Cardiovascular and Metabolic Risk ORIGINAL ARTICLE

Metabolic Effects of Chronic

1 1 RANGANATH MUNIYAPPA, MD, PHD GAIL HALL, RN Consistent with these findings, clinical in- 1 1 SARA SABLE, BS KONG Y. CHEN, PHD 2 6 terventional trials suggest that CB1R antag- RONALD OUWERKERK, PHD NORA D. VOLKOW, MD 3 7 onism reduces body weight, improves ANDREA MARI, PHD GEORGE KUNOS, MD, PHD 2 8 dyslipidemia, and attenuates insulin resis- AHMED M. GHARIB, MD MARILYN A. HUESTIS, PHD 4 1 tance in humans (6,7). MARY WALTER, PHD MONICA C. SKARULIS, MD 5 D D AMBER COURVILLE, PHD, RD 9- ( 9-THC), the primary psychoactive component of , activates peripheral and OBJECTIVEdWe examined if chronic cannabis smoking is associated with hepatic steatosis, central CB1Rs (2). Acute treatment of insulin resistance, reduced b-cell function, or dyslipidemia in healthy individuals. healthy human volunteers with cannabis induces glucose intolerance (8,9). Simi- RESEARCH DESIGN AND METHODSdIn a cross-sectional, case-control study, we larly, short-term (13 days) studied cannabis smokers (n = 30; women, 12; men, 18; 27 6 8 years) and control subjects smoking increases appetite, food intake, n 6 ( = 30) matched for age, sex, ethnicity, and BMI (27 6). Abdominal fat depots and intrahepatic and body weight in healthy men (10). In fat content were quantified by magnetic resonance imaging and proton magnetic resonance b individuals with chronic hepatitis C, daily spectroscopy, respectively. Insulin-sensitivity indices and various aspects of -cell function were marijuana abuse was associated with a derived from oral glucose tolerance tests (OGTT). higher risk for hepatic steatosis (11). These RESULTSdSelf-reported cannabis use was: 9.5 (2–38) years; joints/day: 6 (3–30) [median findings from animal and human studies (range)]. Carbohydrate intake and percent calories from carbohydrates, but not total energy suggest that chronic cannabis use and fi intake, were signi cantly higher in cannabis smokers. There were no group differences in percent CB1R activation may negatively affect total body fat, or hepatic fat, but cannabis smokers had a higher percent abdominal visceral fat metabolic actions of insulin and facilitate (18 6 9vs.126 5%; P = 0.004). Cannabis smokers had lower plasma HDL cholesterol (49 6 14 6 P hepatic steatosis. Cannabis (marijuana) is vs. 55 13 mg/dL; = 0.02), but fasting levels of glucose, insulin, total cholesterol, LDL the most commonly used illicit in cholesterol, triglycerides, or free fatty acids (FFA) were not different. Adipocyte insulin resistance the U.S. with ;17 million current users index and percent FFA suppression during an OGTT was lower (P , 0.05) in cannabis smok- ers. However, oral glucose insulin sensitivity index, measures of b-cell function, or incretin con- (12). Despite such widespread use, rela- centrations did not differ between the groups. tively little is known about the metabolic effects associated with chronic cannabis CONCLUSIONSdChronic cannabis smoking was associated with visceral adiposity and use. To that end, in a cross-sectional, adipose tissue insulin resistance but not with hepatic steatosis, insulin insensitivity, impaired case-control study, we examined if chronic b pancreatic -cell function, or glucose intolerance. cannabis smoking is associated with he- patic steatosis, insulin resistance, reduced b-cell function, and dyslipidemia in healthy individuals. annabinoid receptors (CB1Rand CB1Rs are present in peripheral tissues in- C CB2R) and their endogenous li- volved in energy homeostasis, such as adi- gands, the endocannabinoids, play pose tissue, liver, skeletal muscle, and RESEARCH DESIGN AND an important role in regulating energy pancreas (1,2). CB1R activation promotes METHODS balance, appetite, insulin sensitivity, lipogenesis in the liver and adipose tissue pancreatic b-cell function, and lipid me- (1,2,5), reduces insulin responsiveness in Study design and study subjects tabolism (1–4). Endocannabinoids (anan- skeletal muscle (3), and impairs insulin ac- This cross-sectional, case-control study damide and 2-arachidonoyl glycerol) and tion and secretion in pancreatic b-cells (4). was conducted at the Clinical Research ccccccccccccccccccccccccccccccccccccccccccccccccc Center, National Institutes of Health (NIH), in Bethesda, MD (ClinicalTrials. 