Department of Health and Human Services
Research Plan Portion of Public Health Service Grant Application
submitted April 1996

Protocol: Smoked Marijuana for HIV-Associated Anorexia and Wasting

Principal Investigator: Donald I. Abrams, M.D.
Community Consortium, University of California
San Francisco AIDS Program at San Francisco General Hospital


Our primary aim is to evaluate the safety and efficacy of smoked marijuana as an appetite stimulant for HIV-associated anorexia and weight loss. We propose to do this by conducting two related, sequential studies: a phase I/II randomized, double-blind, placebo-controlled, within-subjects evaluation of smoked marijuana conducted in an inpatient setting at the General Clinical Research Center (GCRC) at San Francisco General Hospital; and a phase II/III randomized, open-label study of smoked marijuana versus dronabinol conducted in an inpatient setting. This is a proposal to conduct the initial inpatient component of these investigations.

The inpatient study will provide data on the effects of moderate (~4%) THC-content marijuana and food intake, as well as safety data on HIV viral burden and immunologic, pulmonary, endocrine and neuropsychological functioning. Primary endpoints include changes in HIV RNA (as measured by bDNA assay), lymphocyte function and cytokine levels, lung capacity and lung volume (as measured by spirometry), serum testosterone, LH and FSH levels, plasma THC levels, body composition, cognitive and motor function (as measured by standard neuropsychological tests) and quality of life variables.

Fifteen subjects will be recruited for this study, which will be comprised of 5 successive stages: a 5-day lead-in phase in which baseline measurements will be obtained while patients are hospitalized in the GCRC at San Francisco General Hospital. This will be followed immediately by a 5-day lead-in period in the GCRC in which patients receive either moderate (~4%) THC-content marijuana cigarettes or placebo marijuana cigarettes. This stage will be followed by a wash-out period of four weeks in which patients are not hospitalized. The wash-out period will be followed by another 5-day lead-in period in the GCRC, followed immediately by a second 5-day period in the GCRC in which patients are crossed-over to placebo marijuana cigarettes or active THC-content marijuana cigarettes, depending upon their initial randomization allocation. Patients will spend a total of 20 days in the GCRC.

If the results of this initial trial indicate an absence of significant adverse effects of smoked marijuana in this population, then estimates of effect-size and response variability obtained from this study will be used to plan and implement a larger, comparative trial evaluating smoked marijuana to dronabinol in an outpatient setting, with sufficient power to answer definitively the question of efficacy.

Protocol: Smoked Marijuana for HIV-Associated Anorexia and Wasting


Our primary aim is to evaluate the safety and efficacy of smoked marijuana as an appetite stimulant for HIV-associated anorexia and weight loss. We propose to do this by conducting two related, sequential studies: a phase I/II randomized, double-blind, placebo-controlled, within-subjects evaluation of smoked marijuana conducted in an inpatient setting at the General Clinical Research Center at San Francisco General Hospital; and a phase II/III randomized, open-label study of smoked marijuana Versus dronabinol conducted in an Outpatient setting. This is a proposal to conduct the initial inpatient component of these investigations.

The inpatient study will provide data on the effects of smoking moderate (~4%) THC-content marijuana on appetite and food intake, as well as safety data on HIV viral burden and immunologic, pulmonary, endocrine, and neuropsychological functioning. Based on the results of the inpatient study, we plan to conduct an outpatient study that will provide comparative data on the effects of smoked marijuana to the licensed oral synthetic cannabinoid preparation, dronabinol (Marinol, Roxane Laboratories, Columbus, OH) on body weight and body composition. The outpatient study will also provide safety data on HIV viral burden and immunologic, pulmonary, and endocrine function when these agents are used to treat patients with HIV-associated anorexia and weight loss over several months.

As the public policy implications posed by the medical use of marijuana are significant, we feel an inpatient study conducted under well-controlled, experimental conditions (as described herein) and an outpatient study conducted under "real world" conditions in primary care settings are needed to evaluate fully the safety and potential efficacy of this highly controversial therapy.

1. Primary Objective

2. Secondary Objectives

3. Primary Endpoints

4. Secondary Endpoints

B. BACKGROUND AND SIGNIFICANCE Largely on the strength of anecdote alone, and in the absence of any supporting data, smoked marijuana has become an increasingly popular treatment for HIV-associated anorexia and weight loss. Moreover, legislation has recently been presented in the State of California to allow physicians to prescribe marijuana for conditions where it has been shown to have medical benefit. HIV-associated wasting is one of the four conditions proposed. Use of smoked marijuana has also been facilitated in the San Francisco Bay Area by the creation of the Cannabis Buyers Club (Goldberg 1996), which sells marijuana to nearly eight thousand clients who present a physician's note documenting that they have HIV infection or another medical condition that may be ameliorated by smoking marijuana. Similar ventures have been established in other cities around the nation. =46or a number of years, providers caring for patients with HIV infection in the San Francisco Bay Area have been concerned about the safety of marijuana smoking by patients with an acquired immune deficiency. Others have been impressed with positive effects in weight gain, mood and quality of life in their patients smoking marijuana.

The Community Consortium, an association of HIV health care providers in the San Francisco Bay Area, was established in March 1985 to encourage communication and collaboration between AIDS researchers at the University of California San Francisco AIDS Program at San Francisco General Hospital and frontline primary care physicians in practice in the community. Shortly after its inception, the Community Consortium developed into one of the nation's pioneer community-based clinical trials organizations. One of our primary goals has always been to investigate agents that are in widespread use in the community in a controlled fashion in order to evaluate their safety and possible efficacy. For example, in 1986 there was increased use of inhaled pentamidine as a prophylaxis against Pneumocystis carinii pneumonia so as to avoid systemic agents that might interact with zidovudine, the first licensed antiretroviral drug. Because of the widespread community interest in and use of this new modality, the Consortium designed and conducted a trial comparing three different doses of aerosolized pentatmidine. The results of this trial led to approval of inhaled pentamidine for Pneumocystis prophylaxis by the U.S. Food and Drug Administration (FDA) in 1989 (Leoung 1990). Similarly, when it became evident that an increasing number of patients with HIV infection were reportedly utilizing inhaled marijuana as treatment (or prophylaxis) for HIV-associated wasting, Community Consortium investigators felt that a clinical trial should be designed to evaluate the safety and efficacy of this potential therapy.

Designing a study of smoked marijuana in AIDS patients presents a challenge that surpasses the normal effort of designing studies of new and promising therapies for HIV/AIDS. The political and emotional aspects of any study involving smoked marijuana complicate the issue logarithmically. In addition, the previously established compassionate use program for the distribution of marijuana through the US Public Health Service was terminated just as we began initial discussions regarding the need for a controlled clinical trial (Cotton 1992). Convinced of the public health necessity to conduct such a study, however, we proceeded in 1993 to develop a pilot study comparing three different strengths of inhaled marijuana to the licensed tetrahydrocannabinol preparation, dronabinol, which is approved for treatment of anorexia associated with HIV-associated wasting (See Appendix A). The 12-week pilot study was to be conducted as an outpatient study and would include body composition measurements as well as evaluations of pulmonary function, endocrine function and HIV viral load. The study was designed with constructive input from the Pilot Drug Evaluation Staff at the FDA, from whom an IND was obtained (#43,542). The pilot study was also approved by the Committee on Human Research at the University of California at San Francisco, and received conditional approval from the California Research Advisory Panel (which reviews all clinical trials involving controlled substances), pending identification of a legal source of marijuana. Multiple complications ensued over the next three years regarding obtaining a legal supply of marijuana to use in the study (Voelker 1994; Lehrman 1995; Grinspoon 1995; Abrams 1995; Steele 1995).

