Phase 2 and 3 Clinical Trial of Oral Bexarotene (Targretin Capsules) for the Treatment of Refractory or Persistent Early-Stage Cutaneous T-Cell Lymphoma

Objectives  To determine the safety and efficacy of oral bexarotene (Targretin capsules; Ligand Pharmaceuticals Incorporated, San Diego, Calif).

Design  The effects of 2 randomized doses of 6.5 mg/m² per day (with crossover for progression) vs 650 mg/m² per day (later modified to 300 mg/m² per day) were evaluated in an open-label, multicenter, phase 2 and 3 study conducted between February 1997 and November 1998.

Setting  Eighteen international cutaneous T-cell lymphoma clinics at academic referral centers.

Patients  Fifty-eight patients with biopsy-proven stage IA through IIA cutaneous T-cell lymphoma that was refractory to (or patients were intolerant of) treatment or had reached at least a 6-month response plateau under at least 2 forms of prior therapy (median of 3.5 prior therapies).

Intervention  Bexarotene (Targretin capsules) administered once daily with meal for 16 weeks or longer.

Main Outcome Measures  Primary end point classification of overall response rate of complete and partial remissions determined by either the Physician's Global Assessment of Clinical Condition or the objective Composite Assessment of Index Lesion Severity. Body surface area, time to response, duration of disease control, time to disease progression, individual index lesion signs and symptoms, and quality of life parameters were secondary outcomes.

Results  Responses (≥50% improvement) were seen in 3 (20%) of 15 patients with an initial dose at 6.5 mg/m² per day (95% confidence interval [CI], 0%-40%), 15 (54%) of 28 patients at 300 mg/m² per day (95% CI, 35%-72%), and 10 (67%) of 15 patients at above 300 mg/m² per day (95% CI, 43%-91%). The rate of progressive disease was 47%, 21%, and 13% at the same dose levels, respectively. Eight (73%) of 11 patients crossing over from 6.5 mg/m² per day to higher doses subsequently responded. The median duration of response from start of therapy could not be estimated for the 15 patients at 300 mg/m² per day owing to low relapse rates in 2 patients (13%); at higher doses it was 516 days. The following drug-related adverse effects were reversible and treatable: hypertriglyceridemia (46 patients [79%]), hypercholesterolemia (28 patients [48%]), headache (27 patients [47%]), central hypothyroidism (23 patients [40%]), asthenia (21 patients [36%]), and leukopenia (16 patients [28%]). No cases of drug-related neutropenic fever, sepsis, or death occurred. Pancreatitis occurred in 3 patients with triglyceride levels higher than 14.69 mmol/L (1300 mg/dL), all of whom were taking 300 mg/m² or more of oral bexarotene per day.

Conclusions  Bexarotene (Targretin capsules) (the first retinoid X receptor–selective rexinoid) was well tolerated and effective as an oral treatment for 15 (54%) of 28 patients with refractory or persistent early-stage cutaneous T-cell lymphoma at doses of 300 mg/m² per day. Hypertriglyceridemia and hypothyroidism require monitoring but are reversible and manageable with concomitant medication.

CUTANEOUS T-cell lymphomas (CTCLs) include extranodal, non-Hodgkin lymphomas arising from well-differentiated T-lymphocyte clones classified by clinical presentation and immunophenotype.^1,2 Mycosis fungoides (MF), with an incidence of 0.37 per 100 000 persons, is the most common form. ^3-6 Mycosis fungoides may gradually evolve from nonspecific lesions such as patches (eczematous), plaques (psoriasiform), follicular papules, hypopigmentation, or exfoliative erythroderma. "Parapsoriasis" is used to describe the condition of patients who have undiagnosed MF as well as that of patients with early MF that never progresses to lymphoma. The histologic pattern of MF mimics other inflammatory dermatoses.^4,5 Single-cell epidermotropism or lymphocytes along the basement membrane are more frequently encountered than the "diagnostic" Pautrier microabscesses.^4,5 Helper and memory skin-homing T lymphocytes expressing CD4 and CD45RO as well as CD8⁺ cytotoxic T cells are within the lesions.^1,6-10

An HLA-restricted, antigen-driven T-cell immune response is hypothesized to underlie the initial process with clonal expansion and accumulation via either loss of apoptosis and/or persistence of antigen stimulation.¹¹ The antigens HLA-DR5 (HLA-Dw11) and HLA-DQB1*03 are increased in MF, which is consistent with a CD4⁺-mediated response to antigen.^12,13 Superantigen stimulation by skin flora (especially Staphylococcus aureus in patients with erythroderma and/or implants), Chlamydia pneumoniae, leprosy, chemicals or immunogens, and viruses has been proposed.^14-21 Cytokines (interleukin [IL] 2, IL-7, IL-8, IL-10, IL-15, tumor necrosis factor [TNF] α, and interferon gamma) and chemokines (interferon-inducible protein 10) stimulate T-cell proliferation and recruit T cells to the epidermis.^22,23 The point at which inflammatory T cells become irreversibly malignant has not been defined by genetic mutations.

