Chemoprevention with systemic retinoids has demonstrated promise in decreasing the incidence of new primary nonmelanoma skin cancers (NMSCs) in immunocompromised post-transplantation recipients. There is limited evidence for the use of systemic retinoids in the nontransplantation patient. To the authors' knowledge, this is the first randomized controlled trial to assess the efficacy of acitretin as a chemopreventive agent in nontransplantation patients at high-risk for NMSC.
The study was designed as a prospective, randomized, double-blind, placebo-controlled clinical trial. To test the possible skin cancer-preventing effect of a 2-year treatment with acitretin, 70 nontransplantation patients aged ≥18 years who had a history of ≥2 NMSCs within 5 years of trial onset were randomized to receive either placebo or acitretin 25 mg orally 5 days per week. The primary outcome measure was the rate of new NMSC development.
Seventy patients were randomized to receive either acitretin alone (N = 35) or placebo (N = 35). During the 2-year treatment period, the patients who received acitretin did not have a statistically significant reduction in the rate of new primary NMSCs (odds ratio, 0.41; 95% confidence interval, 0.15-1.13; 54% vs 74%; P = .13). However, using the incidence of new NMSC, the time to new NMSC, and total NMSC counts, an umbrella test indicated a significant trend that favored the use of acitretin (chi-square statistic, 3.94; P = .047). The patients who received acitretin reported significantly more mucositis and skin toxicities compared with the patients who received placebo.
It is estimated that nonmelanoma skin cancers (NMSC) account for nearly 3.5 million new cancer diagnoses yearly, far exceeding the annual incidence of all other cancers combined.1 Although rarely the primary cause of death, NMSCs result in a large health care expenditure each year and can lead to significant morbidity in some patients because of loss of function and disfigurement. Basal cell carcinomas (BCC) and squamous cell carcinomas (SCC) make up the vast majority of NMSCs and occur at a ratio of approximately 4:1.2
Multiple risk factors for NMSC have been identified through decades of epidemiologic research, including older age, fair skin phenotype, tendency toward sunburn, chronic sun exposure, previous precancerous lesions (actinic keratoses), previous skin cancers, chronic immunosuppression, and infection with the human papilloma virus.3, 4 It has been demonstrated that patients who have a history of NMSC have the greatest risk of developing new primary NMSC.5 The rate of new primary NMSC in patients who have a history of ≥10 NMSCs nears 100%.6
Systemic retinoids have demonstrated promise in reducing the rate of NMSC in certain high-risk populations. In previous randomized controlled trials, the use of systemic retinoids was proven effective in reducing the development of new primary NMSCs or new precancerous lesions in immunocompromised patients after transplantation.7-10 In a trial by Bavinck et al, patients who underwent renal transplantation had a significant reduction in new primary NMSC with acitretin compared with placebo over a 6-month period.7 In another trial, the effect of acitretin over a 2-year period significantly decreased the development of new SCC.10 In a trial led by de Sevaux et al, the utility of acitretin revealed a decrease in the rate of new actinic keratoses, but not in new skin malignancies.9 Systemic retinoids also have demonstrated benefit in preventing NMSC in high-risk, nontransplantation populations, such as patients who receive psoralen-ultraviolet A light treatment and patients with xeroderma pigmentosa.11, 12
Despite the benefit gained through the use of retinoids like acitretin, in patients who undergo transplantation, the utility of these agents for those who do not undergo transplantation has yet to be fully elucidated. In 1997, Levine et al reported that high-risk patients (with >4 previous NMSC) who received either systemic retinol or isotretinoin had no difference in the incidence or the time to first new NMSC compared with patients who received placebo.13 Another trial demonstrated that systemic retinol significantly reduced the risk of new SCC compared with placebo in patients who reportedly were at moderate risk for NMSC.14 However, pooled data from both of those trials did not demonstrate a significant difference in the time to first SCC or BCC.15 In a trial that assessed the chemopreventive effect of low-dose isotretinoin in patients with a history of BCC, no difference was observed in the incidence or the time to first new BCC.16 The use of acitretin, a newer second-generation retinoid, has not been reported to date in high-risk, nontransplantation patients. Here, we report the results from a randomized, placebo-controlled, double-blind, chemoprevention trial assessing the utility of acitretin in patients with a history of multiple NMSCs.