1 From the Diabetes, Endocrinology, and Obesity Branch, National Institute of Diabetes and Digestive and gov Identifier: NCT00428987) and the Kidney Diseases, Bethesda, Maryland; the 2Biomedical and Metabolic Imaging Branch, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, Maryland; the 3Institute of Biomedical Engineering, Johns Hopkins Behavioral Pharmacology National Research Council, Padova, Italy; the 4Clinical Core Laboratory, National Institute of Diabetes and Research Unit (BPRU) in Baltimore, MD. Digestive and Kidney Diseases, Bethesda, Maryland; the 5Clinical Center, National Institutes of Health, The study protocol was approved by the Bethesda, Maryland; the 6Intramural Research Program, National Institute on Drug Abuse, National In- 7 NIH Central Nervous System Institu- stitutes of Health, Baltimore, Maryland; the Laboratory of Physiologic Studies, National Institute on Al- tional Review Board and Institutional Re- cohol Abuse and Alcoholism, National Institutes of Health, Rockville, Maryland; and the 8Intramural Program, National Institute on Abuse and Alcoholism, National Institutes of Health, Bethesda, view Boards of the National Institute of Maryland. Diabetes and Digestive and Kidney Dis- Corresponding author: Monica C. Skarulis, [email protected]. eases-National Institute of Arthritis and Received 6 November 2012 and accepted 5 February 2013. Musculoskeletal and Skin Diseases, and DOI: 10.2337/dc12-2303. Clinical trial reg. no. NCT00428987, clinicaltrials.gov. © 2013 by the American Diabetes Association. Readers may use this article as long as the work is properly all procedures followed were in accor- cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ dance with institutional guidelines. Writ- licenses/by-nc-nd/3.0/ for details. ten informed consent was obtained from care.diabetesjournals.org DIABETES CARE 1 Diabetes Care Publish Ahead of Print, published online March 25, 2013 Metabolic all subjects. Adults who smoked cannabis and orange vegetables and legumes, (mg/dL). OGIS was calculated from at least 4 days per week for the last 6 milk, meat and beans, oils, saturated fat, OGTT using the 3-h OGIS equation months, with a self-reported history of sodium, and calories from solid fats, alco- (18). HIRI was calculated as the product cannabis smoking for at least 1 year, and holic beverages, and added sugars. HEI of total area under curve (AUC) for glu- not seeking active treatment were re- scores range from 0–100, with a higher cose and insulin during the first 30 min of cruited by the National Institute on score indicating a better-quality diet. the OGTT (19). The trapezoidal method Drug Abuse. Exclusion criteria were de- Assessment of body composition, adi- was used to calculate AUC. HIRI = (G 0–30 3 pendence on other than , pose tissue distribution, and hepatic [AUC] I 0–30[AUC]). Adipose tissue sensi- , and cannabis within the past 6 fat content. Body weight was measured tivity to the antilipolytic actions of insulin months of screening visit; history of in- using a digital balance (Scale-Tronix were assessed by the adipocyte insulin re- travenous illegal drug use; alcohol intake 5702; Scale-Tronix, Carol Stream, IL). sistance index (AIRI) (21) and the degree of more than five drinks per day on $3 Body composition was measured by dual- of suppression of plasma FFA during an days in a week (.15 drinks/week); his- energy X-ray absorptiometry with a Lunar OGTT (22). AIRI is the product of the tory of hepatitis B or C or current hepatitis iDXA scanner (GE Healthcare, Madison fasting plasma insulin and fasting plasma A, B, or C; history of diabetes (type 1 or WI). Subcutaneous and visceral adipose FFA (log-transformed) concentrations. type 2), polycystic ovary disease, or other tissue volumes were assessed at the L4–L5 Suppression of lipolysis during an OGTT conditions that may confound study out- spine level by magnetic resonance imag- was assessed as FFAAUC adjusted for comes; professional or collegiate athletes ing using manual image segmentation mean plasma insulin concentrations or active participation in .60 min/day of on a T1-weighted turbo-spin-echo image and percent maximal suppression of vigorous exercise; use of prescribed or as previously described (15). Hepatic fat FFA levels. over-the-counter or herbal/alternative content was measured with single volume b-Cell glucose sensitivity, rate sensi- medications/preparations with effects on magnetic resonance spectroscopy (MRS) tivity, and the potentiation factor were glucose and lipid metabolism; and preg- in an 8 ml volume in the posterior right calculated with the model as described nancy. lobe of the liver as described previously previously (23). In addition, insulino- Cannabis smokers underwent history (16). genic index (IGI) was derived from and physical examination as well as lab- Oral glucose tolerance test. Each sub- plasma insulin and glucose concentra- oratory tests of blood and urine to ensure ject underwent 75-g oral glucose toler- tionsduringanOGTTaspreviouslyde- that they were free of current somatic and ance test (OGTT) after a 10-h overnight scribed (24). psychiatric illness and did not currently fast. Samples for determination of plasma Statistical analyses. The primary out- abuse drugs other than cannabis (or glucose, insulin, C-peptide, free fatty come measures were differences in