level of less than 500 or lowest platelet level of less than 50×109/L, and discontinuation until resolution for infection or bleeding. All drugs were held pending hematologic recovery to granulocyte level of at least 1000 and platelet level of greater than 100×109/L and recovery to grade 2 or less of nonhematologic effects. The recommended dose modifications for nonhematologic toxic effects are as follows: no change for grades 0 to 2, decrease by 1 level for grade 3 after treatment interruption until resolution to at least grade 2 toxic effects, and treatment discontinuation (off-protocol) for grade 4 toxic effects.
Response was evaluated using results of clinical examination and biopsy review.
Complete response (CR) was defined as disappearance of all evidence of severe dysplasia or carcinoma in situ for at least 1 cycle of therapy or 4 weeks. The patient should be free of all symptoms of cancer. Partial response (PR) was defined as 50% or greater decrease in the sum of the products of diameters of all measured lesions persisting for at least 1 cycle of therapy or 4 weeks. No lesion may increase in size, and no new lesion may appear. Stable disease was defined as no change in lesion size and no development of new lesions. Progressive disease (PD) was defined as any increase of greater than 25% in the sum of the products of diameters of any measurable lesions or in estimated size of nonmeasurable lesions, the appearance of an unequivocal new lesion, or progression to invasive carcinoma.
Complete response was defined as complete reversal to nondysplastic squamous epithelium; PR, regression of dysplasia to a lower grade; no response, persistence of pretreatment histologic grade; and PD, progression in grade of dysplasia or to invasive cancer.
Criteria for discontinuing therapy included the following: development of histologic or clinical PD as invasive carcinoma; unacceptable toxic effects defined as unpredictable, irreversible, or grade 3 not resolved with dose reduction; noncompliance by patient with protocol requirements; and patient refusal. All patients were observed for at least 1 month from the last date of treatment.
A total of 36 patients were entered into this study. Of these, 30 reported present or former exposure to tobacco, and 24, history of alcohol use. Of the 23 patients with laryngeal dysplasia, 16 had a history of gastroesophageal reflux, a factor considered by many as contributory to abnormalities in this area.22 Six of the patients had a previous cancer in the upper aerodigestive tract and thus were at risk for second primary tumor development. Twenty-one patients had previous history of dysplasia, and many of these had undergone repeated excisional procedures to address this problem.
The objective response to treatment that was evaluable in 30 patients at 6 months and in 21 patients at treatment completion is shown in Table 4. Six patients could not undergo evaluation at 6 months because of noncompliance,2 ineligibility,1 and toxic effects related to withdrawal2; 1 patient was still undergoing the fourth month of treatment. The percentage of patients with clinical CR or PR at 6 months was 60%, and 27% had stable disease. Dysplasia disappeared in 10 (33%) of the patients, and the degree of histologic dysplasia was reduced in 7 (23%). Overall histologic improvement occurred in 57% of 30 patients. The initial response was clearly visible on gross examination and appeared as thinning of the lesion of leukoplakia with reduction in the area it occupied or reversal of the irregular and edematous appearance of the vocal cords (Figure 2). The histologic changes following treatment were striking with normalization of grossly abnormal epithelium, with dysplasia extending from the basal membrane to the epithelial surface (Figure 3). There was a remarkable difference in terms of response favoring the laryngeal premalignant setting. Nine (47%) of 19 patients demonstrated complete histologic reversal of dysplasia in the larynx (CR), whereas only 1 patient (9%) had a histologic CR in the oral cavity (P=.05). Response rates for clinical CR and PR were 68% and 45% for larynx and oral cavity, respectively (P=.27). At treatment completion (12 months), evaluation for response was possible in 21 patients (including 5 patients with PD earlier than and at 6 months and 1 patient who withdrew at 6 months after achieving a CR). The overall histologic response rate at 12 months was 48%, and the clinical response rate was 57%. The same trend regarding favorable response of laryngeal premalignant lesions was seen at 12 months. A histologic CR was seen in 50% of the patients with laryngeal premalignant lesions, compared with none of the patients with oral cavity lesions (P=.05).
Progression to cancer occurred in 4 patients with laryngeal premalignant lesions, 3 of whom had a history of cancer in the upper aerodigestive tract, and in 4 patients with oral premalignant lesions.
Compliance during treatment was good. Only 2 (5%) of 36 patients withdrew because of noncompliance, and in the rest of the patients compliance was excellent. All of the patients had at least low-grade toxic reactions (Table 5). The most commonly encountered toxic effects were mucocutaneous, including skin dryness, cheilitis, conjunctivitis, arthralgia and myalgia, fatigue, fever, hypertriglyceridemia, and chills. Twenty-three patients (64%) had no more than grade 2 reactions, and 13 (36%) had grade 3 reactions. No grade 4 reactions were seen, perhaps because of the careful tapering of the dose and close observation of the patients with marked toxic effects. Three patients withdrew because of toxic effects, 2 of these because of poorly controlled hypertriglyceridemia, despite dose reduction and antihyperlipidemic therapy.