Survival is predicted by the stage of CTCL at the time of diagnosis.^1,6,24,25 Recent retrospective analyses report that survival among patients with stage IA disease is similar to that of age-matched controls, with decreasing survival with worsening T or skin stage.^1,25-27 Median survival for stage IV is only 1.5 years.²⁸ Large cell transformation within 2 years of diagnosis is associated with poor prognosis.²⁹

No therapy for MF demonstrates survival advantage.^1,30,31 Thus, skin-directed therapies are first used in early skin-limited disease, followed by systemic biological response modifiers, alone or in combination. ^1,30-36 Systemic chemotherapy is reserved for patients with more advanced stages of disease and can precipitate immunosuppression or fatal sepsis.^37,38 Common skin-directed therapies include topical corticosteroids, topical chemotherapy, and UV light or radiation (UV-B, psoralen–UV-A [PUVA], and/or electron beam). Few approved treatments exist for the MF variant of CTCL (MF/CTCL): extracorporeal photopheresis, systemic mechlorethamine hydrochloride, vinblastine sulfate, methotrexate, cyclophosphamide, and a new recombinant targeted fusion protein to the IL-2 receptor, denileukin diftitox (DAB₃₈₉IL2)(Ontak; Ligand Pharmaceuticals Incorporated, San Diego, Calif).^39-41 Topical chemotherapy, phototherapy, and radiation therapy, which are widely used for treating early-stage disease, may promote later development of skin cancer.

Interferon and retinoids, which are commonly used biological response modifiers for CTCL, have response rates around 50% or higher depending on the dose, short duration of response, and dose-limiting toxic effects. ^37,42-45 Retinoids, derived from or related to beta carotene, are steroid hormones and modulate gene expression via interactions with nuclear receptors and DNA transcription factors.^46,47 In human skin, both retinoic acid receptors (RARs) and retinoid X receptors (RXRs) are present. Each receptor has tissue-specific isoforms (α, β, and γ).^48,49 Retinoid X receptors can homodimerize or heterodimerize with RARs or other receptors, such as cholecalciferol, thyroid, and peroxisome proliferator–activated receptors.^50,51 Thus, RXR ligands (or rexinoids) have the potential to influence the transcription of a large number of genes.

Bexarotene [LGD1069] (Targretin capsules; Ligand Pharmaceuticals Incorporated) (an RXR-selective agonist⁵²) is the first rexinoid to be tested in human clinical trials. In a phase 1 study of patients with advanced cancers, therapy with bexarotene led to responses in 2 of 9 patients with CTCL and stabilization of tumor progression in non–small cell lung cancer and head and neck cancer.^53,54 Clinical trials specific to CTCL were initiated for the topical and oral forms of bexarotene. We report safety and efficacy data from the phase 2 and 3 clinical trial conducted in early-stage patients with refractory or persistent disease. Although the use of retinoids as single agents for the treatment of MF/CTCL is reported as far back as the early 1980s,^55-59 bexarotene is the first drug to receive approval for this indication.

A multicenter and multinational, open-label, historically controlled, phase 2 and 3 study of patients with refractory or persistent stage I through IIA MF/CTCL was conducted between February 1997 and the cutoff date of November 20, 1998. Eighteen academic referral centers in the United States, Canada, Australia, and Europe enrolled patients. Tolerability, safety, and antitumor efficacy were assessed at 2 randomly assigned dose levels: 6.5 mg/m² per day or 650 mg/m² per day. All patients signed informed consent and the study was conducted according to good clinical practice guidelines.

According to the inclusion criteria, patients had to be at least 18 years old with stage I through IIA CTCL that was documented within 30 days of study entry by skin biopsy results. The CTCL had to be refractory to therapy, or the patient was intolerant to therapy or had reached a 6-month or greater response plateau under at least 2 of the following qualifying prior therapies: 1 phototherapy (psoralen–UV-A or UV-B) or total body skin electron beam irradiation therapy or 1 topical chemotherapy (mechlorethamine [nitrogen mustard] or carmustine therapy). Interferon or systemic cytotoxic chemotherapy, but not retinoids or topical steroids, qualified entry. Disqualifying prior therapies were systemic antibiotic or topical therapy (for 2 weeks prior); phototherapy (for 3 weeks prior); systemic cancer therapy, electron beam, or other experimental therapy (for 30 days prior); etretinate therapy (for 1 year prior) and other oral retinoid therapies (for 3 months prior).

At baseline, patients had a complete physical examination and staging evaluation including chest radiograph, electrocardiogram, and blood evaluation for Sézary cells. A biopsy of clinically abnormal lymph nodes was recommended but not required. If clinical findings of more advanced CTCL emerged during the study, imaging studies were to be obtained.