in- nicotine or alcohol). Eligible participants acids (FFA), and various incretin concen- sulin sensitivity and hepatic fat content were admitted to the BPRU, where they trations were drawn at 210, 0, 15, 30, 45, between age-, sex-, ethnicity-, and BMI- were continuously monitored and not 60, 90, 120, and 180 min relative to the matched chronic cannabis smokers and allowed to leave the unit or receive visi- glucose load. Plasma was obtained from control subjects. There were no prior tors during their stay to exclude illicit blood samples by centrifugation and im- studies examining the effects of chronic drug use. Following an overnight stay at mediately frozen on dry ice for storage at cannabis use on insulin sensitivity or he- BPRU, subjects were transported to the 2 808C. patic fat content in healthy individuals. In NIH Clinical Center for metabolic studies Laboratory assays. Routine assays for a study of hepatic steatosis (assessed by and imaging studies. After the study serum lipids, plasma glucose, insulin, and biopsy) in hepatitis C virus (HCV)–infected procedures, the subjects were transported hemoglobin A1c (A1C) were performed in individuals, 33% of daily cannabis users back to BPRU. Control subjects with no the Department of Laboratory Medicine at had marked steatosis compared with prior history of cannabis use and with the Clinical Center, NIH. FFA were mea- 16% of nonusers (odds ratio of 2.1) (11). negative urine screens for illicit drug use sured with a Wako HR Series NEFA-HR In healthy individuals, we expected the were recruited by National Institute of kit (Wako Diagnostics, Wako Chemicals effects of chronic cannabis smoking on Diabetes and Digestive and Kidney Dis- USA, Inc., Richmond, VA) and run on a hepatic fat content to be larger than ob- eases in Bethesda, MD. Control subjects COBAS FARA-II analyzer (Roche Diagnos- served in HCV-infected individuals. A were matched with cannabis smokers for tics, Indianapolis, IN). Adiponectin, leptin, sample size of n =60,withn =30ineach sex, ethnicity, age (6 5years),andBMI total peptide YY (PYY), active glucagon- cohort was sufficient to provide 80% (6 2kg/m2). like peptide-1 (GLP-1), and acyl and desacyl power with a =0.05indetectingastan- ghrelin levels were measured in plasma by dardized difference of ;0.75 in hepatic Study procedures ELISA. fat content (estimated SD, 7%) (16). Dietary intake and dietary quality. A Modeling and calculations. Quantita- Similarly, a sample size of n = 60 was suf- 24-h dietary recall was used for evaluation tive Insulin Sensitivity Check Index ficient to detect a standardized difference of total daily energy intake (13). The (QUICKI), Oral Glucose Insulin Sensitiv- of ;1.0 in insulin sensitivity by QUICKI Healthy Eating Index (HEI) 2005 was ity (OGIS), and Hepatic Insulin Resis- (estimated SD, 0.037) (20) and OGIS (es- 2 2 used to assess dietary quality. HEI is tance Index (HIRI) were used as surrogate timated SD, 60 mL z min 1 z m 2) (18) based on the food patterns in the 2005 measures of glucose insulin sensitivity/ with a power of 90% and a = 0.05. The Dietary Guidelines for Americans (14). resistance (17–19). QUICKI was calcu- presence of skewed data were evaluated by The score is a composite measure of 12 lated as defined previously from fasting visual inspection of quantile–quantile energy-adjusted components, including, glucose and insulin values (20). QUICKI = plots, stem and leaf plots, or box plots fi total grains, whole grains, total fruit, 1/(log[I0]+log[G0]), where I0 is fasting and veri ed by the Shapiro-Wilk test for whole fruit, total vegetables, dark green insulin (mU/mL) and G0 is fasting glucose normal distribution. After testing data for

2 DIABETES CARE care.diabetesjournals.org Muniyappa and Associates normality, we used Student unpaired participants in the control group were to- resonance imaging. While cannabis t, paired t, or appropriate nonparametric bacco smokers. Systolic and diastolic smokers had a lower total (P = 0.003) tests for evaluation of differences between blood pressures were significantly higher and subcutaneous (P = 0.008) fat in the the groups. Because of the exploratory na- in cannabis users when compared with abdomen, the percent of visceral fat area ture of this study, no adjustments were control participants (systolic blood pres- (visceral fat area/total abdominal fat area) made for multiple comparisons, and val- sure, 128 6 13 vs. 120 6 12 mm Hg, P = was higher (18 6 9vs.126 5%; P = ues of P , 0.05 were considered to rep- 0.01; diastolic blood pressure, 76 6 10 0.004) when compared with the control resent statistical significance. The vs. 71 6 8mmHg,P = 0.01). group (Fig. 1). However, the percent of statistical software StatsDirect version Total daily energy intake, assessed by visceral fat area was unrelated to age or 2.7.2 (Chershire, U.K.) was used for 24-h dietary recall, was not significantly frequency or duration of cannabis use. data analysis. different between the groups. However, Contrary to our hypothesis, hepatic fat carbohydrate intake and percent calories content as assessed by MRS was not dif- RESULTS from carbohydrates were significantly ferent between the groups (1.1 6 1.3 vs. higher in cannabis smokers (Table 1). 