To evaluate fully the safety and efficacy of smoked marijuana for HIV-associated anorexia and weight loss, we now propose to conduct an intensive, Phase I/II study comparing smoked marijuana versus placebo in an inpatient setting, before proceeding to a larger, Phase II/III efficacy study comparing smoked marijuana to dronabinol conducted in an outpatient setting.

I. Safety Concerns

The potential detrimental effects of marijuana smoking in patients without HIV infection are numerous. Interestingly, most of the literature is at least twenty years old and often conflicting reports can be found, occasionally by the same author. With regard to patients with HIV infection, the impact of smoked marijuana on immune function, pulmonary function, gonadal function and neuropsychiatric status have not yet been described and would seem to be particularly critical safety parameters to examine.

1. Immune Function

Studies of the effect of marijuana on immunity have been contradictory, and, when viewed in the aggregate, difficult to interpret. On the one hand, the major psychoactive component of marijuana, delta-9-tetrahydrocannabinol (THC), has been reported to suppress immune functions such as lymphocyte proliferation, antibody production, natural killer cell activity and macrophage function; to dysregulate production of such pro-inflammatory cytokines as interferon and tumor necrosis factor (TNF) (Friedman 1995); and to confer altered susceptibility in vivo to infection with intracellular organisms such as Legionella pneumophilia (Klein 1994; Klein 1995) and in vivo to herpes simplex virus type-1 infected cells (Cabral 1992; Fischer-Stenger 1992). With Munro's discovery of a new type of cannabinoid receptor present not in the brain but only in peripheral tissues (particularly reticuloendothelial tissues), the potential for interactions of THC with the immune system was further validated (Munro 1993). In a commentary on the discovery of the receptor, an industry pharmacologist poses the question, "Is the second cannabinoid receptor only in the spleen? The level of expression, probably high in the active macrophages, in a region where the outside world meets the immune system, guggests a possible role in inflammatory and immune responses to infection or other foreign antigens." (Iverson 1993).

It has been hypothesized that these effects may be related to THC-induced shifts in the balance of"Thl" and "Th2" cells (Newton 1994). In contrast, and as reviewed by Hollister (Hollister 1992), many of the effects so documented for THC have been observed in conditions, both in vivo and in vitro, in which supraphysiologic doses of the compound are used and/or without inclusion of controls which have similar lipophilic properties. Even relatively simple observations (e.g., that phytohemagglutinin and mixed lymphocyte culture responses are suppressed in young, chronic marijuana smokers) (Nahas 1974), have been difficult to reproduce (White 1975; Lau 1976). More recently, conflicting reports have been generated regarding the impact of THC on levels of TNF-alpha. Whereas some investigators report THC inhibition of TNF (Friedman 1995), another study utilizing ELISA (enzyme-linked immunoabsorbant assay) techniques demonstrated decreased interleukin-6, but increased TNF levels in a mouse macrophage system (Shivers 1994). As these cytokines, particularly TNF, have been implicated in the pathogenesis of HIV-related wasting, determining the impact of THC on levels of TNF and related cytokines in human subjects smoking marijuana would be a significant contribution.

No controlled investigations of the impact of marijuana on immune function in patients with HIV infection have been conducted to date; however, information suggesting impaired cellular immune function is certainly worrisome. Interestingly, suggestions of B lymphocyte modulation and TNF inhibition could, in fact, have potential beneficial effects in patients with HIV-induced immune dysregulation over the long term. Whether a stimulant or suppressant of immune function, marijuana could potentially lead to increased viral burden. This potential effect, as well, has never been investigated in a prospective, controlled fashion. Retrospective analyses from the Multicenter AIDS Cohort Study evaluating outcomes in 1662 seropositive users of psychoactive drugs found that none of the drugs used by participants was associated with enhanced clinical or immunologic expression of HIV infection (Kaslow 1989). Of note, use of marijuana in the preceding two years was reported by 89% of the seropositive men in the cohort. This confirmed a previous observation from our San Francisco General Hospital experience (Roland 1987).

The disparate results on the effects of THC on the immune system may be related to differences in study populations, drug composition, drug concentration, or assay conditions. Taking a conservative approach, it is perhaps reasonable at this juncture to assume that marijuana can exert immunosuppressive effects under certain experimental conditions. From the standpoint of this study, the key question will be whether marijuana exerts demonstrable immunosuppressive effects when administered as three ~4% THC cigarettes da!ly over a period of five days. If warranted, future studies may assess whether chronic exposure to similar doses exert additional effects not observed after short term exposure.

2. Pulmonary Effects

Much of the work on the pulmonary effects of smoked marijuana has been done by Tashkin and his group at the University of California Los Angeles. Again, the published data are less than crystal clear. Tashkin wrote in 1987 that "evidence regarding the potential long-term pulmonary consequences of regular marijuana smoking is mixed. Several studies conducted during the past decade on whole animals and isolated cell systems exposed to marijuana smoke, as well as some clinical observations, suggest that marijuana can be harmful to the lung. Conversely, human studies carried out abroad have failed to find any evidence of respiratory dysfunction or disease in long-term heavy users of marijuana" (Tashkin 1987). However, when comparing effects in tobacco and marijuana smokers, Tashkin has demonstrated that marijuana smoke can be harmful to the lungs. In a chronic smoking experiment, young healthy volunteers developed decreases in forced expiratory volume in one second (FEV[1]), in maximal mid-expiratory flow rate, in plethysomographic specific airway conductance and diffusing capacity (Tashkin 1976). In subsequent studies comparing the effects of marijuana smoke to tobacco, his group has reported that marijuana users have significant pulmonary symptoms (cough, sputum, wheeze, bronchitis episodes); detrimental effects on specific airway conductance and resistance (Tashkin 1987); an inconsistent effect on nonspecific airway hyperresponsiveness (Tashkin 1993) a high prevalence of abnormal airway appearance and histologic findings including squamous metaplasia and dysplasia (Tashkin 1987); an increased alveolar cellular response composed predominantly of alveolar macrophages (Tashkin 1987); a substantially greater respiratory burden of carbon monoxide and tar than those smoking a similar quantity of tobacco (Wu 1988); and a trend towards increased alveolar macrophage DNA damage (Sherman 1995). In a recent report, Tashkin's group defines a differential effect of tobaccco and marijuana on lower respiratory tract and circulating T-lymphocyte subpopulations (Wallace 1994). Tobacco was found to have a significant effect toward decreasing percentages of bronchoalveolar CD4 cells and increasing CD8 cells, thus yielding lower CD4:CD8 ratios, whereas marijuana use decreased the percentage of bronchoalveolar CD8 cells. Both tobacco and marijuana increased bronchoalveolar B-lymphocyte populations. In the peripheral blood, marijuana (but not tobacco use) was associated with increased percentages of CD4 cells, decreased CD8 lymphocytes and an elevated CD4:CD8 ratio -- effects frequently sought through therapeutic intervention in patients with HIV infection.

Concerns about the potential risk of sinopulmonary infection from smoking marijuana have been longstanding. Aspergillus contamination of marijuana has been noted and presumed to be the cause of possible clinical infection (Kagen 1981; Schwartz 1985). More recent studies have demonstrated decreased ability ofpulmonary alveolar macrophages to destroy Candida albicans ( Sherman 1991) and suppressed resistance to Legionella pneumophilia secondary to THC (Klein 1994, Klein 1995). These findings would be of concern with regard to an increased possibility of pulmonary infectious processes in patients with acquired immune deficiency utilizing smoked marijuana. No prospective analysis of such risk has been conducted, although a retrospective review examining risk factors for the first episode ofbacterial pneumonia among HIV positive injection drug users with a prior history of Pneumocystis carinii pneumonia found that "smoking illicit drugs" was a significant risk (Caiaffa 1994). Unfortunately, marijuana was included together with cocaine and crack smoking in this analysis. In addition, virtually all of the injection drug users evaluated also smoked cigarettes, further confounding the analysis. Despite the statistical significance and the higher prevalence of marijuana use (88%) compared to cocaine (26%) and crack (9%) smoking, the data supporting the claim that smoking illicit drugs had the strongest effect on the risk of bacterial pneumonia in this narrow subpopulation of AIDS patients would be strengthened if the individual smoked agents had been analyzed separately.