Women of childbearing age were disqualified if they had positive results for a pregnancy test, and contraception was advised for both men and women. Electrolyte and chemical analysis findings, complete and differential blood cell counts, thyroid function test results with thyrotropin (TSH) levels, urinalysis results, and fasting triglyceride and cholesterol levels were acquired at baseline and during the study. Adverse events were reported. Following protocol amendment, ophthalmologic slitlamp examinations were performed at baseline and every 12 weeks.

Bexarotene (4-[1-(5,6,7,8-tetrahydro-3,5,5,8,8-pentamethyl-2-naphthalenyl)ethenyl]benzoic acid) was supplied by Ligand Pharmaceuticals Incorporated as a micronized formulation in 10-mg or 75-mg capsules. Capsules were self-administered once daily with the evening meal. Plasma levels of bexarotene were determined at single time points on day 1, week 2, week 4, and every 4 weeks thereafter during therapy.

Patients were randomly assigned to either 6.5 mg/m² per day (low dose) or 650 mg/m² per day (high dose) under the original protocol. Blinding was not possible because of the number of capsules given. The starting dose was reduced to 500 mg/m² per day and then to a final recommended initial dose of 300 mg/m² per day (optimal dose) based on response and dose-limiting toxic effects. Dose increases for efficacy (if no unacceptable toxic effect was occurring) and dose reductions (for toxic effect) were allowed. The dose of 6.5 mg/m² per day was not reduced. Patients who progressed after 8 weeks or failed to respond after 16 weeks could cross over to high-dose therapy.

Efficacy was based on the intent-to-treat population of all patients enrolled. No validated instruments for clinical assessment of CTCL exist. The Physician's Global Assessment of Clinical Condition (PGA) (Table 1) and the Composite Assessment of Index Lesion Severity (CA) grading scales were proposed as primary end points. Signs and symptoms used were erythema, scaling, plaque elevation, and hypopigmentation or hyperpigmentation and were graded on a scale of 0 to 8:

A scale of 0 to 18 was used to grade lesion size by square centimeter (0, 0 [no measurable area]; 1, >0 and ≤4; 2, >4 and ≤10; 3, >10 and ≤16; 4, >16 and ≤25; 5, >25 and ≤35; 6, >35 and ≤45; 7, >45 and ≤55; 8, >55 and ≤70; 9, >70 and ≤90; 10, >90 and ≤110; 11, >110 and ≤130; 12, >130 and ≤155; 13, >155 and ≤180; 14, >180 and ≤210; 15, >210 and ≤240; 16, >240 and ≤270; 17, >270 and ≤300; and 18, >300) The response rate was the number of patients classified with a clinical complete response (CCR) or partial response (PR) divided by the total number of patients in the study. Trained clinical evaluators assessed the same patients throughout the study.

The PGA was the investigator's subjective assessment of the overall improvement or worsening in disease compared with the baseline assessment (Table 1). A PR by PGA required that at least a 50% improvement be sustained for 2 or more study visits at least 4 weeks apart. Similarly, a CCR required the disappearance of all clinical disease activity.

The CA evaluated up to 5 representative index lesions using a 0- to 8-point scale for each clinical sign or symptom erythema, scaling, plaque elevation, and hypopigmentation or hyperpigmentation) and measurements of area (converted to a scale of 0-18) (Table 2).The physician was blinded to CA response because it was calculated from the case report form. The CA (the ratio of summation (Σ) of all clinical signs for these index lesions at each visit compared with baseline) included cutaneous tumors and all extracutaneous manifestations of disease. Complete clinical remission required a CA ratio of 0 with no evidence of disease (such as nodes, tumors, or visceral involvement). Partial remission was defined as (1) a CA ratio of 0.5 or lower without new clinically abnormal lymph nodes, (2) lower than 25% progression of existing clinically abnormal lymph nodes, (3) no new cutaneous tumors, and (4) no new pathologically positive lymph node or visceral disease in an area previously documented to be negative. Progressive disease was defined as (1) a 25% or higher increase in CA ratio, (2) a 25% or higher increase in the number or area of clinically abnormal lymph nodes, (3) new cutaneous tumor, or (4) new pathologically positive lymph node or visceral disease.

Primary end point classification was defined as the overall response rate according to either PGA or CA. A primary end point classification responder could not first progress by either PGA or CA prior to the confirmed response.

Secondary end points were body surface area (BSA) involvement, time to response, time to disease progression, duration of disease control, response of patients before and after crossing over to higher-dose therapy, and individual index lesion signs and symptoms including pruritus. The 6-item Spitzer quality of life questionnaire (validated for survivors in palliative care and hospice settings) was administered at baseline and each month to assess symptoms.⁶⁰ The first 5 items measure patients' activities, daily living, health, support, and outlook. Item 6 measures overall quality of life by a visual analog scale. Patients also answered a nonvalidated, CTCL-specific quality of life questionnaire. Photographs of index lesions were taken monthly. The addition of postbaseline histologic specimens to confirm response was recommended but not required.