1.5 6 1.9%; P = 0.59) (Fig. 1). In the Clinical characteristics, cannabis Diet quality was measured using the combined cohort, age (age: r =0.26;P = use, and energy intake of study HEI-2005. As seen in Table 1, the quality 0.05) and percent TBF (r =0.35;P = subjects of diet consumed by cannabis smokers 0.007) was associated with hepatic fat This study included age-, sex-, race-, and was lower than the control group, reflect- content. Similar to percent visceral fat, he- BMI-matched chronic cannabis smokers ing differing food choices. Despite differ- patic fat content was unrelated to age or (n = 30) and control subjects (n = 30). The ences in total diet quality, the percent frequency or duration of cannabis use. median age of cannabis smokers was 25 total daily calories from fat and alcohol years (25th–75th percentile: 21–28 were not significantly different between Chronic cannabis use, metabolic years). Age at first cannabis smoking was the groups. In both groups, women con- parameters, and insulin sensitivity 15 6 3 years (range, 11–20). Participants sumed less alcohol (% energy intake) than Fasting plasma glucose and insulin as well smoked an average of 10 6 8jointsper men (control subjects: 1.7 6 2vs.56 as a measure of glycemic control, A1C, day (range 3–30) for 12 6 9 years (range, 11; cannabis: 1.9 6 4 vs. 2.7 6 3). were comparable between the groups 2–38). The majority of cannabis smokers (Table 2). Fasting levels of HDL choles- in our cohort were African American Cannabis use, adiposity, and body terol (P = 0.02) were lower in cannabis (73%) and men (60%) (Table 1). The me- fat distribution smokers (women: 49 6 12 vs. 61 6 12 dian BMI of the cannabis smokers was 26 In addition to being age- and BMI- mg/dL; men: 49 6 12 vs. 51 6 12 mg/dL). kg/m2 (range, 19–42). The participants in matched, total body fat (TBF) content However, LDL cholesterol, triglycerides, the control group were matched for age, (regional percent fat) was similar between and FFA levels were comparable between sex, ethnicity, and BMI (Table 1). Seven- the groups (28 6 12 vs. 28 6 11%; P = thegroups(Table2).Likewise,circulat- teen of the cannabis smokers (56%) also 0.75) (Fig. 1). Relative distribution of ab- ing levels of high-sensitivity C-reactive smoked , whereas only two dominal fat was evaluated by magnetic protein(cannabis:women,5.66 9.3 and men, 0.8 6 0.7 mg/L; control subjects: women, 1.5 6 1.9, men, 1.2 6 1.4 mg/L), Table 1dDemographics, cannabis use, energy intake and dietary quality in chronic leptin (cannabis: women, 29.5 6 20.0 cannabis users and control nonusers and men, 5.9 6 8.4 ng/mL; control sub- jects: women, 28.2 6 18.4 and men, 5.1 6 Control subjects Cannabis smokers 5.4 ng/mL), and adiponectin were similar Variables (n =30) (n =30) P value (Table 2). Although, levels of liver trans- aminases, alanine aminotransferase, and Age (years) 27 6 7276 80.47aspartate aminotransferase were not dif- Female/males 12/18 12/18 ferent, plasma levels of alkaline phospha- African American (n)2222tase were significantly higher in cannabis Whites 7 7 smokers (Table 2). BMI (kg/m2)276 6276 60.91Measures of hepatic insulin resis- Obese, BMI .30 (%) 26 26 tance, QUICKI, HIRI, and OGIS, a mea- Cannabis use sure of peripheral glucose disposal, were Age at first use (years) 15 6 3 not significantly different between the Years of use 12 6 9 groups (Table 3). However, indices of ad- Joints per day 9.5 6 8.4 ipose tissue insulin resistance, AIRI, and Energy intake (kcal/day) 2,423 6 2,280 2,980 6 1,689 0.33 FFA suppression index were significantly Protein (% energy intake) 15.3 6 4.3 11.2 6 4.6 0.002 higher in cannabis smokers (Table 3). Carbohydrate (% energy intake) 50.9 6 10.8 58.8 6 15.8 0.02 Fat (% energy intake) 32.2 6 8.5 29.6 6 11.7 0.07 Chronic cannabis use, insulin Alcohol (% energy intake) 3.7 6 8.6 2.4 6 3.2 0.47 secretion, and pancreatic HEI 56.7 6 11.3 48.4 6 7.9 0.002 b-cell function Data shown are unadjusted means 6 SD. P values indicate significance for comparisons between chronic Basal and postglucose load absolute in- cannabis smokers and age-, sex-, race-, and BMI-matched control subjects with Student t or Wilcoxon signed- sulin secretory rates were comparable rank test, as appropriate. between the groups. In order to assess care.diabetesjournals.org DIABETES CARE 3 Metabolic effects of cannabis

concentrations of ghrelin and active GLP-1 were similar between the groups (data not shown).