3. Endocrine Effects

It has been demonstrated that men with HIV infection may become hypogonadal during the course of their disease (Dobs 1988; Croxson 1989; Raffi 1991; Christeff 1992). It is conceivable that agents such as marijuana may have an impact on changes in male sex hormones. Testosterone synthesis in rats is decreased by THC. In a four week exposure study in men conducted in the 1970's, a decrease in luteinizing hormone (LH) occurred first, followed by decreases in testosterone and follicle-stimulating hormone (FSH). A more recent evaluation in ten chronic marijuana users demonstrated that basal and stimulated levels of LH were reduced in comparison to age-matched controls, with no difference in FSH and prolactin levels and responses (Vescovi 1992). The authors postulate that chronic marijuana use may selectively impair the hypothalamic mechanism regulating LH secretion. In the current study, this information is relevant as hypogonadism has been associated with AIDS-related wasting (Coodley 1994). Marijuana smoking could precipitate or exacerbate the hypogonadal state necessitating testosterone replacement therapy.

4. Neuropsychiatric Effects

The question of the acute neuropsychological effects of marijuana intoxication on healthy subjects has been addressed in several studies in recent years. Research indicates that acute cognitive effects do exist. They may vary based on the amount of prior exposure to marijuana (Marks 1989), and may (Solowij 1995; Block 1993; Solowij 1991; Varma 1988) or may not be confounded by chronic cognitive effects of long term, heavy exposure to the substance (Lundquist, 1995; Carlin 1977; Weckowicz 1977; Weckowicz 1973).

Research has generally found that cognitive impairment covaries with the amount of marijuana exposure and plasma THC levels. Most studies indicate that this impairment may last anyyvhere from 65 minutes (Heishman 1989) to 24 hours (Leirer 1991; Heishman 1990) after exposure. Heishman (1990) found that on the day of exposure subjects exhibited decreased performance on a serial addition/subtraction task. Their performance returned to baseline the following day. Frank (1976) noted a slight acute effect, but no significant difference between marijuana exposed and control group patients over time on an unnamed "computation" task. Research is conflicting regarding the acute effects of marijuana on tho Digit Symbol Subtest of the Wechsler Adult Intelligence Scale--Revised (WAIS-R) (1981), a task of visual-motor speed and novel learning. Some studies indicate that impairment is present (Azorlosa 1991; Heishman 1989), and others report a normal performance in marijuana-exposed subjects (Chait 1990; Chait 1985; Jones 1976). On an unnamed complex visual matching task, Jones (1976) found a pattern of slowing of response time during initial THC exposure followed by a return to baseline performance, despite continued THC exposure. In addition, several subjects (3 of 12) showed no impairment in performance on the task during the thirty day study. This suggests both acclimation to THC, as well as a possible practice effect on the measure. While one study found no effect of marijuana on a computerized task of divided attention (Heishman 1989), others who have explored this cognitive domain have found impairment in subcomponents of the task (Azorlosa 1991; Barnett 1985; Chait 1990). Frank (1976) gave their subjects the Trail Making Test, another measure of executive functioning, on three occasions during a 36 day hospital admission. They found no significant differences between marijuana-exposed and control subjects, but noted a slight practice effect on the test. Studies using a computerized program for central and peripheral light reaction time (Heishman 1990; Marks 1999) both found impairment with acute marijuana exposure. Research indicates that acute marijuana exposure impairs ability to accurately judge brief time intervals (Chait 1990; Chait 1985; =46crraro 1980), and to rapidly sort cards into complex (by suit), and simple (equal stacks, face down) categories. In addition Chait (1995) found that marijuana significantly decreased the number of words generated on a free recall task.

Other measures have been employed in the investigation of the present topic. Each of these has been explored by one investigator, who has found no significant difference between baseline and acute marijuana exposure performances: a computerized version of The Buschke Selective Reminding Task (Chait 1990), an unidentified task of auditory memory (Jones 1976), a task of logical reasoning and a "two letter search" (Heishman 1990).

There are shortcomings in the methodology, of the studies cited above. A number of the measures outlined above are presented in a non-standardized manner. Each of the above mentioned studies of marijuana use also addressed the use of other drugs and alcohol in the sample. Still, subjects often varied within studies on the amount of alcohol and other substances consumed, both by history, and at the time of research. =46inally studies which did not require subjects to remain in a controlled setting during the course of the research may also have allowed exposure to substances/events which could have skewed results.

Most studies cited above attempted to carefully screen out for medical illness or conditions which may contribute to cognitive impairment. No study as yet has addressed the cognitive effects of the use of marijuana in a population of HIV-positive patients who are using the substance to control symptoms of their illness.

II. Efficacy Studies

1. HIV-associated Wasting and Current Treatments

The prognostic significance of weight loss in HIV infection has long been recognized. Early antiretroviral therapy trials used weight gain as an efficacy parameter. Despite an increased armamentarium of licensed antiretroviral agents, weight loss and wasting continue to be problematic for patients with advanced HIV disease. A CPCRA analysis of opportunistic events occurring within the 6 months prior to death in a cohort of 1883 HIV-infected individuals showed that the wasting syndrome and Mycobacterium avium Complex share second place behind Pneumocystis carinii pneumonia as the most frequent premorbid diagnoses (Chan 1995). The 6-month cumulative mortality in another CPCRA cohort of HIV-infected individuals diagnosed with wasting syndrome was 41.7 percent, for a relative risk of 2.5 (Neaton 1995).

Wasting syndrome is also a frequent initial disease progression diagnosis in patients with HIV disease. In an analysis of another CPCRA cohort comparing outcomes by gender, 11 percent of the 768 women and 12 percent of the 3779 men developed wasting as their first progression-of-disease event after 15 months of followup (Melnick 1994). Other analyses of this cohort found relationships among baseline weight, percent weight change, disease progression, and survival. For instance, Gibert examined a cohort of 2,076 persons with CD4+ cell counts of <500/mm[3] and found weight and percent weight change to be important predictors of mortality independent of CD4+ cell count (Gibert 1995). Similarly, Wheeler analyzed the same cohort and showed that a decrease in body weight over 4 months correlated with an increased risk of opportunistic complications during the same 4-month period (Wheeler 1995). Moreover, a decrease in weight over 4 months was associated with the occurrence of an opportunistic complication in the subsequent year, independent of CD4+ cell count.

Loss of lean body mass has likewise been identified as an important predictor of negative outcome in patients with advanced HIV infection. Kotler (1989) analyzed depletion of body cell mass (nonadipose cellular mass, predominantly muscle and viscera) in wasting AIDS patients. The investigators noted that body cell mass values at death were 54 percent of normal and suggested that there is a critical level of body cell mass needed to support life. They postulated that HIV disease-related complications might cause depletion below the critical body cell mass limit, and that interventions that successfully maintain body cell mass may prolong survival.

A recent analysis by Suttmann (1995) suggests that loss of body cell mass is a more significant predictor of negative outcome than a drop in CD4+ cell count to less than 50/mm[3]. Measuring body cell mass by bioelectrical impedance analysis (BIA) in 39 AIDS patients, significant differences in survival were noted. Patients with body cell mass less than 30 percent of body weight at entry into the study survived 335 days compared to 527 days for patients with greater body cell mass (p<0.05).

