Truth Seeker
New Member
Abstract
Purpose To compare the effects of cannabis extract (CE), delta-9-tetrahydrocannabinol (THC), and placebo (PL) on appetite and quality of life (QOL) in patients with cancer-related anorexia-cachexia syndrome (CACS).
Patients and Methods Adult patients with advanced cancer, CACS, weight loss (≥ 5% over 6 months), and Eastern Cooperative Oncology Group (ECOG) performance status (PS) ≤ 2 were randomly assigned (2
1) to receive CE (standardized for 2.5 mg THC and 1 mg cannabidiol) or THC (2.5 mg) or PL orally, twice daily for 6 weeks. Appetite, mood, and nausea were monitored daily with a visual analog scale (VAS); QOL was assessed with the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire C30 (composite score: questions 29 and 30). Cannabinoid-related toxicity was assessed every 2 weeks.
Results Of 289 patients screened, 243 were randomly assigned and 164 (CE, 66 of 95 patients; THC, 65 of 100 patients; and PL, 33 of 48 patients) completed treatment. At baseline, groups were comparable for age (mean, 61 years), sex (54% men), weight loss (32% ≥ 10%), PS (13% ECOG = 2), antineoplastic treatment (50%), appetite (mean VAS score, 31/100 mm), and QOL (mean score, 30/100). Intent-to-treat analysis showed no significant differences between the three arms for appetite, QOL, or cannabinoid-related toxicity. Increased appetite was reported by 73%, 58%, and 69% of patients receiving CE, THC, or PL, respectively. An independent data review board recommended termination of recruitment because of insufficient differences between study arms.
Conclusion CE at the oral dose administered was well tolerated by these patients with CACS. No differences in patients' appetite or QOL were found either between CE, THC, and PL or between CE and THC at the dosages investigated.
INTRODUCTION
Anorexia and weight loss contribute to cancer-related fatigue, functional loss, impaired survival, and intolerance of treatment.1,2 Efforts to palliate these conditions include studies of the endocannabinoid system, which modulates appetite3 through cannabinoid receptor—related processes.4 Hyperphagic effects of cannabinoids5 and hypophagic actions of selective cannabinoid receptor antagonists6 have been reported.
Cannabis sativa contains over 60 cannabinoids, including delta-9-tetrahydrocannabinol (dronabinol, THC) and cannabidiol (CBD).7 THC, its major psychoactive component, is an antiemetic medication.8 CBD reportedly reduces the psychotropic effects of THC and has anti-inflammatory effects.9
Cannabinoids reputedly stimulate appetite, both historically10 and in recent studies of human volunteers11-13 and AIDS patients.14 Studies of patients with multiple sclerosis15-17 or pain18 have evaluated oral mixtures of THC and CBD or whole-plant cannabis extract (CE), replacing smoked marijuana. Data from four dose-finding and phase II studies of 161 patients with cancer-related anorexia-cachexia syndrome (CACS) suggest cannabinoids' potential at fixed doses of 2.5 mg of THC twice to three times daily;19-21 however, megestrol acetate palliated anorexia better than THC.22 We investigated the effects of CE and THC on appetite and quality of life (QOL) in patients with CACS.
PATIENTS AND METHODS
This multicenter, phase III, randomized, double-blind, placebo (PL)-controlled, three-arm, parallel study, which was sponsored by the Institute for Clinical Research, Berlin, adhered to the Good Clinical Practice Program,23 European Union Directive 91/507/EEC,24 and the Declaration of Helsinki.25 Participating centers obtained approval from their local institutional ethical review boards and health authorities.
Participants
Physicians recruited adult patients with advanced incurable cancer who were candidates for appetite stimulation, having had, within the past 6 months, involuntary weight loss of ≥ 5% not explained by other diseases or recent surgery. Eligible patients gave written informed consent to participate; could feed themselves; received no enteral or parenteral nutrition; had taken no anabolic agents, gestagens, cannabinoids, or corticosteroids (except for < 20 mg prednisolone for < 5 consecutive days) within the past month; and had no significant cause of secondary anorexia2 or psychiatric disorder (substance abuse or schizophrenia). Patients had an estimated life expectancy of 3 months; an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of ≤ 2; creatinine, bilirubin, and transaminase values no higher than 3× the maximum normal value; and unchanged antineoplastic therapy for 4 weeks and unchanged supportive treatment (analgesics, sedatives, tranquilizers, and anticholinergics) for 1 week before baseline assessments.
Intervention
After 7 to 14 days of baseline assessment, eligible patients were randomly assigned to treatment with PL, CE, or THC for 6 weeks. Patients received a 2-week supply of capsules to take orally twice daily (1 hour before lunch and before dinner or at bedtime, preferably with milk).
All C sativa plants were harvested at once in Switzerland and processed into a fluid extract (Verein-für-Krebsforschung, Arlesheim, Switzerland) to prepare gelatin capsules containing CE (2.5 mg THC and 1 mg CBD) or only THC (2.5 mg). PL capsules containing standardization medium (Hüls-AG, Marl, Germany) were indistinguishable from the active capsules. Encapsulation was uniformly processed by Scherer-GmbH (Eberbach, Germany). The Unfallkrankenhaus (Berlin, Germany) packed the drug.
Objectives
We tested the hypotheses that appetite and QOL improve significantly in patients treated with THC or whole-plant CE compared with PL26 and that CE and THC have equivalent effects on appetite and QOL.
Outcome Measures
Assessments were performed every 2 weeks during clinic visits at screening, at random assignment when treatment began (week 0), and at weeks 2, 4, and 6; patients kept a diary. At each visit, patients underwent an examination evaluating vital signs and ECOG PS and a test for urinary cannabinoids.