CONCLUSIONSdIn the current study, we demonstrated that chronic cannabis smokers had relative visceral adiposity and adipose tissue insulin re- sistance but not hepatic steatosis, glucose insulin insensitivity, impaired pancreatic b-cell function, glucose intolerance, or dyslipidemia compared with age-, sex-, ethnicity-, and BMI-matched control in- dividuals. Our study results suggest that chronic, daily cannabis use may have dif- ferential tissue-specific effects on insulin sensitivity, but these effects appear to have minimal impact on glucose or lipid metabolism.

Marijuana use and clinical characteristics of study participants Cannabis is the most frequently used illicit drug, especially in younger individuals, and accounts for the main reason for admission into treat- ments (12). Moreover, during the past few years there have been further increases in the prevalence of marijuana use in the U.S. (12). Consequently, understanding the chronic metabolic effects of marijuana smoking are important, a topic that has not been adequately studied. We designed this study to metabolically phenotype healthy chronic cannabis users. Marijuana users in the general population are more likely to be younger (,30 years old), be male, have an average age at first use of ;18 years, and to frequently use other illicit drugs (12). In this study, we ex- cluded individuals who currently abused other illicit drugs other than cannabis and Figure 1dTotal adiposity and regional fat distribution in chronic cannabis smokers and control with history of alcohol abuse or depen- subjects. A: Abdominal total fat, subcutaneous fat, and visceral fat content. B: TBF (regional dency. In 2010, an ;6.9 million current percent), visceral fat (percent of total abdominal fat), and hepatic fat content. Values shown are $ 6 , fi marijuana users used the drug on 20 means SE. *P 0.05, paired t test; ns, not signi cant. days in the past month (12). However, as noted in the Coronary Artery Risk Devel- insulin secretion in response to a dynamic Incretin secretion in chronic opment in Young Adults (CARDIA) Study change in plasma glucose levels, we mea- cannabis smokers (25) and Third National Health and Nu- sured rate sensitivity and IGI. IGI, the Incretins play an important role in regu- trition Examination Survey (NHANES III) ratio of the rise of the insulin and glucose lating appetite and pancreatic b-cell func- participants (26), only 1–3% of current concentrations over basal level at 30 min tion. To examine if chronic cannabis marijuana smokers are heavy users. In (DI/DG), and rate sensitivity were not dif- smoking affects incretin levels, we mea- our cohort of daily cannabis smokers, ferent between the groups (Table 3). In- sured fasting and post-OGTT incretin the dose and frequency of marijuana use sulin secretion and its relationship to concentrations. Fasting levels of acyl- (average ;10 joints/day) and chronicity plasma glucose concentrations were eval- ghrelin (10.6 6 2 vs. 13.2 6 8.2 pg/mL; (;12 years) was substantial. uated by model-derived b-cell glucose P = 0.29), desacyl ghrelin (282 6 152 vs. Marijuana use is associated with sensitivity parameter. Other measures of 279 6 200 pg/mL; P = 0.64), PYY (68.2 6 higher daily caloric intake. In the b-cell function, b-cell glucose sensitivity, 22.4 vs. 77.9 6 46.2 pg/mL; P = 0.44), NHANES III and CARDIA study, heavy and potentiation factor in chronic canna- and active GLP-1 (4.5 6 3.8 vs. 6.1 6 7.2 cannabis users had ;20% higher calorie bis smokers were also similar to control pmol/L; P = 0.52) were not different be- intake than nonusers (25,26). The in- subjects. tween the groups. Also, post-OGTT crease in calories was from higher intake

4 DIABETES CARE care.diabetesjournals.org Muniyappa and Associates

Table 2dMetabolic parameters in chronic cannabis users and control nonusers relatively rich in simple sugars was signif- icantly higher in marijuana smokers. fi Control subjects Cannabis smokers These ndings are consistent with human Variables (n =30) (n =30) P value and animal studies demonstrating that stimulate food intake, spe- Metabolic parameters cifically highly palatable sweet-tasting Fasting plasma glucose (mmol/L) 4.8 6 0.4 4.8 6 0.5 0.49 foods (28). Cannabis smokers in our study Fasting plasma insulin (pmol/L) 22.6 6 15.9 41.4 6 45.7 0.07 exhibited characteristics typically ob- 6 6 Hemoglobin A1c (A1C, %) 5.4 0.4 5.4 0.3 0.64 served in marijuana smokers in the general A1C (mmol/mol) 35.2 6 4.2 35.7 6 3.7 