Data suggest that decreased energy intake is probably the most important contributor to wasting (Macallan 1995; Grunfeld 1992). However, present treatments for HIV-associated anorexia are limited. Megestrol acetate, a progestational hormone, has been approved as an appetite stimulant that produces weight gain in patients with the wasting syndrome (VonRoen 1994; Oster 1994). The weight gain resulting from megestrol use has been predominantly fat and not the more relevant lean body mass component. Dronabinol (delta-9-tetrahydrocannabidiol, Marinol) was also approved by the FDA for the treatment of the anorexia associated with HIV-associated wasting (Gorter 1992; Struwe 1993; Beal 1995). The limited indication stems from the fact that although appetite was increased in the dronabinol recipients compared to controls, the impact on weight gain was not statistically significant. A recently completed pharmacokinetic study of the combination of megestrol and dronabinol compared to each single agent alone concluded that there was no additive weight gain effect when dronabinol was given in combination with megestrol acetate, or any weight gain effect of dronabinol when given alone (dose 2.5 mg twice daily for twelve weeks) (Timpone 1996). Analysis of the visual analog scale for hunger (VASH) response showed improvement within the first week of the trial on all stud), arms without a statistically significant difference in the therapies.

2. Prior Studies of Cannabinoids and Appetite Stimulation

An extensive literature regarding the impact of smoked marijuana on appetite has accumulated, although much of it was published 20 to 50 years ago. Anecdotal accounts of increased food intake have always been reported by marijuana smokers. A study evaluating body weight and caloric intake in 12 "casual" and 15 "heavy" marijuana users reported weight gains of 2.8 and 3.7 pounds respectively during the first five days of a 21 day smoking trial (Greenberg 1976). Control subjects, without access to marijuana or other drugs but otherwise exposed to the identical ward experimental conditions, gained only 0.2 pounds during the same time interval. Despite the absence of sophisticated body composition analysis technologies (such as BIA or DEXA), the investigators believed that water retention was not a major factor in the weight gain, contradicting an earlier report by a member of our group (Benowitz 1975). An increase in caloric intake accompanied the initial weight gain; however, caloric intake was noted to decrease subsequently during the remainder of the 21 day trial. Foltin has investigated the impact of smoking marijuana on food intake in men confined in residential laboratories under various conditions (Foltin 1986). He reported that smoking marijuana compared to placebo was more likely to lead to increased food intake during social access periods when the subject interacted with others as opposed to when the marijuana was smoked in isolation. The average mean caloric daily intake increased by 758 kcal (p<0.001) across the nine subjects under active marijuana as compared to placebo conditions during social access periods. The increased intake was noted to be consumed between meals as snack foods. In a subsequent study of six subjects confined to the residential laboratory for 13 days, smoked active marijuana significantly increased total daily caloric intake by 40%, with increased food intake evident during both private and social periods (Foltin 1988). The increase was again noted to be confined to between meal snacks as a consequence of an increased number of snacking occasions, principally in the sweet, solid item category. The investigators noted that during this short trial, smoking active marijuana significantly increased body weight. Weight increased an average of 3 kg over the three day active smoking periods and decreased subsequently by nearly the same amount over the three day placebo smoking periods. In their experience, smoking marijuana led to reduced physical activity levels and increased sleeping time which could explain why increases in body weight during periods of active marijuana smoking were greater than predicted by caloric intake alone.

3. Oral THC versus Smoked Marijuana

Dronabinol was approved in 1985 by the FDA for use as an anti-emetic in patients receiving cytotoxic chemotherapy (Doblin 1991). In 1992, it was approved for treatment of anorexia in patients with HIV-associated wasting. The experience of patients using dronabinol has been mixed. Variable absorption after oral ingestion is well recognized, leaving some patients with a minimal drug effect. Smoking is a more efficient route of delivery with a more desirable pharmacokinetic pattern for patients seeking to maximize any potential appetite-stimulating effect (Agurell 1986). The smoking curve tends to parallel the effects of intravenous administration at about half the concentration. Plasma drug levels and heart rate peak within 15 minutes of an inhaled dose and rapidly decline. Oral THC increases heart rate to a lesser extent than smoked marijuana with a peak occurring two to three hours after ingestion (Chait 1992). Heart rate remained elevated four hours after ingestion. It is the combination of these features -- delayed onset, prolonged duration and less ability to titrate the effect -- that has led patients with HIV infection to turn in increasing numbers to utilization of smoked marijuana as a self-medication for this condition.

A recent report describes a number of small related trials comparing different routes of THC administration for their effects on appetite stimulation (Mattes 1994). Eleven subjects received single doses by oral, sublingual and inhaled routes. In another component of the trial, a 2.5 mg dose was administered twice daily for three days by oral and rectal suppository routes. There was a high level of variability in plasma drug levels, particularly in subjects with oral drug administration. Two of the 11 (18%) subjects in the oral trial of the multiroute study and 18/57 (32%) of subjects in the larger, acute oral study showed no detectable plasma THC or metabolite level in the four hours following ingestion of the active drug. For patients receiving oral THC in this study, the time course of peak levels and the concentration area under the curve (AUC) values were highly variable across subjects. The suppository led to the highest AUC, followed by inhalation. The metabolite AUC was highest for the oral and lowest following inhalation. Mean energy intake after inhalation of THC was 2719 +/- 359 kcal; 481 kcal more than after oral dosing and 603 kcal greater than sublingual dosing effects. (This analysis eliminates two inhalation subjects who experienced "highs" that caused them to sleep through lunch and had the lowest daily intakes.) The investigators conclude that inhalation of THC led to more consistent elevations of plasma drug concentration and tended to promote intake, although due to the sample size the results were not statistically significant. They note, as well, that "because this form of delivery requires smoking the drug and is not currently legal, it is objectionable to many patients." It is evident that patients with HIV infection are risking potential legal consequences in hopes of reversing the devastation of the wasting syndrome, and that they and their health care providers would clearly benefit from concrete information on the potential risks and benefits of this controversial treatment modality.


Investigations conducted in the General Clinical Research Center (GCRC) at San Francisco General Hospital by members of our group have already made significant contributions toward the understanding and treatment of HIV-associated wasting. Using many of the same techniques and measurements being proposed in this application, Drs. Schambelan and Mulligan first defined the anabolic effects of recombinant human growth hormone (rhGH) for treatment of wasting (Mulligan 1993) (see Appendix B). The initial pilot study investigated six HIV positive men with a mean weight loss of 19% of their baseline body weight treated with rhGH 0.1 mg/kg/day over seven days of treatment. Six healthy seronegative controls without weight loss were also evaluated. All subjects were hospitalized in the GCRC where they consumed a constant metabolic diet. The trial was divided into a 5-day precontrol period followed by successive 7 day baseline and rhGH treatment periods. The endpoints included the effect of rhGH on body weight, nitrogen and electrolyte excretion, energy expenditure, substrate oxidation, and integrated lipid and carbohydrate metabolism. Results included a prompt and progressive increase in body weight during the seven days of rhGH treatment of 2.0+/- 0.3 and 1.6+/-0.2 kg in HIV positive and HIV negative controls, respectively. Urinary nitrogen excretion declined within the first 24 hours, with a sustained decrease throughout the rhGH treatment period. Resting energy expenditure was shown to increase by 7.5% in both groups. The authors concluded that short term rhGH treatment increased both protein anabolism and protein-sparing lipid oxidation, effects that would likely lead to an increase in body cell mass if sustained during chronic therapy. This, in fact, was realized in the 178 patient, multi-center, double-blind, placebo-controlled Phase III trial that was subsequently conducted under the leadership of members of our group (Schambelan 1994). Using dual energy X-ray absorptiometry (DEXA) evaluations, the authors determined that twelve weeks of rhGH treatment led to a significant and sustained increase in weight (+1.6 +/-0.5 kg) and an even greater increase in lean body mass (+3.0 kg +/-0.4 kg), accompanied by a 1.7 +/-0.2 kg decrease in fat mass. In contrast, the placebo group showed no change in weight or body composition compared to baseline. Treadmill work performance increased significantly only in the rhGH treated group and correlated directly with changes in lean body mass, as did quality of life measures that reflected physical activity.