Safety analyses were based on all patients exposed to at least 1 dose of bexarotene. Related events included "possibly related," "probably related," and "yes, related" adverse events. Adverse events were categorized as mild, moderate, moderately severe, and severe and were tabulated according to the Coding Symbols for Thesaurus of Adverse Reaction Terms (COSTART 5)⁶¹ dictionary. For a given patient, all adverse event terms mapping to the same COSTART 5 glossary term were grouped to 1 COSTART 5–preferred adverse event term.

Baseline variables including demographics, extent of disease, prior therapies, laboratory values, symptoms, and signs of disease were summarized for each dose group. A successful clinical trial was defined as a response rate (CCR + PR) of at least 20%, with the lower boundary of the 95% confidence interval [CI] excluding the conservative estimate of a theoretical maximal spontaneous response rate of 5%.

Fifty-eight patients enrolled at 18 international CTCL clinics at academic referral centers were included in an intent-to-treat analysis of safety and efficacy. Seven of 65 patients screened did not meet inclusion/exclusion criteria and did not enroll. In addition, 1 patient with small cutaneous tumors enrolled and was considered a protocol violation.

The demographics in Table 3 reflect known disease characteristics.⁶² Baseline variables were balanced among the 3 initial dose groups as determined by χ² statistics (categorical variables) and 1-way analysis of variance (continuous variables). Of the 58 patients at entry 17 (29%) had stage IA disease (<10% BSA involvement of patches and/or plaques [T1]), 34 (59%) had stage IB disease (≥10% BSA involvement without nodes [T2]), and 6 (10%) had stage IIA disease (either T1 or T2 with clinical adenopathy and insignificant node pathologic characteristics). The median time from first clinical manifestation to entry was 10.5 years and from diagnosis to entry, 6.6 years.

Fifteen centers enrolled at least 1 patient at an optimal starting dose of 300 mg/m² per day. The percentage of BSA involvement was slightly higher in this dose group. At least 1 response was seen at each of the 7 centers enrolling more than 1 patient.

Prior therapies are shown in Table 4. Of the 58 patients, 45 (78%) were refractory to 2 or more therapies; 10 (17%) were refractory to 1 and intolerant to at least 1 prior therapy for CTCL. The optimal dosing group received a median of 3 prior therapies (range 2-8), and 18% had 6 prior therapies or more. Irradiation or phototherapy (56 patients [97%]) and topical nitrogen mustard (54 patients [93%]) were common and 34 patients (59%) had received prior systemic therapies, such as interferon (22 patients [38%]). Six patients (11%) had received combination chemotherapy, systemic corticosteroids, isotretinoin, or the recombinant IL-2 diphtheria fusion protein, Ontak.^40,63

Analysis findings of response rate and toxic effects are given for 3 initial dose arms (Figure 1). Although patients were to be randomized to either low dose (6.5 mg/m² per day) or high dose (650 mg/m² per day, then 500 mg/m² per day), the final optimal starting dose was adjusted to 300 mg/m² per day and was received by 28 patients. Fifteen patients were randomized to 6.5 mg/m² per day, and 15 received initial doses above 300 mg/m² per day. An interim analysis in September 1997 showed a significant difference between the rate of progressive disease in patients receiving 6.5 mg/m² per day (47%) (5 [71%] of 7 patients) vs those on high-dose therapy (19%) (2 [7%] of 29 patients) (P = .045, Fisher exact test). At doses of 6.5 mg/m² per day, 0 of 7 patients had responded compared with 12 (41%) of 29 patients at doses above 300 mg/m² per day. Thus, randomization was discontinued. The duration of treatment was planned for at least 16 weeks, but was extended if treatment was beneficial and no unacceptable toxic effects were occurring.

The study was not powered to compare responses based on dose, but rather for point estimates and 95% CIs for each treatment group. Overall response rates (PR + CCR) for each starting dose as determined by the protocol-defined end points are shown (Figure 1). Three (20%) of 15 patients with starting doses of 6.5 mg/m² per day (95% CI, 0%-40%), 15 (54%) of 28 patients with starting doses of 300 mg/m² per day (95% CI, 35%-72%), and 10 (67%) of 15 patients with starting doses above 300 mg/m² per day (95% CI, 43%-91%) met response criteria by primary end point classification. Clinical complete responses occurred in 2 (7%) of 28 patients starting at 300 mg/m² per day and in 4 (27%) of 15 patients starting either 500 or 650 mg/m² per day. At the optimal dose, responses were similar by stage: stage IA (2 [50%] of 4 patients), stage IB (11 [52%] of 22 patients), and stage IIA (2 [67%] of 3 patients). There was no evidence of preferential response by age, sex, or race.