A primary end point was appetite change from baseline to week 6, assessed with a visual analog scale (VAS; 0 mm = worst; 100 mm = best). Appetite values were calculated as the mean of daily appetite VAS scores for the 7 days of week 2 in each biweekly period. The second primary end point was the change in QOL from baseline to week 6 (composite score [mean] of questions 29 [Global Health Status] and 30 [QOL] on the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire C30 [EORTC QLQ-C30], transformed into a single functional scale [range, 0 to 100]).27
The EORTC QLQ-C3027 was completed biweekly. Three other variables were monitored daily with a VAS: (1) patients' estimation of food intake during the previous 24 hours; and patients' current (2) feeling of nausea and (3) mood.
Two ad hoc EORTC QLQ-C30 modules, Anorexia-Cachexia and Cannabinoid Toxicity (CannTox), were completed biweekly. The first was translated from the Functional Assessment of Anorexia/Cachexia Therapy.28 Its internal consistency (Cronbach's α = .76) and test-retest reliability (Pearson correlation between screening and randomization, r = 0.68) were good, as was the correlation (Spearman partial rank) of the total score with the EORTC QLQ-C30 question about appetite loss (r = −0.59; P < .0001) and with VAS scores for appetite (0.36 mm; P < .0001). The CannTox module was based on the Drug Reaction Scale of adjectives describing mood, physical feelings, and perceptions of mental or cognitive functions of healthy volunteers under the influence of cannabis.29 Patients complied with study protocol if they took at least 90% of the prescribed medication over the study period, as based on the returned empty capsule blisters.
Adverse events (AEs) were classified according to Common Toxicity Criteria.30 CBCs and biochemical and urine profile analyses were performed at each visit. In cases of study drug—related AEs, the dose was reduced to one capsule daily or was interrupted until the AEs resolved, at which point patients again took two capsules or resumed treatment with one capsule daily at bedtime. Serious AEs were evaluated by the Safety Review Committee (F.S., M.S., and T.C.).
Sample Size
Sample size was calculated with reference to a study of THC used to treat AIDS-related anorexia,14 which showed a mean increase in appetite of 15 mm from baseline (standard deviation [SD], 15 mm) on a VAS in the verum group and 7.4 mm in the PL group. Half of SD may be considered clinically significant.31
To compare CE and THC with PL, a two-tailed test of differences from baseline to week 6 in appetite score was planned (significance level, 5%; power, 90%; expected between-group difference, 7 mm). For the noninferiority test of CE against THC, a one-tailed test was planned (significance level, 2.5%; power, 90%; maximum lower deviation of CE, 6 mm). For both tests, an SD of 15 mm in all three groups was assumed. The required sample was 133 patients per verum group and 68 patients in the PL group (treatment allocation in a 12 ratio). Assuming a dropout rate of 25%, the total sample was increased to 445 patients.
After a blinded interim analysis of 46 patients to assess data variability, the analysis plan was adjusted to a group sequential, adaptive trial with the permission of each center's ethical review board. An independent data review board recommended that the trial be closed after the first unblinded interim analysis of fully monitored data from 156 assessable patients because of insufficient differences in the primary end point between the PL and verum arms; recruitment stopped with 289 patients screened.
Treatment Assignment, Random Assignment, and Blinding
Patients were randomly assigned at the week 0 visit, after the baseline period. Random assignment lists, stratified by center, were prepared by a naive statistician using SAS software (SAS Institute, Cary, NC).32 Centers received study drug sets in multiples of five, together with matching sealed envelopes containing individual treatment assignments. Investigators remained blinded until the study ended, with individual unblinding permitted only for safety reasons. The statistician and data manager who were managing random assignment, unblinding, and related decisions were naive to clinical evaluations and uninvolved in data management or analysis.
Statistical Analyses
All analyses were performed with SAS.32 One interim analysis was performed for the independent data review board (156 patients) and one for presentation (216 patients).33
Descriptive statistics were used for demographic and baseline variables. For efficacy analyses, the groups of patients enrolled before the first interim analysis, in between the two analyses, and after the second interim analysis were treated as independent samples of the whole population34,35; the three resulting test statistics were properly combined, holding an overall one-sided significance level of P = .025 partitioned into P = .00025, P = .00691, and P = .02221, respectively, regarding the three successive analyses.
The confirmatory test for superiority of any verum arm over PL was based on the intent-to-treat (ITT) population. For confirmation of noninferiority of CE to THC, the per-protocol (PP) population was used.
An a priori ordered cascade of statistical testing of six null hypotheses was applied36 in each interim and final analysis as follows: H01-3 for appetite, CE versus PL, then THC versus PL, then CE versus THC; and H04-6 for QOL in the same sequence. The significance of each statistical test could be deduced confirmatively only if the actual and all higher order tests had yielded P values smaller than the actual critical significance level.
For the ITT population, missing values were substituted by the nearest-neighbor approach.37 Sensitivity analysis additionally analyzed all efficacy parameters by the last observation carried forward method and by the worst-case approach.38
Verum and PL groups were compared using analysis of covariance, including treatment as a fixed factor and the VAS or QOL baseline values as continuous covariates. Center effects were ignored because of the many centers with only a few patients.39 Additionally, one-sided 97.5% CIs for differences were given.
All successive tests were considered exploratory. For mood, nausea, daily food intake, and EORTC QLQ-C30 items, changes from baseline to weeks 2, 4, and 6 were analyzed analogous to the primary end points. Additional exploratory analyses were performed in subgroups based on preliminary evidence of distinguishable CACS with high versus low C-reactive protein (CRP), in different tumor types, by sex, and by degree of weight loss.1,2
For all safety parameters, descriptive statistics over time was performed. P values were calculated as flagging devices to mark conspicuous, potentially clinically relevant differences. Treatment groups were compared regarding hazard rates of AEs by Cochran-Mantel-Haenszel χ2 tests and regarding continuous data by analysis of variance. The proportions of patients not completing the study were compared between treatment groups by the χ2 test.