MATERIALS AND METHODS
To be eligible for this trial, patients had to be aged ≥18 years, were required to have a history of ≥2 NMSCs and to have received previous treatment for all visible SCC and BCC, and could not have received any retinoids within 1 year of registration. Approval by the investigational review boards of participating centers before patient enrollment was required, and all participants were required to provide written informed consent before trial entry. Initially, this clinical trial included organ-transplantation recipients who were receiving immunosuppressive agents, and we planned to stratify patients by transplantation status. However, because of low accrual of this subset of patients caused by exclusion from the study based on laboratory abnormalities, subsequently, the trial was modified to exclude further enrollment of transplantation recipients.
Patients were excluded if they had a creatinine level >1.5 times the upper limit of normal, a serum cholesterol level >250 mg/dL, a serum triglyceride level >2.5 times the upper limit of normal, or elevated liver function tests (LFTs). Patients also were excluded if they were women of childbearing potential or if they had a life expectancy <5 years.
The clinical trial was designed as a prospective, randomized, double-blind, placebo-controlled trial and was approved by the Mayo Clinic Institutional Review Board according to US federal guidelines. Patients were stratified according to age, number of skin cancers within the past 5 years, date of the most recent skin cancer occurrence, transplantation status (initial patients), patient-reported sunburn susceptibility (Fitzpatrick skin-type), and assessment of visible skin damage. Assessment of visible skin damage involved categorization based on the presence of actinic keratoses, lentigines, wrinkling, and telangiectasia. Minimal, moderate, or extensive skin damage was defined as 1, 2 or 3, or 4 of these findings, respectively. Next, the patients were randomized to 1 of 2 treatment arms (acitretin vs placebo).
The acitretin and placebo capsules were provided in blinded fashion by Roche Pharmaceuticals (Basel, Switzerland). Patients were provided either 25 mg acitretin or an identical appearing placebo. Capsules were taken orally 5 days weekly for 2 years.
Initial history, skin examination, and laboratory studies were obtained less than 90 days before registration. Follow-up visits were scheduled at 1 month, 6 months, 12 months, 18 months, and 24 months. At the time of trial initiation and at each subsequent follow-up, the skin was examined by the patients' dermatologists.
The effect of acitretin on the development of new primary NMSC was assessed according to the incidence of new SCC or BCC in these high-risk patients between the start of the trial through 2 years. The effect of acitretin also was assessed according to the time to first new NMSC based on the date of the last NMSC resection and the time on study. To be included as developing a new primary NMSC within the first year, a patient must have had an NMSC documented from the start of the trial to either the 6-month or 12-month follow-up visit. The development of a new primary NMSC within the second year was defined as having a documented NMSC from 12 months to either the 18-month or 24-month follow-up visit. Review of the medical records up to 2 years after the date on study was completed for all patients, including those who discontinued treatment. The degree of actinic damage from the start of the study to its completion also was assessed.
Adverse events were graded at each follow-up visit using the Common Terminology Criteria for Adverse Events grading system (version 2.0; National Cancer Institute, Bethesda, MD).17 The following symptoms were specifically monitored: alopecia, skin toxicities, mucositis, arthritis, fatigue, headaches, nausea, and vomiting. Monitoring of biochemical fluctuation was determined by serum creatinine, cholesterol, triglyceride, potassium, and LFTs at every follow-up visit.
The primary outcome measure was the rate of new primary NMSC development. All hypothesis testing was carried out using a 2-sided alternative and a 5% type I error rate using the SAS System for Unix software package (version 9; SAS Inc., Cary, NC).