Eleven of 15 patients receiving low-dose therapy progressed or did not improve and crossed over to high-dose therapy. The response rate (CCR + PR) was 18% (2 of 11 patients) prior to crossover compared with 73% (8 of 11 patients) after. Incidence of progressive disease was initially 64% compared with 18% after crossover.

The median total BSA involvement for patients remaining in the study improved over time (Figure 2). The change in BSA involvement for an 81-year-old man with stage IIA patch and/or plaque MF involving 48% of the trunk is shown in Figure 3A and his appearance is shown in Figure 3B. Although he had received standard therapies and combination chemotherapy, he experienced a complete response that has been ongoing for more than 3 years (Figure 3 and Figure 4).

Four of 7 CCR patients had paired biopsy speciments taken from baseline and clinically resolved lesions: the results of 2 biopsies showed no evidence of CTCL and 2 were equivocal. Clinically palpable nodes resolved in 2 of these patients at 300 mg/m² per day, but there was no change in median aggregate lymph node area based on the 6 patients enrolled who had adenopathy. Four of 6 patients had circulating abnormal lymphocytes, and in the patient with the highest counts (14.2% and 20.1%), all abnormal cells resolved by week 12.

The Kaplan-Meier method⁶⁴ was used to calculate the likelihood of the event at any given time for the patients remaining in the study with the results given in Table 5. The median time to onset of response (ie, at least 50% improvement) was 8.1 weeks (range, 3.9-16.3 weeks) at optimal doses vs 13.1 weeks at high doses where pauses in therapy occurred. The median time to progressive disease (by primary end point classification) was 13.6 vs 30.0 weeks at 6.5 vs 300 mg/m² per day initial dosing, respectively, and 73.7 weeks at above 300 mg/m² per day. The duration of disease control for responders was calculated from onset of the patient's relapse (or last date of clinical evaluation documenting continuation of response) minus the date of the first day in the study (Table 5). None of the 3 responders at low initial dose relapsed (data censored after crossover), and only 2 (13%) of 15 responding patients at optimal dose relapsed. Five of 10 responding patients receiving doses above 300 mg/m² per day had relapsed, indicating a projected median of 73.7 weeks. In summary, the rate of relapse following response to bexarotene was low, and the responses were durable.

Index lesion pruritus consistently decreased from mild to moderate at baseline compared with mild to absent by week 16. Pruritus continued to improve during subsequent weeks independent of antihistamine and/or antipruritic use (data not shown). Patients' status was rated quite high at baseline by the Spitzer questionnaire⁶⁰ with little room for improvement. Among primary end point classification responders, 18 (95%) of 19 patients rated their change in CTCL as moderately or much improved, and 16 (84%) of 19 patients were moderately to very satisfied with bexarotene treatment at week 16 (Figure 5).

Single time point plasma samples were collected approximately 12 to 24 hours after bexarotene administration. All doses except that of 6.5 mg/m² per day resulted in measurable plasma bexarotene levels; concentrations were dose dependent. There was no evidence of long-term bexarotene accumulation for therapy up to 64 weeks. Because of dose adjustments, the plasma bexarotene concentrations were normalized for absolute dose prior to assessing the impact of coadministered medications. Concurrent administration of gemfibrozil (Lopid; Parke-Davis, Morris Plains, NJ), but not atorvastatin or levothyroxine, was associated with a higher mean dose-normalized concentration of bexarotene (data not shown). Previous studies have shown that area under the plasma bexarotene concentration–time curves were proportional to doses at the dose administered for this study.^53,54

Although the duration of therapy was to be at least 16 weeks, 11 (19%) of the 58 patients were treated for 40 weeks or more and 16 patients (28%) remained on drug therapy at the time of data cutoff. The most common reasons for withdrawal were adverse events (18 [43%] of 42 withdrawals), 15 of which were attributed to bexarotene, and either progressive disease or withdrawal of consent (6 patients [14%] each). Of 28 patients starting treatment at 300 mg/m² per day, 7 withdrew because of adverse events and 3 withdrew because of disease progression.

There were no deaths during the trial or within 4 weeks of discontinuing therapy. Three deaths within 3 months of therapy discontinuation were attributed to infection or progressive disease. Three cases of acute pancreatitis occurred in association with very high triglyceride levels (>14.7 mmol/L [1300 mg/dL]) in 1 patient receiving a dose of 300 mg/m² per day and 2 patients receiving doses above 300 mg/m² per day. All patients recovered, and no cases of pancreatitis occurred following an institution of strict monitoring and antilipid management. The squamous cell carcinoma that occurred in 4 (7%) of the 58 enrolled patients was attributed to prior therapy.