RESULTS
Patients
Patients were recruited from October 1999 to September 2002 in 30 centers in Germany, Switzerland, and the Netherlands. Average weight loss was 11.9%; 46% of patients lost less than 10% of weight. Patient characteristics (Table 1) and the percentage of dropouts and the reasons for dropout (Fig 1) were similar between treatment arms. Compliance with treatment was similar between arms (CE, 49%; THC, 44%; and PL, 60%; P > .15); 84 patients had major protocol violations (16, 22, and nine patients on CE, THC, and PL, respectively) and/or less than 90% intake of the study medication (14, 10, and four patients on CE, THC, and PL, respectively), were missing primary end point data (three, one, and three patients on CE, THC, and PL, respectively), had THC in their serum at baseline (zero, three, and zero patients on CE, THC, and PL, respectively), or had other protocol deviations (four, three, and one patient on CE, THC, and PL, respectively). The PP analysis set comprised 80 patients (33%; 32, 31, and 17 patients on CE, THC, and PL, respectively), who were mostly men and older and had better baseline values for mood, nausea, daily food intake, and EORTC QLQ-C30 scores.
Appetite and Overall QOL
Pooled over all samples of the adaptive design, patients taking CE and PL, after 6 weeks, showed a similar mean improvement in appetite (mean, 5.4 mm; SD, 24.7 mm; and mean, 5.8 mm; SD, 23.8 mm, respectively), whereas patients taking THC showed only a 0.6-mm improvement (SD, 18.5 mm; Fig 2). No verum arms improved significantly compared with PL (combined test statistics: P = .46 for comparison with CE; P = .95 for comparison with THC). The correlations of VAS appetite scores with the EORTC QLQ-C30 categoric scale of Appetite Loss in the three treatment groups were r = −0.55 for the absolute scores and r = −0.45 for the changes from baseline (Spearman rank correlation, P < .0001 at each comparison). Increased appetite (best biweekly value over baseline) was reported by 75%, 60%, and 72% of patients receiving CE, THC, and PL, respectively (P = .068).
Mean overall QOL after 6 weeks showed nearly no change from baseline and was not different for PL compared with CE (P = .80) or THC (P = .43) from combined test statistics (mean score change ± SD: CE, 1.1 ± 19.2; THC, 5.1 ± 21.2; and PL, 3.0 ± 19.5; Fig 3). The test for noninferiority of CE compared with THC (PP population) showed no significant difference (P = .90, combined).
Other Symptoms, Functional Domains of QOL, and Body Weight
From patients with a baseline VAS nausea score of ≥ 30 mm one CE (40 of 95 patients), THC (36 of 96 patients), or PL (16of 46 patients), an improvement, which was defined as a VAS score less than 30 mm, was seen in 61%, 50%, and 40%, respectively (P = .367). The respective results for mood were 60%, 46%, and 64% (P = .461) in patients on CE (33 of 95 patients), THC (25 of 96 patients), and PL (12 of 46 patients). A steady-state or slight deterioration was observed in all treatment arms for EORTC QLQ-C30 functional scales and for the following items: physical, role, emotional, cognitive, and social functioning; and dyspnea, diarrhea, and financial problems. All treatment arms showed a 5% improvement in the overall score (arithmetic mean) on the Anorexia-Cachexia EORTC QLQ-C30 module until week 2, followed by another 5% improvement until week 6 in the PL group, steady-state with THC, and worsening by 2.5% with CE. No differences between groups in body weight at baseline or week 6 (average, 61 kg) or in weight loss (average, 600 g in 6 weeks) were reported.
Subgroup Analyses
Among patients receiving CE or THC compared with PL, appetite scores higher than the average score over the whole ITT population were found in exploratory analyses of subgroups for women (6.8 mm for CE, 3.1 mm for THC, and −2.8 mm for PL) and for patients with tumors other than hematologic-lymphogenic; head, neck and lung; or GI-urogenital (1.8 mm for CE, 3.1 mm for THC, and −2.1 mm for PL). Patients with CRP values ≥ 10 mg/L at baseline (1.4 for CE; 2.5 for THC; and −5.5 for PL) had mean overall QOL scores higher than the ITT average. Patients with various degrees of weight loss showed no differences.
AEs
There were no differences between treatment arms for vital signs or ECOG PS. There were 526 AEs (CE, n = 238; THC, n = 197; and PL, n = 91); their relationship to the study medication was none or unlikely for 415 AEs (CE, n = 201; THC, n = 144; and PL, n = 70), probable for 90 AEs (CE, n = 28; THC, n = 45; and PL, n = 17), and likely for 20 AEs (CE, n = 9; THC, n = 7; and PL, n = 4). Temporary or permanent dose reductions were necessary in 78 patients (CE, n = 34; THC, n = 30; and PL, n = 14).
AEs occurring more than 10 times were nausea (CE, n = 23; THC, n = 21; and PL, n = 11), fatigue (CE, n = 16; THC, n = 14; and PL, n = 4), pain (CE, n = 11; THC, n = 17; and PL, n = 5), anemia (CE, n = 9; THC, n = 14; and PL, n = 6), dizziness (CE, n = 9; THC, n = 11; and PL, n = 7), dyspnea (CE, n = 9; THC, n = 7; and PL, n = 2), diarrhea (CE, n = 6; THC, n = 7; and PL, n = 2), and obstipation (CE, n = 6; THC, n = 7; and PL, n = 2). Table 2 lists the AE hazard rates. Of all AEs, 241 were mild (CE, n = 104; THC, n = 101; and PL, n = 36), 227 were moderate (CE, n = 113; THC, n = 71; and PL, n = 43), and 57 were severe (CE, n = 21; THC, n = 24; and PL, n = 12); one AE was undetermined. Severe AEs were mainly dizziness, nausea/vomiting, and dyspnea.