As a substudy of the multicenter trial, Drs. Mulligan and Schambelan conducted an assessment of the role of energy intake in changes in weight and body composition that occur with rhGH therapy (Tai 1996). Twelve subjects enrolled at San Francisco General Hospital kept 7-day food diaries prior to and at the end of three months of treatment. In this subset of patients, weight and fat-free mass increased and fat decreased as was observed in the larger nationwide cohort. These changes in body composition occurred in the absence of any significant increases in energy, protein, fat or carbohydrate intake. A mean overall 200 kcal/day/increase in energy intake was observed, but this was not statistically significant. Hence, the authors concluded that in contrast to appetite-stimulating therapies, treatment with rhGH resulted in weight gain in the absence of any substantial net increase in energy intake. The loss of fat, coupled with persistent increases in lipid oxidation, suggests that utilization of endogenous fat stores as a supplemental source of fuel synthesis and maintenance of fat-free mass occurs during rhGH treatment.

1. Qualifications of investigators

a. Donald I. Abrams, M.D.: Dr. Abrams is a Professor of Medicine at the University of California, San Francisco and the Assistant Director of the AIDS Program at San Francisco General Hospital. Dr. Abrams has been involved in the care of individuals with HIV infection since clinical manifestations of the epidemic became apparent in San Francisco in 1981. He was involved in the early description of the epidemiology and clinical features associated with AIDS and co-authored a number of studies describing the first recognized immune and hematologic abnormalities in patients with Kaposi's sarcoma and opportunistic infections. He is recognized as one of the first individuals to describe the syndrome of persistent generalized lymphadenopathy (PGL). He has numerous publications which define this AIDS-related condition and which describe the natural history and progression of PGL to AIDS.

Dr. Abrams is also chairman of the Community Consortium, an association of more than 250 health care providers who care for the majority of patients with HIV disease in the San Francisco Bay Area (see also page 65). Dr. Abrams has been the chairman of the Consortium since its inception in 1985, and he was a prime motivator in the development of the Consortium's community-based clinical trials program. Dr. Abrams was a co-principal investigator and designer of one of the Community Consortium's first studies -- an evaluation of aerosolized pentamidine for prophylaxis of Pneumocystis carinii pneumonia (PCP) which ultimately led to the FDA's granting of an indication for aerosolized pentamidine as primary and secondary prophylaxis of PCP. This was the first approved regimen for PCP prophylaxis and resulted in a lead article published in the New England Journal of Medicine (Abrams 1990). Dr. Abrams is also the chair of the HIV Therapies Committee of the Community Programs for Clinical Research on AIDS (CPCRA) (see also page 62). This committee is charged with developing and implementing the antiviral therapies agenda of this national NIH-funded, multicenter, collaborative clinical research program. Dr. Abrams served as protocol chair of CPCRA 002, which compared ddI to ddC in HIV-infected patients who were intolerant of or had failed zidovudine therapy. The results of this study also resulted in a lead article in the New England Journal of Medicine (Abrams 1994).

Dr. Abrams became interested in therapeutic approaches for patients with HIV-associated wasting in the late 1980's. In 1989, the Community Consortium conducted a small placebo-controlled trial of megestrol acetate in patients with wasting. The study was conducted in collaboration with nutrition experts from the University of California at Berkeley who were interested in evaluating the nature of the weight gained using BIA. More recently, Dr. Abrams has pursued the potential usefulness of anabolic steroids in patients with wasting syndrome. In 1995, he designed a study to evaluate the effect of testosterone replacement on lean body mass in men with HIV-related hypogonadism (see Appendix C). The trial was designed to evaluate two routes of administration of testosterone (intramuscular and transdermal) compared to placebo in effecting changes in body composition in men with HIV and hypogonadism. This Community Consortium trial was planned in close collaboration with Drs. Mulligan and Schambelan, and included body composition measurements using BIA and DEXA conducted at the GCRC at San Francisco General Hospital. This 24-week outpatient study was launched in October 1995.

Dr. Abrams' interest in wasting therapies has extended to the national level where he is the protocol chaff of the CPCRA's proposed wasting protocol. After participating in an analysis of CPCRA patients which demonstrated the poor prognostic implications of a loss of even 5% of baseline body weight (Gibert 1995; Wheeler 1995), he became involved in developing a study to prevent frank wasting by intervening earlier in the course of weight loss. The CPCRA study was designed to compare the impact of two anabolic agents: recombinant human growth hormone [Serostim, Serono Laboratories, Norwell, MA] and oxandralone (an oral anabolic steroid) [Oxandrin, Biotechnology General, Iselin, NJ] (See Appendix D). In a factorial design, patients with a weight loss of at least 5% from baseline will be randomized to either subcutaneous injections of growth hormone or oral oxandralone with the addition of either megestrol acetate [Megace, Bristol-Myers Squibb Company, Princeton. NJ] or placebo. The primary endpoint of interest is survival and progression of disease. A subset of patients will undergo sophisticated body composition measurements. However, the primary objective is to determine whether these weight increasing interventions actually translate into a significant additional clinical benefit by slowing disease progression or prolonging survival.

b. Morris Schambelan, M.D.: Dr. Schambelan is a Professor of Medicine at the University of California, San Francisco, Chief of the Division of Endocrinology and Program Director of the GCRC at San Francisco General Hospital. Dr. Schambelan has extensive experience conducting metabolic ward studies and, over the past five years, has conducted a series of studies of the mechanisms underlying HIV-associated wasting and detailed investigations of the effects of anabolic and anticytokine therapies in this setting. Dr. Schambelan is protocol chair for the first study in patients with HIV-associated wasting conducted by the AIDS Clinical Trials Group of the National Institute of Allergy and Infectious Diseases.

c. Kathleen Mulligan, Ph.D.: Dr. Mulligan is Assistant Adjunct Professor of Medicine at the University of California, San Francisco. Dr. Mulligan is a co-investigator with Dr. Schambelan in studies of body composition and energy metabolism in patients with HIV-associated wasting. As part of her doctoral research in nutrition sciences, Dr. Mulligan designed and conducted studies of energy balance that included rigorous assessment of energy intake and food choices. She is protocol chair for a study specifically targeting women with HIV-associated weight loss conducted by the AIDS Clinical Trials Group of the National Institute of Allergy and Infectious Diseases.

d. Neal Benowitz, M.D.: Dr. Benowitz is Professor of Medicine, Psychiatry and Biopharmaceutical Sciences and Chief, Division of Clinical Phamacology and Experimental Therapeutics at the University of California, San Francisco. He was recently elected president of the American Society for Clinical Pharmacology and Therapeutics. Dr: Benowitz has devoted his career to clinical investigation of drugs of abuse. While his current research interest is focused primarily on nicotine and tobacco, in the 1970's he conducted a number of studies on the clinical phamacology of THC and marijuana. He has published 10 research papers and several book chapters on the human pharmacology of THC. He is an expert in drug metabolism, pharmacokinetics and pharmacodynamics, the pharmacology of the inhaled drugs of abuse in general, as well as clinical aspects of substance abuse.