Adverse events occurring with a 10% or more incidence, regardless of relatedness to the study drug, are given in Table 6. One or more adverse events (mostly mild to moderate severity) occurred in 57 (98%) of the 58 study patients. Dose-limiting toxic effects were present in 41 (71%) of the 58 patients with incidence related to dose. A higher number of moderately severe events occurred at doses above 300 mg/m² per day, but severe events were rare: diarrhea (2 patients) and desquamation (1 patient).

Adverse events with overall incidence of 25% or more that were at least possibly related to bexarotene are shown in Figure 6. Pruritus was recorded as an adverse event in 15 (26%) of the 58 patients, but its relation to drug vs disease condition could not be determined.

Hyperlipidemia occurred in 46 (79%) of the 58 patients and was considered dose limiting in 25 patients (43%). Rises in lipids were rapid (within 1 to 2 weeks). Twenty patients had peak triglyceride level elevations of higher than 3.4 to 4.5 mmol/L (300 to ≤400 mg/dL); 9 in the range of higher than 4.5 to 5.65 mmol/L (400 to ≤500 mg/dL); and 8 with levels higher than 5.65 mmol/L (500 mg/dL). Hypercholesterolemia was present in 20 (71%) of 28 patients and 10 (67%) of 15 patients beginning therapy at 300 and above 300 mg/m² per day, respectively. For all dose groups, the median values were 8.1 mmol/L (314 mg/dL) for cholesterol and 5.3 mmol/L (472 mg/dL) for triglyceride levels at 2 to 4 weeks. With antilipid therapy, levels remained elevated at 6.9 and 3.7 mmol/L (267 and 333 mg/dL) for cholesterol and triglyceride levels at 12 to 16 weeks, respectively.

In the 58 patients, other adverse events related to bexarotene included headache (27 patients [47%]), hypothyroidism (23 patients [40%]), asthenia (21 patients [36%]), and leukopenia (16 patients [28%]) (Figure 6). Leukopenia was dose related, occurring as early as 2 to 4 weeks, and was due to a decrease in neutrophils rather than lymphocytes. Neutropenia occurred in 9 (38%) of 24 patients (only 24 had paired samples) at 300 mg/m² per day, required dose reduction in 2 patients (10%), and growth factor support in 1 patient only. There were no reported events of related neutropenic fever or sepsis, indicating that the neutropenia was not associated with fever or septic sequelae.

Oral bexarotene therapy was associated with alteration in thyroid axis function as previously reported.⁶⁵ Levels of TSH fell from 1.63 mIU/L (n = 55) median at baseline to 0.36 mIU/L (n = 22) at more than 2 to 4 weeks. Although 2 (4%) of 58 patients had low TSH values at baseline, 28 (48%) and 25 (43%) had abnormally low values during the 2 postbaseline periods. Changes in total thyroxine levels generally corresponded to those of TSH. Twenty-four (41%) of 58 patients were administered levothyroxine at some time during the study based on their symptoms or fall in levothyroxine levels. Symptoms of hypothyroidism included asthenia, fatigue, cold intolerance, and constipation and were reversed with thyroid hormone replacement or within a week of discontinuing bexarotene therapy.

Elevation of hepatic function test results (aspartate transaminase or alanine transaminase levels in the range of >2.5 to ≤5.0 × the upper limit of normal) were reported as a related adverse event in 6 (10%) of 58 patients. Transaminase level elevation was dose related with the highest incidence and severity in the 300 mg/m² per day initial dose group.

After the onset of this trial, cataract formation in dogs and rats treated with high doses of bexarotene occurred. Slitlamp eye examinations were performed in 31 (53%) of the 58 patients at baseline and in 54 patients (93%) at least once. The baseline incidence of lens opacities in at least 1 eye was high at 65% (20 of 31 patients). In only 2 of 18 patients, a new lens opacity occurred (study day 94) or possible subtle worsening of bilateral cataracts was noted. The incidence of lens opacities was sporadic and not related to dose or duration of treatment; loss of visual acuity was not reported.

Although rare, CTCLs are disfiguring and symptomatic. Early diagnosis of MF is elusive, and once made, no curative therapy currently exists. Mycosis fungoides may have an indolent course with long-standing skin involvement or may progress to aggressive systemic and lethal lymphoma.^{1,25,26,28,30} Only a few of the available therapies for MF and CTCL are approved by the Food and Drug Administration (FDA); most have cumulative toxic effects over time.^33,34,63 A randomized, National Cancer Institute–sponsored study conducted in the 1980s demonstrated that multidrug chemotherapy was no better than sequential conservative topical therapy in prolonging overall survival, despite a slightly higher initial response rate.³¹

Patients with early-stage MF who either fail to respond to or relapse after skin-directed therapies require administration of systemic agents, and biological response modifiers (cytokines or retinoids) are favored for disease control.^{1,30,32-34,66,67} Bexarotene is the first of a new class of agents (the rexinoids) that are selective for RXRs. As with retinoids, bexarotene is contraindicated in pregnant women or in women planning pregnancy. Bexarotene is effective in patients who failed to respond to or had been treated with other oral retinoids.⁶⁸ Bexatotene's mechanism of action in CTCL is not known, although RXR agonists have been shown to modulate growth and differentiation and apoptosis in a variety of cellular systems ^69-71 and to interact with intracellular signaling pathways and transcription factors.^47,72