In total, 82 serious AEs30 emerged (CE, n = 32; THC, n = 33; and PL, n = 17); diagnoses recorded at least five times were dyspnea (CE, n = 5; THC, n = 8; and PL, n = 3), tumor progression (CE, n = 5; THC, n = 8; and PL, n = 2), vomiting (CE, n = 8; THC, n = 8; and PL, n = 1), worsening of general well-being (CE, n = 2; THC, n = 9; and PL, n = 1), death (CE, n = 4; THC, n = 6; and PL, n = 1), pain (CE, n = 2; THC, n = 4; and PL, n = 4), fever (CE, n = 3; THC, n = 2; and PL, n = 2), diarrhea (CE, n = 3; THC, n = 2; and PL, n = 0), and exsiccation (CE, n = 1; THC, n = 3; and PL, n = 1). Thirteen serious AEs were life threatening (CE, n = 6; THC, n = 6; and PL, n = 1), 68 required hospitalization (CE, n = 25; THC, n = 27; and PL, n = 16), and one was unexpected by the physician (CE, n = 1; THC, n = 0; and PL, n = 0). No differences were found for the CannTox scales for dizziness, feeling good, feeling high, hallucinations, heart beating, panic attacks, feeling active, or walking insecurely.
DISCUSSION
This is the first phase III trial in patients with CACS comparing the effects of cannabinoids with PL and standardized CE, an integral total product of medical cannabis. We found no differences between the three groups over 6 weeks of treatment for the primary end points of appetite and QOL, for cannabinoid-related toxicity, or for secondary end points such as mood or nausea. Like the recent North Central Cancer Treatment Group (NCCTG) phase III trial using the same dose of THC, in which 49% of patients receiving THC had better appetite at least once during the study,22 our study showed biweekly appetite scores higher than baseline values at some point during the trial in more than 50% of patients in all three groups. Like previous trials for symptom control, our study showed significant PL effects.26 No differences in toxicity ascribed to cannabinoids were found between treatment arms in either this or the NCCTG phase III trial.22
Our rationale for choosing the daily dose of THC 5 mg was based on the following: (1) the study results available in the late 1990s, which suggested that THC 2.5 mg twice daily is the optimal dose19,20; (2) the lack of evidence of a clear dose-effect relationship requiring a trial design of individual dose titration; and (3) regulatory realities in this first trial with CE as a schedule 1 substance approved in the European Union, namely, the great caution regarding psychotropic adverse effects of cannabinoids. In a PL-controlled phase II trial with THC 0.1 mg/kg/d for 7 days, 20 of 54 patients dropped out; six patients dropped out as result of drowsiness, thinking problems, and panic reactions.19 In a comparison of four doses of THC (2.5 or 5 mg, once or twice daily) over 6 weeks, 10 of 42 patients (eight taking 5 mg) dropped out because of drowsiness, memory problems, or mood changes. Only patients receiving THC 2.5 mg twice daily (47%) reported increased appetite.20 In a study of 19 patients taking THC 2.5 mg three times daily over 4 weeks, four patients dropped out, and 13 of 18 patients had increased appetite.21
The secondary end point, weight, requires cautious interpretation because, in contrast to other trials lasting 8 to 12 weeks or longer, this trial was only 6 weeks long; however, in some of those trials, weight gain was observed after 4 weeks.47,48 The degree of weight loss in our patients was lower (11.9%) than in other trials (17%47 and 13%48). Of our patients, 46% had less than 10% weight loss, whereas in the NCCTG trials,22,49 40% lost less than 5 pounds (approximately 7%).
Our trial reflects the challenges of a symptom-control trial in patients with advanced cancer, including a wide SD of symptom scores; substantial PL effect; the clinical reality of interfering symptoms and complications, adverse effects, and interactions of other medications; and a dropout rate of approximately 20% within 6 weeks. Nonetheless, we think the effects detected represent the actual effects with CACS in this real-life setting.40
Pathophysiologic features of CACS may partially explain the limited therapeutic effect of CE.41 Cannabinoids appear to influence cytokines9,42 but to insufficiently target the proinflammatory cytokines mediating CACS (tumor necrosis factor alpha, interleukin-1, and interleukin-6).43 THC reduces GI motility,44 which often is already impaired in patients with CACS.45 Efforts are being made to characterize distinct CACS subtypes based on CRP levels,46 pronounced early satiety, or chronic nausea,45 but our exploratory analysis detected no trend toward a better response to CE among putative subgroups of CACS or patients with different tumor types or degrees of weight loss.
Limitations of this study include the lack of intrapatient dose escalation of CE to test whether CBD may protect from dose-limiting THC adverse effects. The 6-week duration was short but sufficient to detect the potential of CE to change appetite loss and other symptoms. Although all investigators were trained in the procedures to be used in obtaining outcome measures, variations among centers may have affected uniformity of the data. Our stratification and block randomized design should control adequately for this potential limitation.
In conclusion, CE was well tolerated by patients but was associated with a higher hazard rate for AEs than PL. CE and THC, each with THC 2.5 mg, taken twice daily improved appetite and QOL no more than PL in patients with advanced cancer.
Source, Graphs and Figures: Comparison of Orally Administered Cannabis Extract and Delta-9-Tetrahydrocannabinol in Treating Patients With Cancer-Related Anorexia-Cachexia Syndrome: A Multicenter, Phase III, Randomized, Double-Blind, Placebo-Controlled Clinical Trial From the Ca
Purpose To compare the effects of cannabis extract (CE), delta-9-tetrahydrocannabinol (THC), and placebo (PL) on appetite and quality of life (QOL) in patients with cancer-related anorexia-cachexia syndrome (CACS).