0.67 population. In particular, the degree and Lipids (mmol/L) duration of cannabis exposure in our Total cholesterol 4.2 6 0.7 3.9 6 1.0 0.17 study participants provided us the oppor- LDL cholesterol 2.4 6 0.7 2.2 6 0.8 0.26 tunity to clearly assess any significant met- HDL cholesterol 1.4 6 0.3 1.3 6 0.4 0.02 abolic effects of chronic and heavy Triglycerides 0.8 6 0.4 1.0 6 0.7 0.24 marijuana use in healthy individuals. FFA (mEq/L) 578 6 210 689 6 339 0.16 Hepatic enzymes (IU/L) Effects of marijuana on adiposity Alkaline phosphatase 59 6 16 69 6 12 0.004 and abdominal fat distribution ALT 28 6 11 26 6 16 0.47 D9-THC, the lipophilic active ingredient AST 17 6 7186 10 0.56 of cannabis, is known to accumulate in hsCRP (mg/L) 1.3 6 1.6 2.7 6 6.3 0.31 adipose tissue of heavy marijuana users Adiponectin (mg/mL) 5.3 6 4.0 3.8 6 4.5 0.10 (29). This may explain adipose tissue– Leptin (ng/mL) 14.7 6 16.9 15.3 6 18.2 0.78 specific effects of chronic marijuana use. D Data shown are unadjusted means 6 SD. P values indicate significance for comparisons between chronic In in vitro studies, 9-THC has been cannabis smokers and age-, sex-, race-, and BMI-matched control subjects with Student t or Wilcoxon signed- shown to increase adipocyte hypertrophy rank test, as appropriate. ALT, alanine aminotransferase; AST, aspartate aminotransferase; hsCRP, high- and lipogenesis (30). Based on extant lit- sensitivity C-reactive protein. erature, mainly from in vitro and animal of all macronutrients. Specifically, the fre- Consistent with other studies, the quality studies (2,3), we hypothesized that quency and amount of consumption of of diets consumed by cannabis users was chronic cannabis users may have higher soda, cheese, salty snacks, pork, and alco- poor (27). Furthermore, the percent of amounts of hepatic fat and abdominal vis- hol was higher in cannabis users. daily calories derived from carbohydrates ceral fat. Using MRS, we did not observe any significant differences in hepatic fat Table 3dIndices of insulin sensitivity/resistance, insulin secretion, and b-cell function in content between chronic cannabis users chronic cannabis users and control nonusers and age-, sex-, ethnicity-, and BMI- matched nonusers. In one recent epide- Control subjects Cannabis smokers miological study of several hundred Variables (n = 30) (n = 30) P value patients with hepatitis C infection, daily cannabis use was found to be a predictor Measures of insulin sensitivity/ of hepatic steatosis (by microscopy) (11). resistance However, there was a higher prevalence of Insulin sensitivity, skeletal hepatic steatosis in nonmarijuana users muscle/hepatic with HCV infection, suggesting that mar- QUICKI 0.415 6 0.039 0.400 6 0.061 0.23 ijuana use may exacerbate the pathologi- 2 2 OGIS (ml min 1 m 2) 444 6 67 459 6 75 0.48 cal effects of HCV infection in the liver. HIRI (Matsuda) 32.0 6 17.5 32.2 6 22.5 0.90 Alkaline phosphatase, but not hepatic Insulin sensitivity, adipocyte transaminase, levels were higher in can- AIRI 61.7 6 43.1 110 6 113 0.05 nabisusersinourstudy.Thisisconsistent FFA suppression (%) 91 6 5866 10 0.02 with a prior report demonstrating ele- FFA suppression index vated hepatic enzymes and hepatomegaly 6 6 (FFAAUC/insulinAUC) 153 92 289 295 0.01 in cannabis smokers (31). But for this sol- Measures of insulin secretion itary study, there are no other human IGI, OGTT 155 6 119 127 6 97 0.36 studies that help interpret or refute the Basal insulin secretion rate likelihood of a false-positive finding of 2 (pmol z min 1 z m2)696 27 79 6 46 0.29 higher levels of alkaline phosphatase. 2 Total insulin secretion (nmol z m 2)646 25 56 6 25 0.16 In animal studies, chronic CB R stim- 2 1 Glucose sensitivity(pmol z min 1 z m2 ulation by endocannabinoids favors adi- 2 z mmol 1) 185 6 117 148 6 81 0.12 posity independent of its effects on 2 Rate sensitivity (pmol z m2 z mmol 1)1,2546 817 1,418 6 981 0.43 appetite stimulation, as deduced from the 6 6 Potentiation (fold) 1.12 0.34 1.16 0.58 0.59 opposite effects of chronic CB1R blockade. Data shown are unadjusted means 6 SD. P values indicate significance for comparisons between chronic This contrasts with the effects of chronic cannabis smokers and age-, sex-, race-, and BMI-matched control subjects with Student t or Wilcoxon marijuana use as reported in NHANES III signed-rank test, as appropriate. (27), the National Epidemiologic Survey care.diabetesjournals.org DIABETES CARE 5 Metabolic effects of cannabis on Alcohol and