For primary analysis, we used the Fisher exact test to compare the rate of new NMSC development between the acitretin and placebo treatment arms.18 The numbers of new NMSCs and maximum adverse event grades were compared using nonparametric Wilcoxon tests.19 A multivariate logistic regression model was constructed to examine which clinical and demographic characteristics were related to the development of a new NMSC.20
The time to new NMSC development was measured from the date of the last NMSC resection and from the date on study and was compared between the acitretin and placebo treatment groups using Kaplan-Meier survival estimates and log-rank and Wilcoxon tests. A multivariate Cox model was used to test for a consistent treatment effect after adjusting for baseline characteristics.21 With a small sample size, the power of each individual test is low. The O'Brien umbrella test was used to determine whether the combined tests indicated a significant trend toward favoring acitretin.22, 23 The variables that were included in this test for trend included the incidence of new NMSCs, the time to developing new NMSCs, and the total NMSC count.
The original design was to have 110 patients per treatment arm, which would provide 80% power to detect a 33% difference in NMSC incidence rates. The attained sample size of 35 patients per group provided 51% power to detect a difference of incidence in NMSC of 11% versus 33%. This sample size provided 82% power to detect a difference in NMSC of 5% versus 33% or 11% versus 43%.
In total, 73 patients were enrolled from May 26, 2000 to April 16, 2004, when the trial closed because of slow accrual. Patient flow is illustrated in Figure 1. In all, 70 nontransplantation patients, including 35 in the acitretin arm and 35 in the placebo arm, initiated protocol treatment and were included in the intent-to-treat analysis. Analysis was completed on the entire available patient cohort, including the 2 initially enrolled transplantation recipients and with the exclusion of those 2 patients. Virtually identical results were observed with both cohorts. Data are presented below for the nontransplantation patient cohort. The demographics, clinical, and other patient characteristics were well balanced (see Table 1).
Table 1. Distributions of Baseline Characteristics
Attrition rates were relatively high in both groups. In total, 53 patients completed the 2-year study according to protocol. Reasons for patients stopping treatment were similar across the 2 treatment arms, and the most common reason was refusal of further treatment (acitretin arm, 9 of 35 patients [26%]; placebo arm, 6 of 34 patients [17%]; P = .22). Of the 9 patients in the acitretin arm who stopped therapy early, 5 reported grade 1 and 2 skin toxicities and alopecia, 1 reported dyspepsia, 1 stopped because of a feeling of polypharmacy, 1 had recurrence of a previously resected invasive SCC and refused further treatment, and 1 refused further treatment because of multiple, stable, nondermatologic medical comorbidities. Two patients in the acitretin arm were lost to follow-up and had no retrievable information throughout the 2-year period. After these 2 patients were censored, the primary endpoint remained unchanged. Of the 6 patients in the placebo arm, 2 refused further treatment because of nonspecific arthritic symptoms, 1 stopped because of an exacerbation of chronic psoriasis, 1 patient died from other medical comorbidities, and 2 stopped for unclear reasons.
For the primary outcome measure (Fig. 2), the rate of new primary NMSC development, although the acitretin arm fared better numerically, there was no statistically significant difference between the placebo arm versus the acitretin arm (odds ratio, 0.41; 95% confidence interval [CI], 0.15-1.13; P = .13). Likewise, no statistically significant difference was observed in the time to developing a first new NMSC between the treatment arms from the date of study initiation. The time to the first new NMSC for the 2 treatment arms is illustrated in Figure 3, and Table 2 provides the incidence of new NMSC at each evaluation interval. Table 2 indicates that 46% of patients in the acitretin arm had no NMSC at all compared with 26% of patients in the placebo arm. At 6 months, the difference still was in favor of acitretin (23% vs 40%). Although the numerical differences appeared large, the small sample size prohibited a definitive statement regarding efficacy. The time to the first new NMSC also was analyzed from the date of the last NMSC resection and produced similar results.
Table 2. Patients Who Developed a First, New Nonmelanoma Skin Cancer by Date and Treatment Arm
A multivariate logistic analysis was performed to assess the risk factors for developing a new NMSC (Table 3). After adjusting for other baseline characteristics, patients on the acitretin arm had numerically lower rates of developing new NMSC; however, the difference did not reach statistical significance (odds ratio, 0.33; 95% CI, 0.10-1.04; P = .06). No other characteristics identified were associated with a lower rate of NMSC.