In the initial oral leukoplakia treatment trial conducted in 1986 by Hong et al,12 high-dose isotretinoin therapy was shown to be highly effective against oral leukoplakia, but the few cases with moderate or severe dysplasia included in this trial appeared to resist single-agent retinoid therapy. The experience of others with single-agent retinoids for the treatment of laryngeal dysplasia is limited to 2 small studies: one with the synthetic derivative of retinol, etretinate, and the other with retinyl palmitate.23,24 The conclusion from these small studies was that retinoids appear to be active against laryngeal dysplasia, but histologic evaluation was only documented in the first study and revealed only partial reversal of severe dysplasia. Therefore, single-agent retinoids do not appear to be able to reverse advanced premalignant lesions of the upper aerodigestive tract. Our results suggest that a combination regimen including interferon alfa shows remarkable activity in the setting of laryngeal dysplasia. Responses in the setting of laryngeal dysplasia occurred regardless of smoking status or presence of gastroesophageal reflux. The contribution of tobacco use to the results of treatment is unclear, but it did not seem to affect responses. Of our 36 patients, 17 were former smokers, 13 were current smokers (of whom 5 quit on entry into the study), and 6 had no smoking history. Among current smokers, there were 5 histologic CRs, 2 PRs, 2 with stable disease, and 3 with histologic PD. Thus, major histologic response was observed even in patients who underwent persistent carcinogenic exposure. The role of gastroesophageal reflux as an etiologic factor for dysplasia of the larynx could not be formally studied, since no routine assessment of this risk factor, such as using pH probe, was performed. Nevertheless, 16 of the 23 patients with laryngeal dysplasia had symptoms or clinical history suggestive of gastroesophageal reflux and were treated with omeprazole sodium and behavioral modifications. The contribution of antireflux measures to the treatment result cannot be assessed herein, although responses were observed regardless of the presence of reflux.
The cause of the differential response between oral cavity and larynx lesions is unclear. It is possible that the biological features underlying the disturbed phenotype in both sites are different in terms of degree of genomic alterations, as is also true for invasive cancer in both sites. We could hypothesize that the accumulation of genetic damage necessary to result in the characteristics of moderate-severe dysplasia in tongue lesions is more pronounced because of their anatomic location and wider exposure to carcinogens, and therefore that the dysplastic tongue lesions are more resistant to chemopreventive intervention. In contrast, laryngeal lesions might be less genetically altered and thus more sensitive to intervention. Biomarker analysis, including analysis of specific gene alterations, generalized genomic instability, the status of retinoic acid receptor-β, and proliferation status, are ongoing, and we hope it will answer some of these questions. Two important findings already have been derived from this analysis. Mao et al25 used polymerase chain reaction analysis of microsatellite DNA sequences in cells from these precancerous lesions to determine the frequencies of loss of heterozygosity at chromosomal loci commonly deleted in head and neck cancer, ie, 9p21, 3p14, and 17p13. Among 9 patients who exhibited loss of heterozygosity at 9p21 in pretreatment biopsy specimens, the genetic loss persisted in 8, including 3 of 4 patients with histologic CRs. This observation suggested that genetic alterations persist in the tissue, despite phenotypic reversal of dysplasia. The second important finding from the biomarker analysis was that high p53 expression at baseline that was maintained after treatment was associated with progression in 12 (86%) of 14 patients, in 6 (43%) of whom invasive carcinoma developed, whereas it developed in none of the patients with low levels of p53 expression during treatment (P<.001).26 Thus, p53 status and responsiveness to biochemoprevention appear to be associated, suggesting that p53 alterations can serve as a prognostic factor for treatment response and also as a target to overcome resistance to intervention.
As a conclusion, our initial hypothesis that more aggressive intervention in advanced oral premalignant lesions would lead to reversal of the resistance to retinoids seems to be defeated at least with the present data, and a different approach should be sought in this type of lesions.
Regarding the encouraging response in the larynx, we believe that this type of intervention with combination of interferon alfa and isotretinoin is valuable and should be further pursued. Different retinoids have shown activity in this setting, and interferon has shown activity in laryngeal papillomatosis in several studies and probably contributed to the complete reversal of dysplasia seen in our study. Based on these data, we have launched a new trial addressing exclusively laryngeal dysplastic lesions in which the same treatment protocol will be used for a year, followed by 2 years of maintenance therapy with fenretinide or placebo. Fenretinide has a different mechanism of action; ie, it has been described as inducing apoptosis,27 an effect potentially important for its chemopreventive activity, and this induction appears to be independent of nuclear retinoid receptors. Fenretinide has shown activity in resistance to other retinoid cell lines and lesions28,29 and has a favorable toxicity profile.30,31 Fenretinide has been shown to have a significant preventive effect against the development of contralateral breast cancer in premenopausal women and ovarian cancer among patients with breast cancer.32,33 Importantly, fenretinide is active in preventing development of new lesions after resection of oral premalignant lesions.30 The interest in using fenretinide as maintenance therapy after completion of induction biochemoprevention lies in its potential to induce apoptosis and thus possibly to eliminate remaining abnormal clones of cells or to reverse resistance of lesions that have not completely responded to the induction therapy. A formal assessment of gastroesophageal reflux as a measure for response to treatment will be included in this trial. We hope that this new study will lead to a more definitive assessment of the value of chemopreventive intervention in this setting.
Accepted for publication June 25, 1999.
Reprints: Vali A. Papadimitrakopoulou, MD, Department of Thoracic–Head and Neck Medical Oncology, University of Texas, M. D. Anderson Cancer Center, 1515 Holcombe Blvd, Box 80, Houston, TX 77030 (e-mail: email@example.com).