Isotretinoin and etretinate are active in MF/CTCL with responses in the range of 50% to 65%.^55-57,73-75 In our study, patients with refractory persistent stage IA through IIA disease had response rates of 67%, 54%, and 20% at the starting doses of 500 to 650, 300, or 6.5 mg/m² per day, respectively. Although higher rates of partial and complete response and significantly lower rates of progressive disease related to the dose level occurred, the study was not powered to determine significant differences between the groups. Although other retinoids may seem to have similar response rates to bexarotene in the published reports, the prior studies were not conducted as rigidly controlled clinical trials and did not use the same response criteria. Patients were not strictly limited to refractory or persistent disease.

Retinoids often have been used in clinical practice in combination with other agents effective in treating MF/CTCL, including PUVA,⁷⁶ interferon,⁷⁷ and combined-modality therapies.^36,73 Individual patients have received bexarotene therapy in combination with other therapies (ie, UV light, photopheresis, interferon alfa, nitrogen mustard), either with or following Ontak treatment with good results, but formal trials have not been conducted (M.D., unpublished observations, 2001).

Toxic effects and adverse events were more common and more severe among patients taking the highest doses. Of concern, lipid levels may rise rapidly within 1 to 2 weeks of starting bexarotene therapy, precipitating pancreatitis. Only 3 patients in this trial had pancreatitis, and all had triglyceride levels above 14.7 mmol/L (1300 mg/dL). For safety reasons, the 300 mg/m² per day initial dose of bexarotene is now recommended. The clinician should screen for a history of diabetes, lipid abnormalities, and/or pancreatitis and should measure fasting lipid levels prior to starting therapy. Patients with high fasting triglyceride levels should undergo therapy with lipid-lowering agents (eg, atorvastatin or fenofibrate) prior to taking Targretin capsules at full doses. Fasting cholesterol and triglyceride levels need to be measured weekly until the lipid levels stabilize. Monitoring should then occur every 1 to 2 months while the patient remains under therapy. The concomitant use of gemfibrozil (Lopid) and bexarotene is not recommended based on the results from pharmacokinetic studies. One should also be aware of potential interactions between hydroxymethyl glutaryl coenzyme A reductase inhibitors (eg, atorvastatin, lovastatin, pravastatin, and simvastatin) and fibric acid derivatives (eg, fenofibrate) or niacin used in combination. Although a few patients undergoing therapy with Targretin capsules received concurrent atorvastatin and a fibric acid derivative to control triglyceride levels, this combination has potential risks (such as severe myopathy, rhabdomyolysis, and acute renal failure), which may outweigh benefits.

In our trial, patients who were randomized to 6.5 mg/m² per day initially responded later to higher doses. Thus, initiation of low-dose therapy and subsequent gradual increases to maximize response and minimize the adverse effects is an acceptable alternative for future studies. Oral bexarotene (0.5-2.0 mg/kg per day) is under investigation for treating plaque psoriasis with mild hypertriglyceridemia reported in 57% of subjects.⁷⁸

With the exception of hyperlipidemia and headache, the adverse effects of rexinoids are distinct from RAR agonist symptoms (cheilitis, alopecia, arthritis, and myalgias). Transient headache or peeling may occur when bexarotene therapy is initiated. Patients receiving bexarotene quickly develop reversible central hypothyroidism manifested by suppressed TSH production and low thyroxine values.⁶⁵ Hypothyroidism (defined as requiring thyroid hormone replacement) occurred in 23 (40%) of the 58 patients with CTCL. This is a far higher incidence than observed in other cancer trials.^53,54 The incidence of hypothyroidism may be higher if the symptoms are specifically solicited or if hypothyroidism is defined by laboratory measurements. Signs and symptoms of hypothyroidism are quite subtle. We found that monitoring free thyroxine levels (but not TSH levels) over time is useful for instituting thyroid hormone replacement therapy, and patients felt better if their thyroid hormone was supplemented.⁶⁵ Hypothyroidism and hyperlipidemia are fully and rapidly reversible within a week or 2 of drug therapy cessation. The mechanism of bexarotene-induced hyperlipidemia is not fully understood,^79,80 but hypothyroidism is known to decrease lipid clearance and may contribute to the hyperlipidemia if not corrected.