Table 3. Logistic Regression for New Nonmelanoma Skin Cancer Development Using Treatment Arm and Baseline Characteristics
Sunburn susceptibility: Type 1 or 2 (vs type 3 or 4)
Age group: ≤59 y (vs ≥70 y)
Skin damage: Extensive (vs minimal or moderate)
Age group: 60-69 y (vs ≥70 y)
A multivariate Cox model (Table 4) indicated that, after adjusting for other baseline characteristics, patients who received acitretin had a significantly longer time to first new NMSC from the date of the last NMSC resection (hazard ratio, 0.48; 95% CI, 0.25-0.92; P = .03). A history of 2 to 5 previous NMSC resections, compared with ≥6 resections, also was associated with a longer time to developing a first new NMSC. Patients ages 60 to 69 years and patients with their most recent NMSC resection within 12 months of study initiation had a shorter time to developing new NMSC.
Table 4. Cox Model of Time to New Nonmelanoma Skin Cancer Development From the Last Nonmelanoma Skin Cancer Resection by Treatment Arm and Baseline Characteristics
Nineteen patients (54%) and 26 patients (74%) developed at least 1 NMSC during the course of the trial in the acitretin and placebo arms, respectively. Analysis of the total number of new SCCs and BCCs separately is illustrated in Figure 4. Noting that 1 patient was an outlier in the placebo group, with 24 documented BCCs during the 2-year period, the total number of NMSCs observed in the acitretin arm was significantly less over the 2-year period (52 vs 119 NMSCs; P = .02). The mean ± standard deviation number of NMSCs per patient in the acitretin and placebo arms was 1.5 ± 2.12 NMSCs and 3.4 ± 4.51 NMSCs, respectively. The degree of actinic damage at each evaluation interval is outlined in Table 5, and a significant difference was not observed between the treatment and placebo arms; however, there was a considerable degree of missing documentation over the course of the 2-year study period.
Table 5. Degree of Actinic Damage by Date and Treatment Arm
Using multiple endpoints assessed in this trial (incidence of new NMSCs, time to new NMSCs, and total NMSC counts), O'Brien's umbrella test favored the use of acitretin (chi-square statistic, 3.94; P = .047). With censorship of the 2 patients in the acitretin arm who were lost to follow-up, this P value was .08.
Table 6 lists the maximum reported toxicities according to the Common Terminology Criteria grading system (version 2.0) by treatment arm.17 Patients in the acitretin arm reported significantly more alopecia, mucositis, and skin toxicities compared with patients in the placebo arm. More patients on the acitretin arm had hypertriglyceridemia and elevated LFTs compared with patients on the placebo arm; however, nearly all events were grade 1, and the differences were not statistically significant between treatment arms (P = .33 and P = .49, respectively).
Table 6. Maximum Reported Common Toxicity Criteria Adverse Event Grades
To our knowledge, this is the first published trial to investigate the utility of acitretin as a chemopreventive agent against NMSC in high-risk, nontransplantation patients. In this randomized, double-blind, placebo-controlled trial, acitretin failed to significantly reduce the incidence or the time to first new NMSC in patients who had ≥2 previous NMSCs. However, there was a significant reduction in total NMSCs, and we observed a trend that favored the utility of acitretin over the 2-year period.
Positive and confirmatory data have been reported regarding the utility of retinoids as chemopreventive agents in immunocompromised patients (ie, renal transplantation recipients), as reported previously.7, 9, 10 However, data supporting the use of these agents for chemoprevention in high-risk, nontransplantation, presumably immunocompetent patients have not been adequately examined.