e. Philip Hopewell, M.D.: Dr. Hopewell is Chief of the Division of Pulmonary and Critical Care Medicine at San Francisco General Hospital and Professor of Medicine at the University o.f California, San Francisco. He was recently elected president of the American Thoracic Society. Dr. Hopewell has considerable experience in clinical investigations involving patients with HIV infection. In addition, he was the principal investigator of a 5-year study comparing various modalities of treatment for chronic obstructive pulmonary disease in which pulmonary function testing such as described in the proposed study served to define major outcome variables.

f. Joseph M. McCune, M.D., Ph.D.: Dr. McCune is an Associate Investigator at the J. David Gladstone Institute of Virology and Immunology, an Associate Professor of Medicine, University of Califomia, San Francisco and the Associate Director of the General Clinical Research Center at San =46rancisco General Hospital. He is also an attending physician in the AIDS Clinic at SFGH, and the supervisor of the Gladstone Institute Flow Cytometry Unit. His research focuses on the pathogenic mechanisms of HIV disease, particularly as they relate to hematopathology and immunopathology.

g. Amelia Wilcox, Ph.D.: Dr. Wilcox is an Assistant Clinical Professor and Attending Psychologist, University of California, San Francisco. Dr. Wilcox has a background in the assessment of neuropsychological impairment, including impairment secondary to HIV infection.

h. Alcia Boccellari, Ph.D.: Dr. Alicia Boccellari is an Associate Clinical Professor of Psychology at the University of California, San =46rancisco and Director of thee Division of Psychosocial Medicine and the Neuropsychology Service at San Francisco General Hospital. She has been actively involved since 1985 in conducting research on the neuropsychological effects of HIV infection in both gay men and intravenous drug users. This work has included both longitudinal studies on the natural history and course of HIV, as well as treatment intervention studies with Compound Q and Peptide T. Dr. Boccellari will serve as consultant to the neuropsychology component of this study.

i. Margaret Nettles, Ph.D.: Dr. Nettles is a Senior Psychologist and Group Supervisor in the Positive Education Project at the Center for AIDS Prevention Studies (CAPS), and an Associate Specialist in the Department of Medicine in the UCSF School of Medicine. She is a Licensed Clinical Psychologist in private practice and a Regional Trainer for the American Psychological Association's HIV Office for Psychology Education. At CAPS, she is engaged in a collaborative research group studying HIV/AIDS risk factors and prevention strategies, and the development and evaluation of behavioral treatments of health problems. Areas of expertise include coping interventions and assessment of health quality of life. She has collaborated in the development of disease-specific health quality of life assessment instruments for several clinical trials involving HIV-associated wasting syndrome.

j. Margaret Chesney, Ph.D.: Dr. Chesney is a Co-Director of the Center for AIDS Prevention Studies (CAPS) and Professor of Medicine in the School of Medicine at the University of California, San Francisco. Dr. Chesney is widely recognized for her research on the development and evaluation of cognitive-behavioral interventions for chronic diseases, including AIDS. Much of her work over the last 15 years has focused on behavioral factors in clinical trials. She chaired the Working Conference on Adherence in AIDS Clinical Trials and is the Co-chair of the Outcomes Subcommittee on Recruitment, Retention and Adherence for the AIDS Clinical Trials Group of the National Institute of Allergy and Infectious Diseases, which examines issues ranging from quality of life to health care costs. Prior to joining UCSF in 1987, Dr. Chesney had over 11 years experience conducting epidemiologic research and clinical trials as the Head of the Department of Behavioral Medicine at Stanford Research Institute. Dr. Chesney has authored or co-authored over 130 scientific publications and book chapters, and her areas of expertise include behavioral science and coping with HIV. Dr. Chesney will serve as consultant for the quality of life assessments in this study.

2. Organizational Expertise

a. General Clinical Research Center: The inpatient study will be performed at the General Clinical Research Center (GCRC) at San Francisco General Hospital. The GCRC is a self-contained organized research unit that is fully staffed by nursing, dietary and core laboratory personnel to support the clinical research project described herein. Facilities at the GCRC include a metabolic kitchen where all meals will be prepared, and dietary intake quantified by research dieticians. Dr. Morris Schambelan is the Program Director of the GCRC.

A dual-energy X-ray absorptiometer (DEXA; Lunar model DPX, Madison, WI) and bioelectrical impedance analyzer (BIA; RJL, Clinton Twp, MI) that belong to Dr. Schambelan's research group will be used for body composition measurements. A room equipped with a Tepco 750 smoke scrubber is available on the GCRC for smoking studies. However, as this room is frequently in use for studies of tobacco and nicotine, funds are requested for a HEPA filter (BioSafety, RS 1000) so another room in the GCRC can be equipped for the proposed study.

b. Community Consortium: The Community Consortium was formed in 1955 and is part of the University of California, San Francisco AIDS Program at San Francisco General Hospital. Its membership includes over 250 physicians, nurses, physician assistants and other licensed health care providers who work at publicly-funded clinics, academic medical centers and health maintenance organizations. Collectively, members of the Community Consortium care for the majority of people with HIV disease in the San =46rancisco Bay Area. Dr. Donald Abrams is the chairman of the Community Consortium.

In 1988, the Community Consortium established a "community-based" clinical research program. Such studies are designed to be conducted in private physicians' offices, hospital-based outpatient clinics and community health centers. In 1989, the Consortium was awarded a 5-year contract from the Community Programs for Clinical Research on AIDS (CPCRA) of the National Institute of Allergy and Infectious Diseases to participate in federally-sponsored, multi-center, community-based clinical trials of promising experimental therapies for HIV/AIDS. In 1994, the Consortium successfully recompeted for a second 5-year contract from the CPCRA. To date, over 65 primary care providers at more than 20 sites in San Francisco, Alameda and Marin counties have enrolled over 2200 patients with HIV disease onto Consortium clinical trials.

Consortium clinical research staff will be responsible for recruitment and screening of subjects, coordinating all study-related measurements and procedures, conducting clinical evaluations of patients, as well as data collection, data management and data analysis.


1. General Study Design: The inpatient study will be a randomized, double-blind, placebo-controlled, within-subjects evaluation of smoked marijuana in persons with HIV disease and mild-to-moderate weight loss. A total of 15 patients will be recruited for this study. They will be randomly assigned to one of two groups receiving either moderate (~4%) THC-content marijuana cigarettes or placebo marijuana cigarettes. The inpatient study will be composed of five successive "stages". The first stage will be a 5-day lead-in phase in which baseline measurements will be obtained while patients are hospitalized in the General Clinical Research Center (GCRC) at San Francisco General Hospital. This will be followed immediately by a 5-day period in the GCRC in which patients receive either active THC-content marijuana cigarettes or placebo marijuana cigarettes. This stage will be followed by a wash-out period of four weeks in which patients are not hospitalized. The wash-out period will be followed immediately by another 5-day lead-in period in the GCRC, followed by a second 5-day period in the GCRC in which patients are crossed-over to placebo marijuana cigarettes or active THC-content marijuana cigarettes, depending upon their initial randomization allocation. Thus, each subject will experience three experimental conditions: lead-in, moderate (~4%) THC-content smoked marijuana, and placebo marijuana.