Related Conditions, the D9-THC induces glucose intolerance in acutely increases plasma ghrelin and lep- National Comorbidity Survey–Replication men (8,9). However, the effects of chronic tin and decreases PYY levels in individuals (32), and the CARDIA (25) studies, which marijuana smoking on glucose disposal, with neuropathic pain (43). Although we show that despite an increase in caloric in- insulin action, and b-cell function have did not assess appetite ratings, fasting and take, prevalence of obesity is lower in can- not been previously investigated. In our post-OGTT levels of ghrelin, GLP-1, and nabis users as compared with nonusers. study, chronic cannabis impaired adipose PYY were unaffected by marijuana smok- Moreover, in NHANES III, higher fre- tissue insulin sensitivity but not insulin ing. It is probable that the duration of quency of marijuana use was significantly sensitivity parameters related to glucose cannabis exposure (acute vs. chronic) associated with lower BMI (27). This neg- metabolism. These tissue-specific effects may differentially affect levels of circulat- ative association of cannabis use and BMI of marijuana may be due to: 1) higher lev- ing appetite hormones. Further studies remained even after adjustment for poten- els of D9-THC and prolonged retention in are needed to confirm these findings. tial confounders such as age, sex, educa- adipose tissue (29); 2) higher sensitivity tion, caloric intake, , and to antilipolytic effects of insulin in adi- Metabolic effects of alcohol use. pose tissue compared with its action in phytocannabinoids versus In our study, nonmarijuana users had the liver (suppression of glucose produc- endocannabinoids comparable amounts of percent TBF, but tion) or skeletal muscle (stimulation of Phytocannabinoids are plant-derived nat- the total abdominal fat content (subcuta- glucose disposal); and 3) varying expres- ural products that are chemically similar neous and visceral combined) was signif- sion of CB1R in insulin-sensitive tissues to cannabinoids and interact with CB icantly lower in cannabis smokers. (e.g., expression of CB1Rs in liver is min- receptors. The two major cannabinoids However, when we examined the relative imal in healthy individuals) (42). from cannabis sativa are D9-THC and D amounts of abdominal fat distribution, Pancreatic CB1R activation leads to CBD. 9-THC is a CB1RandCB2R partial cannabis smokers had significantly higher b-cell death and impairs insulin secretion agonist, and CBD produces a broad range percentage of visceral fat. This finding is in mice (4). Whether chronic marijuana of effects not mediated by either CB1Ror consistent with the observation that mar- smoking affects insulin secretion in hu- CB2R, although in some systems it may ijuana use is associated with larger waist mans is not known. Using C-peptide de- behave as a CB2R inverse agonist (44). circumference (25). In obese individuals, convolution and OGTT modeling, total However, when CB receptor expression endocannabinoid levels are higher in ab- insulin secretion, b-cell glucose sensitiv- is low, D9-THC blocks CB receptor acti- dominal (33–36), but paradoxically lower ity, rate sensitivity, and potentiation of in- vation by other cannabinoids (44). in subcutaneous adipose tissue (37,38). sulin secretion were evaluated. Chronic Chronic cannabis use downregulates These findings have led to the suggestion cannabis smoking does not appear to im- CB1Rs (45) and may induce tolerance to that such depot-specific upregulation of pair b-cell glucose sensitivity, rate sensi- the effects of D9-THC. This is the most (ECS) may favor tivity, or insulin secretion. Thus, chronic likely explanation of the discrepant effects partition of lipids toward visceral and away marijuana smokers manifest normal glu- of cannabis on glucose metabolism in from subcutaneous fat depots during cose tolerance due to intact peripheral in- short-term studies and after chronic ex- states of caloric excess (39). Interestingly, sulin sensitivity and b-cell function. The posure. Although the possibility of CB1R D9-THC–induced decrease in AMP- lack of adverse effects of chronic cannabis desensitization exists, definitive conclu- activated protein kinase, an enzyme that use on glucose tolerance is consistent with sions cannot be made due to lack of activates fatty acid oxidation, is greater in the lower prevalence of type 2 diabetes in data regarding cannabis dose over time visceral adipocytes (40), suggesting a de- heavy