Leukopenia occurred infrequently at only the highest doses. One third of patients had decreases in total leukocyte or neutrophil count below the lower level of normal within 4 weeks of starting therapy. Only 2 patients received growth factor support therapy. No drug-related deaths, neutropenic fever, or sepsis occurred, underlying the advantage of bexarotene oral therapy. The skin of patients with MF is often colonized with S aureus, which poses great risk for sepsis through catheters.^14,81,82 Bexarotene's profile differs from systemic chemotherapy, which may be accompanied by myelosuppression, neutropenia, and sepsis. Chemotherapy is rarely, if ever, indicated for early-stage patients with favorable prognosis.

In stages IA through IIA CTCL, bexarotene monotherapy demonstrated a 54% to 67% response rate at doses of 300 mg/m² or higher per day occurring within the first 2 months in most patients. The FDA approved bexarotene on December 29, 1999, as an oral therapy for refractory CTCL in all stages⁶⁸ and again in June 2000 as a gel formulation for cutaneous manifestations of early-stage refractory or persistent CTCL (D. Breneman, M. Duvic, T. Kuzel, R. Yocum, J. Truglia, V. Stevens, unpublished data). Oral bexarotene treatment is effective in patients whose degree of skin disease or whose failure to respond to skin-directed therapy supports a need for systemic therapy. Bexarotene's profile of adverse effects (hyperlipemia, hypothyroidism, and less commonly, leukopenia) is mild, does not overlap that of other commonly used CTCL treatments, and suggests the feasibility of use in combination with other agents for enhanced efficacy and disease control in future studies. For patients with MF (stages 1-IIA), bexarotene is an effective drug with durable responses, a favorable safety and tolerability profile, and adverse effects that are easily monitored, reversible, and manageable with the use of concomitant medication.

Accepted for publication on January 1, 2001.

This study was funded in part by clinical research grants from Ligand Pharmaceuticals Incorporated, San Diego, Calif, and in part by grants R21-CA74117 (Dr Duvic) and CA16672-22 (M. D. Anderson Cancer Center) from the National Cancer Institute, Bethesda, Md.

Presented in part at the American Society of Hematology (ASH) meeting, New Orleans, La, December 6, 1999.

Members of the Worldwide Bexarotene Study Group include John Aeling, MD, University of Colorado Health Sciences Center, Denver; Martine Bagot, MD, Hôpital Henri Mondor, Creteil, France; Philippe Bernard, MD, Hôpital Robert Debré, Reims, France; Eduardo López Bran, MD, Hospital Universitario San Carlos, Madrid, Spain; Debra Breneman, MD, University of Cincinnati, Cincinnati, Ohio; Gunter Burg, MD, University Hospital Zurich, Zurich, Switzerland; Luis Iglesias Díez, MD, Hospital Doce de Octubre, Madrid, Spain; Madeleine Duvic, MD, M. D. Anderson Cancer Center, Houston, Tex; Rokea el-Azhary, MD, Mayo Clinic, Rochester, Minn; David Fivenson, MD, Henry Ford Hospital, Detroit, Mich; Francine Foss, MD, New England Medical Center, Boston, Mass; Peter Heald, MD, Yale University School of Medicine, New Haven, Conn; Pascal Joly, MD, Hôpital Charles Nicolle, Rouen, France; Youn Kim, MD, Stanford University School of Medicine, Stanford, Calif; Ann G. Martin, MD, Washington University, St Louis, Mo; Marilyn Mehlmauer, MD, Pasadena, Calif; Wilson Miller, MD, Jewish General Hospital, Montreal, Quebec; Jennie Muglia, MD, Rhode Island Hospital, Providence; Patricia Myskowski, MD, Memorial Sloan-Kettering, New York, NY; Elise Olsen, MD, Duke University Medical Center, Durham, NC; Anna Pluzanska, MD, Klinika Chemioterapii Katedry, Lodz, Poland; H. Miles Prince, Peter MacCallum Cancer Institute, East Melbourne, Australia; Jadwiga Rozkiewicz, MD, Katedra i Klinika Chorob Skory i, Gdansk, Poland; Michael D. Tharp, MD, Rush-Presbyterian-St Luke's Medical Center, Chicago, Ill; Bruce Thiers, MD, Medical University of South Carolina, Charleston; Ken Washenik, MD, NYU Medical Center, New York; Gary Wood, MD, University Hospital of Cleveland, Case Western Reserve University, Cleveland, Ohio; and John Zone, MD, University of Utah, Salt Lake City.

The authors would like to thank Victor Stevens, PhD, and Carol Manifold, PharmD (Ligand Pharmaceuticals Incorporated), the study coordinators, and the research fellows participating in this study. The authors wish to acknowledge the help of Gordon R. Loewen, PhD, and Allen Cato III, PhD (Ligand Pharmaceuticals Incorporated) for the pharmacokinetic data and Robert H. Knopp, MD (Department of Medicine, Washington University) for helpful suggestions.

Corresponding author and reprints: Madeleine Duvic, MD, M. D. Anderson Cancer Center, Box 434, 1515 Holcombe Blvd, Houston, TX 77030 (e-mail: mduvic@mail.mdanderson.org).