Patients and Methods Adult patients with advanced cancer, CACS, weight loss (≥ 5% over 6 months), and Eastern Cooperative Oncology Group (ECOG) performance status (PS) ≤ 2 were randomly assigned (2
1) to receive CE (standardized for 2.5 mg THC and 1 mg cannabidiol) or THC (2.5 mg) or PL orally, twice daily for 6 weeks. Appetite, mood, and nausea were monitored daily with a visual analog scale (VAS); QOL was assessed with the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire C30 (composite score: questions 29 and 30). Cannabinoid-related toxicity was assessed every 2 weeks.Results Of 289 patients screened, 243 were randomly assigned and 164 (CE, 66 of 95 patients; THC, 65 of 100 patients; and PL, 33 of 48 patients) completed treatment. At baseline, groups were comparable for age (mean, 61 years), sex (54% men), weight loss (32% ≥ 10%), PS (13% ECOG = 2), antineoplastic treatment (50%), appetite (mean VAS score, 31/100 mm), and QOL (mean score, 30/100). Intent-to-treat analysis showed no significant differences between the three arms for appetite, QOL, or cannabinoid-related toxicity. Increased appetite was reported by 73%, 58%, and 69% of patients receiving CE, THC, or PL, respectively. An independent data review board recommended termination of recruitment because of insufficient differences between study arms.
Conclusion CE at the oral dose administered was well tolerated by these patients with CACS. No differences in patients' appetite or QOL were found either between CE, THC, and PL or between CE and THC at the dosages investigated.
INTRODUCTION
Anorexia and weight loss contribute to cancer-related fatigue, functional loss, impaired survival, and intolerance of treatment.1,2 Efforts to palliate these conditions include studies of the endocannabinoid system, which modulates appetite3 through cannabinoid receptor—related processes.4 Hyperphagic effects of cannabinoids5 and hypophagic actions of selective cannabinoid receptor antagonists6 have been reported.
Cannabis sativa contains over 60 cannabinoids, including delta-9-tetrahydrocannabinol (dronabinol, THC) and cannabidiol (CBD).7 THC, its major psychoactive component, is an antiemetic medication.8 CBD reportedly reduces the psychotropic effects of THC and has anti-inflammatory effects.9
Cannabinoids reputedly stimulate appetite, both historically10 and in recent studies of human volunteers11-13 and AIDS patients.14 Studies of patients with multiple sclerosis15-17 or pain18 have evaluated oral mixtures of THC and CBD or whole-plant cannabis extract (CE), replacing smoked marijuana. Data from four dose-finding and phase II studies of 161 patients with cancer-related anorexia-cachexia syndrome (CACS) suggest cannabinoids' potential at fixed doses of 2.5 mg of THC twice to three times daily;19-21 however, megestrol acetate palliated anorexia better than THC.22 We investigated the effects of CE and THC on appetite and quality of life (QOL) in patients with CACS.
PATIENTS AND METHODS
This multicenter, phase III, randomized, double-blind, placebo (PL)-controlled, three-arm, parallel study, which was sponsored by the Institute for Clinical Research, Berlin, adhered to the Good Clinical Practice Program,23 European Union Directive 91/507/EEC,24 and the Declaration of Helsinki.25 Participating centers obtained approval from their local institutional ethical review boards and health authorities.
Participants
Physicians recruited adult patients with advanced incurable cancer who were candidates for appetite stimulation, having had, within the past 6 months, involuntary weight loss of ≥ 5% not explained by other diseases or recent surgery. Eligible patients gave written informed consent to participate; could feed themselves; received no enteral or parenteral nutrition; had taken no anabolic agents, gestagens, cannabinoids, or corticosteroids (except for < 20 mg prednisolone for < 5 consecutive days) within the past month; and had no significant cause of secondary anorexia2 or psychiatric disorder (substance abuse or schizophrenia). Patients had an estimated life expectancy of 3 months; an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of ≤ 2; creatinine, bilirubin, and transaminase values no higher than 3× the maximum normal value; and unchanged antineoplastic therapy for 4 weeks and unchanged supportive treatment (analgesics, sedatives, tranquilizers, and anticholinergics) for 1 week before baseline assessments.
Intervention
After 7 to 14 days of baseline assessment, eligible patients were randomly assigned to treatment with PL, CE, or THC for 6 weeks. Patients received a 2-week supply of capsules to take orally twice daily (1 hour before lunch and before dinner or at bedtime, preferably with milk).
All C sativa plants were harvested at once in Switzerland and processed into a fluid extract (Verein-für-Krebsforschung, Arlesheim, Switzerland) to prepare gelatin capsules containing CE (2.5 mg THC and 1 mg CBD) or only THC (2.5 mg). PL capsules containing standardization medium (Hüls-AG, Marl, Germany) were indistinguishable from the active capsules. Encapsulation was uniformly processed by Scherer-GmbH (Eberbach, Germany). The Unfallkrankenhaus (Berlin, Germany) packed the drug.
Objectives
We tested the hypotheses that appetite and QOL improve significantly in patients treated with THC or whole-plant CE compared with PL26 and that CE and THC have equivalent effects on appetite and QOL.
Outcome Measures
Assessments were performed every 2 weeks during clinic visits at screening, at random assignment when treatment began (week 0), and at weeks 2, 4, and 6; patients kept a diary. At each visit, patients underwent an examination evaluating vital signs and ECOG PS and a test for urinary cannabinoids.