A. Use of placebo control: Most of the previous studies of the effects of cannabinoids on appetite have utilized placebo controls. A placebo-controlled design is proposed herein to diminish the potential for bias in the ascertainment of certain outcomes. For example, subjects' food intake patterns may differ between the first 5-day period (the lead-in phase) and the second 5-day period (the intervention phase) due to their adjustment to hospitalization in the GCRC, rather than to the intervention(s). Similarly, as there is such an abundance of anecdotal evidence that marijuana smoking increases appetite, subjects might be inclined to increase their caloric intake due to expectation rather than a true physiologic effect. Finally, the use of the placebo control will enable us to provide accurate estimates of the side effects of smoked marijuana, which would be necessary to any scientific consideration of its widespread use for treatment of HIV-associated anorexia and wasting.

B. Choice of single dose/strength of marijuana: Plasma THC levels can vary considerably among subjects smoking the same strength of marijuana, depending on how deeply subjects inhale, how long they hold their breath before exhaling, how many inhalations they take, and the length of time they take to smoke each marijuana cigarette (Agurell 1986; Azorlosa 1992). In addition, smokers may smoke marijuana differently to obtain desired doses of THC. Thus, although it is possible under well-controlled, experimental conditions to standardize (to some degree) how patients smoke marijuana, such standardization is unlikely to be achieved in clinical practice. Moreover, such variation in plasma THC levels among study subjects that can result from differences in their smoking patterns makes comparisons in outcomes in patients smoking different strengths of marijuana difficult to interpret. For these reasons, we have chosen to evaluate a single strength (~4%) of marijuana in this investigation, rather than to conduct a dose-ranging study of two or more doses/strengths of marijuana. Marijuana of this strength most closely approximates that which is commonly available "on the street" and can readily be supplied by the National Institute of Drug Abuse. Dose response issues may be addressed in part by examining the relationship between THC levels and the primary endpoints.

C. Rationale for specific measurements: A variety of measurements of the safety and efficacy of smoked marijuana will be used in the inpatient study. Some of these measurements are best obtained only under well-controlled, experimental conditions in an inpatient setting; others are more relevant to the chronic use of marijuana and can easily be obtained in an outpatient setting. The measurements that are best obtained in an inpatient setting include assessments of appetite, daily food intake, resting energy expenditure and neuropsychological functioning. Laboratory data indicate that the effects of' smoked marijuana on these parameters occurs relatively quickly and lasts from 2-8 hours. The inpatient setting also permits us to measure plasma THC levels as a means to assess the total dose delivered. Measurements that are more relevant to the chronic use of marijuana include body composition, pulmonary function measured by spirometry, testosterone levels, HIV RNA, lymphocyte function and quality of life. These measurements are proposed in the inpatient study to determine if there are any measurable short-term effects of marijuana on these parameters. More importantly, they will provide a complete picture of the safety profile of smoked marijuana when used as an appetite stimulant for HIV-associated wasting when data from both the inpatient and outpatient studies are available for analysis and dissemination.

1. Appetite: Self reported hunger will be assessed using Visual Analog Scales, as described by Hetherington and Rolls. Because previous studies (Foltin 1986) suggest that cannabinoids increase snack food intake, the Hetherington and Rolls VAS items will be supplemented with items inquiring about hunger for various types of foods, as suggested by deWit (deWit 1996) (see Appendix D). The VAS items are sensitive to small changes in ratings, and can be used repeatedly with the same participant, thus allowing for frequent assessment of self-reported hunger (Schoeller 1995).

2. Food intake: Obtaining accurate quantitative food intake data has long presented a challenge to researchers in human nutrition (e.g., Schoeller 1995; Bingham 1987). Use of recall methods frequently results in distortion, substitution, or omission of food items consumed and, overall, underestimation of daily food intake (Acheson 1980). Errors in estimating portion sizes have been found to typically range from 20-50% (e.g., Bingham 1987). Frequently, such errors result in overestimation of consumption of food items considered to be healthful and underestimation of consumption of "bad" foods. More quantitative information can be obtained when research subjects actually weigh and measure food items consumed, but such activity frequently alters eating behavior and, thus, provides inaccurate estimates of"habitual" intake. Certainly, it is conceivable that use of marijuana or dronabinol could further compromise the reliability of food intake data obtained by either recall or food intake diaries. Notably, none of the published studies of dronabinol in patients with HIV infection have included quantitative data on energy intake. Although patients in these studies subjectively report increased appetite, no corresponding increases in weight have occurred.

Based on these considerations, the controlled inpatient setting in the proposed study provides a unique opportunity to rigorously evaluate the effects of smoked marijuana on self-selected food intake in a way that requires a minimum of subject cooperation. In previous studies of marijuana in healthy subjects, similar controlled approaches have been use to demonstrate significant effects of marijuana on food intake (e.g., =46oltin, 1988; Greenberg, 1976; and Foltin, 1986).

3. Body composition: The composition of changes in weight, should then, occur during either the lead-in or intervention periods, will be characterized by perforating measurements of dual-energy X-ray absorptiometry (DEXA) and bioelectrical impedance analysis (BIA).

4. Resting energy expenditure: We have observed that resting energy expenditure (REE), which is elevated in many patients with HIV infection, accounted for approximately two-thirds of the daily energy requirement in a group of HIV-infected men housed in the GCRC and fed weight-maintaining diets (Mulligan 1993). Thus, quantitative information on the effect, if any, of marijuana on this important component of total daily energy expenditure will contribute to our evaluation of the effects of marijuana on overall energy balance. To date, we are unaware of any published results of studies of the effects of marijuana on REE. Certainly, it is known that REE is affected by a wide variety of other drugs. For example, significantly increased rates of REE have been reported in response to diazepam (Kirvela 1994), scopolamine/morphine (Kirvela and Kanto 1992), and fentanyl/flunitraxepam (Binder 1991), whereas REE increases with nicotine (Arcavi 1994) and caffeine (Arciero 1995). The design of the proposed study provides an excellent opportunity to measure the acute effects of marijuana on REE and to evaluate its potential effects on energy balance.

5. THC levels: Plasma delta 9 THC levels will be measured by the Center for Human Toxicology at the University of Utah. The assays will be performed by Gas Chromatography - Mass Spectrometry. With a 1 ml sample, THC levels of 0.5 mg/ml can be measured, which is well below the expected levels after smoking a marijuana cigarette. The Center for Human Toxicology has a long-standing contract with NIDA for THC assays. The trough and peak levels will be used to construct a daily estimate of exposure. We recognize that sampling only the trough and peak concentrations is not adequate to fully define the area under the plasma concentration time curve (which is the optimal estimate of exposure), but due to the large number of other blood samples that must be taken, we felt limited to collect only six samples for THC measurement per day. We will be able to construct from these data a crude dose estimation. Expired carbon monoxide (CO) levels will be measured by using an Ecolyzer. Expired CO levels throughout the day will provide another measure of smoke exposure. 6. Neuropsychological functioning: All subjects will receive evaluations which will include a structured demographic interview, the Profile of Mood States (POMS), and serial neuropsychological testing. Demographic and psychiatric measures to be included in the full battery include:

Two special concerns were taken into consideration in the development of the neuropsychological component of this study:

Neuropsychological measures to be administered include:

The above items were selected for inclusion in the present study because they are standardized measures which assess the function of a variety of cognitive domains (i.e., attention and concentration, visual scanning ability, visual learning ability, immediate and delayed verbal memory for word lists, executive functions, visual motor skills, motor speed, coordination and strength, and reaction time). The test items are among the neuropsychological measures less influenced by educational level. Many of these cognitive functions have been previously explored in marijuana administration studies (see above), and several have yet to be addressed. None has yet been addressed in a marijuana administration study conducted on a sample of HIV-positive individuals who are receiving the substance in a controlled, hospital setting.

The administration of all measures to be included in the battery is expected to take approximately 40 minutes. These procedures are expected to represent minimal risk to subjects. Each subtest is brief, and the potential for provoking anxiety or frustration is kept to a minimum.