Several aspects of the current trial may account for the lack of benefit observed in the study arm. Most important, as a pilot study, the current trial was designed to detect a large effect size (33% difference in incidence rates), but this was not achieved. The observed effect size was 25%, which, at the least, is worthy of further confirmatory investigation. This lack of power may account for the impressive yet nonsignificant odds ratio observed regarding the cumulative incidence of NMSC in the acitretin arm (odds ratio, 0.33; 95% CI, 0.10-1.04; P = .06). Second, although there was not a statistically significant difference, the rates of attrition were greater in the study arm and may have affected the results. Only 4 of 9 patients who withdrew in the acitretin arm had documented NMSC after discontinuation. Of these, only 1 patient had more than 1 NMSC (ie, 4 NMSCs [3 BCCs and 1 SCCs]) over the remaining 2-year period. This is intriguing, because other trials demonstrated a more profound rebound effect with the development of numerous NMSCs once retinoid therapy was discontinued.7, 10, 24-26 This suggests that the current trial participants may have constituted an inherently lower risk population despite an NMSC history similar to that of previously studied transplantation populations.10, 25 Approximately 50% of patients in the current trial had only 2 to 5 previous NMSCs; this lower risk population, relative to the remaining patients with ≥6 previous NMSCs, also may account for the decreased efficacy in the overall cohort. Third, the benefit observed with the test of trend was no longer statistically significant after we censored the 2 patients in the acitretin arm who were lost to follow-up.
Despite the primarily negative results, the results from this trial support the finding that acitretin decreased the total number of SCCs and BCCs compared with placebo. In several trials, systemic retinoids have been more effective in suppressing new SCCs compared with new BCCs.10, 14, 16 However, in the current trial, acitretin appeared to be similarly efficacious in suppressing both SCC and BCC. The lack of effect on cumulative incidence rates probably is not related to the acitretin dose. In the current trial, we used acitretin at a dose of 25 mg orally for 5 days per week, a dose that was selected based on the positive results from 3 randomized controlled trials assessing the agent's efficacy in renal transplantation recipients.7, 9, 10 In the trial reported by de Sevaux et al, no difference in NMSC development was observed when higher dose acitretin was compared with lower dose acitretin. However, only 3 of 14 patients in that trial were able to maintain high-dose therapy (0.4 mg/kg daily).9
In the current trial, we did not observe a statistically significant difference in any of the well known side effects of retinoids on liver function disturbances and/or hyperlipoproteinemia. More important, these laboratory abnormalities were not the cause of the high attrition rates in both treatment arms. This contrasts with the laboratory alterations reported in the study by George et al, when higher doses (up to 50 mg daily) of acitretin were used.10 The majority of patients in the acitretin arm who did have adverse effects in the current trial had the common mucocutaneous side effects of retinoid use.27 Although not life-threatening, these adverse effects can be sufficiently bothersome to patients that they discontinue treatment.
There has been increasing evidence that systemic and topical retinoids may be associated both with an increased risk of developing lung cancer and with mortality.28-30 The systemic retinoid, isotretinoin, has been associated with increased mortality in a large, randomized chemoprevention trial that involved patients with previously resected, stage I nonsmall cell carcinoma. This association, however, was based on a subgroup analysis and was restricted to current smokers.29 It also has been suggested that beta-carotene, a less closely related compound, increases the risk of lung cancer development as well as mortality in smokers.31 To date, there has been no large-scale trial involving systemic acitretin that has demonstrated a similar risk. In the current trial, we did not stratify patients according to smoking status, and the trial was not designed to assess mortality.
On the basis of available data, should acitretin be recommended as a chemopreventive agent in high-risk, nontransplantation patients? Although the current data support the finding that acitretin decreases total tumor burden, this benefit does not appear to correlate with a decrease in the incidence of or a clinically meaningful delay in new NMSC development. In addition, the toxicities from oral retinoids can be poorly tolerated. Thus, the risk-to-benefit for considering systemic therapy with acitretin for chemoprevention of NMSC may be greater in patients who have a higher burden of disease. Hopefully, more efficacious and less toxic alternative forms of chemopreventive medications will be developed for the nontransplantation population.
This work was supported by the US National Institutes of Health (grant CA-124477; principal investigator, Charles L. Loprinzi, MD) and by National Cancer Institute Community Clinical Oncology Program grant CA-37404.