A primary end point was appetite change from baseline to week 6, assessed with a visual analog scale (VAS; 0 mm = worst; 100 mm = best). Appetite values were calculated as the mean of daily appetite VAS scores for the 7 days of week 2 in each biweekly period. The second primary end point was the change in QOL from baseline to week 6 (composite score [mean] of questions 29 [Global Health Status] and 30 [QOL] on the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire C30 [EORTC QLQ-C30], transformed into a single functional scale [range, 0 to 100]).27
The EORTC QLQ-C3027 was completed biweekly. Three other variables were monitored daily with a VAS: (1) patients' estimation of food intake during the previous 24 hours; and patients' current (2) feeling of nausea and (3) mood.
Two ad hoc EORTC QLQ-C30 modules, Anorexia-Cachexia and Cannabinoid Toxicity (CannTox), were completed biweekly. The first was translated from the Functional Assessment of Anorexia/Cachexia Therapy.28 Its internal consistency (Cronbach's α = .76) and test-retest reliability (Pearson correlation between screening and randomization, r = 0.68) were good, as was the correlation (Spearman partial rank) of the total score with the EORTC QLQ-C30 question about appetite loss (r = −0.59; P < .0001) and with VAS scores for appetite (0.36 mm; P < .0001). The CannTox module was based on the Drug Reaction Scale of adjectives describing mood, physical feelings, and perceptions of mental or cognitive functions of healthy volunteers under the influence of cannabis.29 Patients complied with study protocol if they took at least 90% of the prescribed medication over the study period, as based on the returned empty capsule blisters.
Adverse events (AEs) were classified according to Common Toxicity Criteria.30 CBCs and biochemical and urine profile analyses were performed at each visit. In cases of study drug—related AEs, the dose was reduced to one capsule daily or was interrupted until the AEs resolved, at which point patients again took two capsules or resumed treatment with one capsule daily at bedtime. Serious AEs were evaluated by the Safety Review Committee (F.S., M.S., and T.C.).
Sample Size
Sample size was calculated with reference to a study of THC used to treat AIDS-related anorexia,14 which showed a mean increase in appetite of 15 mm from baseline (standard deviation [SD], 15 mm) on a VAS in the verum group and 7.4 mm in the PL group. Half of SD may be considered clinically significant.31
To compare CE and THC with PL, a two-tailed test of differences from baseline to week 6 in appetite score was planned (significance level, 5%; power, 90%; expected between-group difference, 7 mm). For the noninferiority test of CE against THC, a one-tailed test was planned (significance level, 2.5%; power, 90%; maximum lower deviation of CE, 6 mm). For both tests, an SD of 15 mm in all three groups was assumed. The required sample was 133 patients per verum group and 68 patients in the PL group (treatment allocation in a 1
2 ratio). Assuming a dropout rate of 25%, the total sample was increased to 445 patients.After a blinded interim analysis of 46 patients to assess data variability, the analysis plan was adjusted to a group sequential, adaptive trial with the permission of each center's ethical review board. An independent data review board recommended that the trial be closed after the first unblinded interim analysis of fully monitored data from 156 assessable patients because of insufficient differences in the primary end point between the PL and verum arms; recruitment stopped with 289 patients screened.
Treatment Assignment, Random Assignment, and Blinding
Patients were randomly assigned at the week 0 visit, after the baseline period. Random assignment lists, stratified by center, were prepared by a naive statistician using SAS software (SAS Institute, Cary, NC).32 Centers received study drug sets in multiples of five, together with matching sealed envelopes containing individual treatment assignments. Investigators remained blinded until the study ended, with individual unblinding permitted only for safety reasons. The statistician and data manager who were managing random assignment, unblinding, and related decisions were naive to clinical evaluations and uninvolved in data management or analysis.
Statistical Analyses
All analyses were performed with SAS.32 One interim analysis was performed for the independent data review board (156 patients) and one for presentation (216 patients).33
Descriptive statistics were used for demographic and baseline variables. For efficacy analyses, the groups of patients enrolled before the first interim analysis, in between the two analyses, and after the second interim analysis were treated as independent samples of the whole population34,35; the three resulting test statistics were properly combined, holding an overall one-sided significance level of P = .025 partitioned into P = .00025, P = .00691, and P = .02221, respectively, regarding the three successive analyses.
The confirmatory test for superiority of any verum arm over PL was based on the intent-to-treat (ITT) population. For confirmation of noninferiority of CE to THC, the per-protocol (PP) population was used.
An a priori ordered cascade of statistical testing of six null hypotheses was applied36 in each interim and final analysis as follows: H01-3 for appetite, CE versus PL, then THC versus PL, then CE versus THC; and H04-6 for QOL in the same sequence. The significance of each statistical test could be deduced confirmatively only if the actual and all higher order tests had yielded P values smaller than the actual critical significance level.
For the ITT population, missing values were substituted by the nearest-neighbor approach.37 Sensitivity analysis additionally analyzed all efficacy parameters by the last observation carried forward method and by the worst-case approach.38
Verum and PL groups were compared using analysis of covariance, including treatment as a fixed factor and the VAS or QOL baseline values as continuous covariates. Center effects were ignored because of the many centers with only a few patients.39 Additionally, one-sided 97.5% CIs for differences were given.
All successive tests were considered exploratory. For mood, nausea, daily food intake, and EORTC QLQ-C30 items, changes from baseline to weeks 2, 4, and 6 were analyzed analogous to the primary end points. Additional exploratory analyses were performed in subgroups based on preliminary evidence of distinguishable CACS with high versus low C-reactive protein (CRP), in different tumor types, by sex, and by degree of weight loss.1,2
For all safety parameters, descriptive statistics over time was performed. P values were calculated as flagging devices to mark conspicuous, potentially clinically relevant differences. Treatment groups were compared regarding hazard rates of AEs by Cochran-Mantel-Haenszel χ2 tests and regarding continuous data by analysis of variance. The proportions of patients not completing the study were compared between treatment groups by the χ2 test.