7. Spirometry: Because airways obstruction has been the only consistent lung function abnormally detected in marijuana smokers, we will focus on measurements of forced expiratory flows. Subjects will be instructed not to use any bronchodilator medications for the 12 hours before the test. Measurements will be made with subjects seated and a nose clip in place. At least three full forced expiratory maneuvers will be obtained according to the criteria of the American Thoracic Society. A SensorMedics 2450 Pulmonary Function Laboratory Unit will be used for the measurements. If there is evidence of airways obstruction, a second set of measurements will be obtained 15 minutes after inhalation of two puffs of albuterol. Data obtained will include the forced vital capacity (Fvc), forced expiratory volume in 1 second (FEV 1), peak flow rate (PFR) and forced expiratory flows (FEF) at 25%, 50% and 75% of the forced vital capacity. Because of their reproducibility and better standardization, the main outcome measures used will be the FEV 1 and the FEV1/FVC.

8. Testosterone levels: Serum testosterone levels have been noted to decrease in men with chronic marijuana use, although the time course of this decline has not been established. In the proposed inpatient phase of our investigations, we will examine the short-term effect of marijuana smoking on tho hypothalamic-pituitary-gonadal axis by measuring total testosterone, luteinizing hormone and follicle stimulating hormone levels utilizing assays available in the GCRC core laboratory.

9. Lymphocyte function: We will use many of the same assays that have been used previously in the literature addressing the effects of marijuana on immune function. These assays (listed below) are relatively straightforward and easy to standardize. Moreover, data derived from them will be directly comparable to previous data reported in the literature. In the case of each assay, we propose to minimize the effects of inter- and intra-assay variation by obtaining three baseline determinations from each subject and then three additional readings at the end of each phase of placebo and active treatment. The assays that will be run for all subjects include:

a. Mitogen stimulation: Peripheral blood mononuclear cells (PBMC) will be purified by ficoll hypaque density sedimentation, incubated in tissue culture medium (RPMI 1640 medium supplemented with 10% fetal calf serum), and stimulated with phytohemagglutinin. Proliferation indices will be read at day 7 by measuring incorporation of [3]H-thymidine.

b. Mixed lymphocyte culture: PBMC will be co-cultivated with irradiated, allogenic stimulator cells in a 1:1 ratio and incubated for 7 days in tissue culture medium. Stimulation indices will be determined at day 7 by measuring incorporation of [3]H-thymidine.

c. Cytokine production: The ability of PBMC from subjects to produce "Th1 cytokines" (e.g., IL2, FNy) or "Th2 cytokines" (e.g., IL4, IL6) will be assayed by several approaches. In one, cells will be stimulated in vitro with PMA and ionomycin for 24 hours and the tissue culture supernatant will be assayed for the presence of secreted cytokines by ELISA. Simultaneously, cells from the stimulated cultures will be assayed for the presence of cytokine ( IL2, TNF, IL4, IL6, ILS and gamma interferon) mRNA using a bDNA-based assay format (Chiron). In a second, single-cell approach, PBMC will be either stimulated with PMA and ionomycin or left untreated for 4 hours, in the presence of brefeldin A (an inhibitor of protein secretion). After permeabilization of the cells, accumulated intracellular cytokine will be detected by flow cytometry using fluoresceinated anti-cytokine antibodies.

10. HIV viral load: Analysis of HIV RNA will be determined utilizing the branched DNA (bDNA) technology (Chiron Corporation, Emeryville, CA). The Chiron "ultrasensitive" second generation assay will be utilized. This method is able to detect as few as 500 copies/ml of HIV viral RNA. The "ultrasensitive" assay correlates well with the currently available Chiron bDNA assay which has 10,000 copies/ml as its lower limit of detection. Baseline determinations will be collected in duplicate to assure greater accuracy of the initial determination. From a recent Community Consortium trial evaluating the impact of antiretroviral therapies utilizing the Chiron bDNA assay (Follansbee 1996), we know that significant responses of viral load to therapeutic interventions can occur within 7 days of the initiation of treatment; hence determinations after five days of smoking marijuana or placebo could be expected to show variation if a significant effect occurs.

11. Quality of life: Quality of life assessment is considered an important component of the evaluation of patient outcomes, especially for HIV infection where the progress of the disease is debilitating and painful (Bozzette 1995; Lubeck 1992). Moreover, functional outcomes are significantly associated with HIV mortality, even after controlling for CD4 count (Stanton 1994). Thus, quality of life variables are an important outcome in the proposed study. We have selected a set of self-report scales that assess psychological functioning, a key component in quality of life, and one that may be influenced by the treatment under investigation. Included among the quality of life variables to be assessed are depression, anger, fatigue, anxiety and positive well-being (see Appendix F).

2. Statistical Considerations

The goal of the inpatient study is to broadly estimate the safety and potential effectiveness of smoked marijuana when used to treat patients with mild-to-moderate HIV-associated weight loss. If the results of this initial trial indicate an absence of significant adverse effects of smoked marijuana in this population, then estimates of effect-size and response variability obtained from this study will be used to plan and implement a larger, comparative trial evaluating smoked marijuana to dronabinol in an outpatient setting, with sufficient power to answer definitively the question of efficacy.

A total of 15 subjects will be evaluated in this study. This number represents a balance between the number of subjects needed to broadly estimate the safety and potential efficacy of the intervention, and the feasibility of conducting an intensive, inpatient study of HIV-infected patients.

The choice of a cross-over design was motivated by the need to maximize available physical resources within the GCRC, and to increase the acceptability of an inpatient placebo-controlled trial to potential study subjects. This design uses each subject as their own control, under the assumptions of a limited intervention order effect and an adequate wash-out period. Allowing each subject to serve as their own control will serve to reduce bias due to individual subject characteristics. This design will also provide data to investigate within and across patient variability, the importance of intervention order and the strength of carryover effects.

The primary outcome used for power calculation for this study is change in appetite as measured by change in average energy intake (lcal/Kg/day). An initial lead-in period will be observed to estimate energy intake in the inpatient setting. An average of energy intake in kcal/Kg/day over the last three days of the lead-in phase will be used to create the baseline measurement. It is thought that the last three days of the lead-in phase will represent a stabilization of energy intake in the new environment.

Similarly, an average of energy intake in kcal/Kg/day over the last three days of the intervention phases will be used to create the response measurement. The definition of a significant response for this study will be an 8 kcal/Kg/day increase in the average energy intake compared to the lead-in period. In a 65 kg person this works out to approximately 500 kcal/day or approximately 20% of estimated daily energy requirements (estimated roughly as 1.5 x resting energy expenditure).

Estimates of individual variability in energy intake used in sample size calculations were obtained from a study of 13 healthy controls utilizing the same facilities that are to be used in the proposed study. These values were inflated by 15% to adjust for the reduced health of the study subjects. Variability in individual response to placebo or active treatment was assumed to be 150% greater than the individual variability and equal for both intervention groups.

This study has been powered such that with 15 evaluable patients, the probability that there exists a greater than 20% response rate when no responses were observed is less than 5%.

3. Study Population: Fifteen subjects with HIV infection and mild-to-moderate (5-15%) weight loss will be recruited for this study. We chose to include patients with some weight loss to ascertain whether there may be a trend towards improvement in the appetite and food intake parameters in patients with the condition of interest, rather than restrict the study to relatively healthy HIV-positive subjects with minimal (<5%) or no weight loss. We have limited the total amount of weight loss so as not to enroll subjects too debilitated to complete the study.

A. Subject Selection

Inclusion Criteria: Exclusion Criteria:

B. Subject Recruitment

C. Treatment Plan