RESULTS
Patients
Patients were recruited from October 1999 to September 2002 in 30 centers in Germany, Switzerland, and the Netherlands. Average weight loss was 11.9%; 46% of patients lost less than 10% of weight. Patient characteristics (Table 1) and the percentage of dropouts and the reasons for dropout (Fig 1) were similar between treatment arms. Compliance with treatment was similar between arms (CE, 49%; THC, 44%; and PL, 60%; P > .15); 84 patients had major protocol violations (16, 22, and nine patients on CE, THC, and PL, respectively) and/or less than 90% intake of the study medication (14, 10, and four patients on CE, THC, and PL, respectively), were missing primary end point data (three, one, and three patients on CE, THC, and PL, respectively), had THC in their serum at baseline (zero, three, and zero patients on CE, THC, and PL, respectively), or had other protocol deviations (four, three, and one patient on CE, THC, and PL, respectively). The PP analysis set comprised 80 patients (33%; 32, 31, and 17 patients on CE, THC, and PL, respectively), who were mostly men and older and had better baseline values for mood, nausea, daily food intake, and EORTC QLQ-C30 scores.
Appetite and Overall QOL
Pooled over all samples of the adaptive design, patients taking CE and PL, after 6 weeks, showed a similar mean improvement in appetite (mean, 5.4 mm; SD, 24.7 mm; and mean, 5.8 mm; SD, 23.8 mm, respectively), whereas patients taking THC showed only a 0.6-mm improvement (SD, 18.5 mm; Fig 2). No verum arms improved significantly compared with PL (combined test statistics: P = .46 for comparison with CE; P = .95 for comparison with THC). The correlations of VAS appetite scores with the EORTC QLQ-C30 categoric scale of Appetite Loss in the three treatment groups were r = −0.55 for the absolute scores and r = −0.45 for the changes from baseline (Spearman rank correlation, P < .0001 at each comparison). Increased appetite (best biweekly value over baseline) was reported by 75%, 60%, and 72% of patients receiving CE, THC, and PL, respectively (P = .068).
Mean overall QOL after 6 weeks showed nearly no change from baseline and was not different for PL compared with CE (P = .80) or THC (P = .43) from combined test statistics (mean score change ± SD: CE, 1.1 ± 19.2; THC, 5.1 ± 21.2; and PL, 3.0 ± 19.5; Fig 3). The test for noninferiority of CE compared with THC (PP population) showed no significant difference (P = .90, combined).
Other Symptoms, Functional Domains of QOL, and Body Weight
From patients with a baseline VAS nausea score of ≥ 30 mm one CE (40 of 95 patients), THC (36 of 96 patients), or PL (16of 46 patients), an improvement, which was defined as a VAS score less than 30 mm, was seen in 61%, 50%, and 40%, respectively (P = .367). The respective results for mood were 60%, 46%, and 64% (P = .461) in patients on CE (33 of 95 patients), THC (25 of 96 patients), and PL (12 of 46 patients). A steady-state or slight deterioration was observed in all treatment arms for EORTC QLQ-C30 functional scales and for the following items: physical, role, emotional, cognitive, and social functioning; and dyspnea, diarrhea, and financial problems. All treatment arms showed a 5% improvement in the overall score (arithmetic mean) on the Anorexia-Cachexia EORTC QLQ-C30 module until week 2, followed by another 5% improvement until week 6 in the PL group, steady-state with THC, and worsening by 2.5% with CE. No differences between groups in body weight at baseline or week 6 (average, 61 kg) or in weight loss (average, 600 g in 6 weeks) were reported.
Subgroup Analyses
Among patients receiving CE or THC compared with PL, appetite scores higher than the average score over the whole ITT population were found in exploratory analyses of subgroups for women (6.8 mm for CE, 3.1 mm for THC, and −2.8 mm for PL) and for patients with tumors other than hematologic-lymphogenic; head, neck and lung; or GI-urogenital (1.8 mm for CE, 3.1 mm for THC, and −2.1 mm for PL). Patients with CRP values ≥ 10 mg/L at baseline (1.4 for CE; 2.5 for THC; and −5.5 for PL) had mean overall QOL scores higher than the ITT average. Patients with various degrees of weight loss showed no differences.
AEs
There were no differences between treatment arms for vital signs or ECOG PS. There were 526 AEs (CE, n = 238; THC, n = 197; and PL, n = 91); their relationship to the study medication was none or unlikely for 415 AEs (CE, n = 201; THC, n = 144; and PL, n = 70), probable for 90 AEs (CE, n = 28; THC, n = 45; and PL, n = 17), and likely for 20 AEs (CE, n = 9; THC, n = 7; and PL, n = 4). Temporary or permanent dose reductions were necessary in 78 patients (CE, n = 34; THC, n = 30; and PL, n = 14).
AEs occurring more than 10 times were nausea (CE, n = 23; THC, n = 21; and PL, n = 11), fatigue (CE, n = 16; THC, n = 14; and PL, n = 4), pain (CE, n = 11; THC, n = 17; and PL, n = 5), anemia (CE, n = 9; THC, n = 14; and PL, n = 6), dizziness (CE, n = 9; THC, n = 11; and PL, n = 7), dyspnea (CE, n = 9; THC, n = 7; and PL, n = 2), diarrhea (CE, n = 6; THC, n = 7; and PL, n = 2), and obstipation (CE, n = 6; THC, n = 7; and PL, n = 2). Table 2 lists the AE hazard rates. Of all AEs, 241 were mild (CE, n = 104; THC, n = 101; and PL, n = 36), 227 were moderate (CE, n = 113; THC, n = 71; and PL, n = 43), and 57 were severe (CE, n = 21; THC, n = 24; and PL, n = 12); one AE was undetermined. Severe AEs were mainly dizziness, nausea/vomiting, and dyspnea.
In total, 82 serious AEs30 emerged (CE, n = 32; THC, n = 33; and PL, n = 17); diagnoses recorded at least five times were dyspnea (CE, n = 5; THC, n = 8; and PL, n = 3), tumor progression (CE, n = 5; THC, n = 8; and PL, n = 2), vomiting (CE, n = 8; THC, n = 8; and PL, n = 1), worsening of general well-being (CE, n = 2; THC, n = 9; and PL, n = 1), death (CE, n = 4; THC, n = 6; and PL, n = 1), pain (CE, n = 2; THC, n = 4; and PL, n = 4), fever (CE, n = 3; THC, n = 2; and PL, n = 2), diarrhea (CE, n = 3; THC, n = 2; and PL, n = 0), and exsiccation (CE, n = 1; THC, n = 3; and PL, n = 1). Thirteen serious AEs were life threatening (CE, n = 6; THC, n = 6; and PL, n = 1), 68 required hospitalization (CE, n = 25; THC, n = 27; and PL, n = 16), and one was unexpected by the physician (CE, n = 1; THC, n = 0; and PL, n = 0). No differences were found for the CannTox scales for dizziness, feeling good, feeling high, hallucinations, heart beating, panic attacks, feeling active, or walking insecurely.
DISCUSSION
This is the first phase III trial in patients with CACS comparing the effects of cannabinoids with PL and standardized CE, an integral total product of medical cannabis. We found no differences between the three groups over 6 weeks of treatment for the primary end points of appetite and QOL, for cannabinoid-related toxicity, or for secondary end points such as mood or nausea. Like the recent North Central Cancer Treatment Group (NCCTG) phase III trial using the same dose of THC, in which 49% of patients receiving THC had better appetite at least once during the study,22 our study showed biweekly appetite scores higher than baseline values at some point during the trial in more than 50% of patients in all three groups. Like previous trials for symptom control, our study showed significant PL effects.26 No differences in toxicity ascribed to cannabinoids were found between treatment arms in either this or the NCCTG phase III trial.22
Our rationale for choosing the daily dose of THC 5 mg was based on the following: (1) the study results available in the late 1990s, which suggested that THC 2.5 mg twice daily is the optimal dose19,20; (2) the lack of evidence of a clear dose-effect relationship requiring a trial design of individual dose titration; and (3) regulatory realities in this first trial with CE as a schedule 1 substance approved in the European Union, namely, the great caution regarding psychotropic adverse effects of cannabinoids. In a PL-controlled phase II trial with THC 0.1 mg/kg/d for 7 days, 20 of 54 patients dropped out; six patients dropped out as result of drowsiness, thinking problems, and panic reactions.19 In a comparison of four doses of THC (2.5 or 5 mg, once or twice daily) over 6 weeks, 10 of 42 patients (eight taking 5 mg) dropped out because of drowsiness, memory problems, or mood changes. Only patients receiving THC 2.5 mg twice daily (47%) reported increased appetite.20 In a study of 19 patients taking THC 2.5 mg three times daily over 4 weeks, four patients dropped out, and 13 of 18 patients had increased appetite.21
The secondary end point, weight, requires cautious interpretation because, in contrast to other trials lasting 8 to 12 weeks or longer, this trial was only 6 weeks long; however, in some of those trials, weight gain was observed after 4 weeks.47,48 The degree of weight loss in our patients was lower (11.9%) than in other trials (17%47 and 13%48). Of our patients, 46% had less than 10% weight loss, whereas in the NCCTG trials,22,49 40% lost less than 5 pounds (approximately 7%).
Our trial reflects the challenges of a symptom-control trial in patients with advanced cancer, including a wide SD of symptom scores; substantial PL effect; the clinical reality of interfering symptoms and complications, adverse effects, and interactions of other medications; and a dropout rate of approximately 20% within 6 weeks. Nonetheless, we think the effects detected represent the actual effects with CACS in this real-life setting.40
Pathophysiologic features of CACS may partially explain the limited therapeutic effect of CE.41 Cannabinoids appear to influence cytokines9,42 but to insufficiently target the proinflammatory cytokines mediating CACS (tumor necrosis factor alpha, interleukin-1, and interleukin-6).43 THC reduces GI motility,44 which often is already impaired in patients with CACS.45 Efforts are being made to characterize distinct CACS subtypes based on CRP levels,46 pronounced early satiety, or chronic nausea,45 but our exploratory analysis detected no trend toward a better response to CE among putative subgroups of CACS or patients with different tumor types or degrees of weight loss.
Limitations of this study include the lack of intrapatient dose escalation of CE to test whether CBD may protect from dose-limiting THC adverse effects. The 6-week duration was short but sufficient to detect the potential of CE to change appetite loss and other symptoms. Although all investigators were trained in the procedures to be used in obtaining outcome measures, variations among centers may have affected uniformity of the data. Our stratification and block randomized design should control adequately for this potential limitation.
In conclusion, CE was well tolerated by patients but was associated with a higher hazard rate for AEs than PL. CE and THC, each with THC 2.5 mg, taken twice daily improved appetite and QOL no more than PL in patients with advanced cancer.
Source, Graphs and Figures: Comparison of Orally Administered Cannabis Extract and Delta-9-Tetrahydrocannabinol in Treating Patients With Cancer-Related Anorexia-Cachexia Syndrome: A Multicenter, Phase III, Randomized, Double-Blind, Placebo-Controlled Clinical Trial From the Ca