marijuana smokers (26). Likewise, (stable vs. scaling up), especially at the pot-specific effect. Although not known, heavy marijuana use did not impair fasting time of metabolic assessment. The abil- such depot-specific actions of phytocanna- glucose levels in the CARDIA study (25). ity of D9-THC to block or activate CB binoids may explain the effects of chronic receptors, differential downregulation cannabis smoking on regional fat distribu- Marijuana use, blood pressure, and of CB1Rs, role of non- CB1Ractivation tion. In addition, non–D9-THC phytocan- dyslipidemia by phytocannabinoids, varying tissue- nabinoids (e.g., [CBD]) may Chronic marijuana users had higher specific expression of CB receptors, and also play a role in adipogenesis (41). How- blood pressure but near-normal lipid the development of tolerance to some ever, the ratio of D9-THC versus CBD ex- parameters compared with nonusers. effects of D9-THC may explain the differ- ceeded 10:1 in the vast majority of Similar findings have also been reported ential effects of ECS upregulation in ro- marijuana samples seized in Maryland from the NHANES III and CARDIA study dents and chronic cannabis use in (M. A. ElSohly, personal communication). (25,26). One potential confounder in in- humans. This makes it unlikely that non–D9-THC terpreting these findings is the frequent cannabinoids such as CBD contributed to association of tobacco smoking in chronic Limitations the relative visceral adiposity observed in cannabis smokers in our and other’sstud- Our study has several limitations aside our study. These novel and intriguing find- ies (12,25). from the small cohort size and lack of ings from our study should be considered multiplicity corrections for exploratory hypothesis-generating that need further Marijuana use and appetite- outcomes. First, the accuracy of self- confirmation in larger studies. modifying incretins reported substance use is frequently ques- Appetite stimulating actions of cannabis tioned due to substantial underreporting. Effects of marijuana use on insulin are well established, but the effects of However, in our study, cannabis use was action and secretion chronic cannabis use on appetite hor- elicited by experienced investigators and Daily smoking of marijuana for 7 days or mones are unknown. In a recent study, underreporting minimized through assur- acute intravenous administration of administration of medicinal cannabis ances of confidentiality, lack of adverse

6 DIABETES CARE care.diabetesjournals.org Muniyappa and Associates consequences, appropriate recall cues, manuscript. A.M. participated in the data intake and body weight of humans living and other confirmatory biological markers analysis and approved the final manuscript. in a residential laboratory. Appetite 1988; for confirmation (46). Even if underrepor- K.Y.C. participated in the data analysis and 11:1–14 fi writing of the manuscript and approved the 11. Hézode C, Zafrani ES, Roudot-Thoraval F, ted, cannabis use was signi cant in our fi study. Second, although the effect of to- nal manuscript. N.D.V. and G.K. participated et al. Daily cannabis use: a novel risk in the design of the study and writing of the factor of steatosis severity in patients with bacco smoking on insulin sensitivity is manuscript and approved the final manu- chronic hepatitis C. Gastroenterology equivocal (47,48), the high prevalence of script. M.A.H. participated in the design and 2008;134:432–439 concurrent tobacco use in cannabis users conduct of the study, collection of data, and 12. Results from the 2010 National Survey on is inevitably a major confounder. None- writing of the manuscript and approved the Drug Use and Health: Summary of Na- theless, the metabolic parameters were final manuscript. M.C.S. participated in the tional Findings [Internet]. Rockville, MD, not different between tobacco users and design and conduct of the study, collection of Substance Abuse and Mental Health Serv- nonusers in cannabis users. Third, con- data, and writing of the manuscript and ap- ices Administration. Available from http:// sumption of a diet rich in carbohydrates proved the final manuscript. M.C.S. is the www.samhsa.gov/data/nsduh/2k10nsduh/ with high glycemic index may have had guarantor of this work and, as such, had full 2k10results.htm. Accessed 12 February independent effects on adipose tissue in- access to all the data in the study and takes 2013 responsibility for the integrity of the data and sulin sensitivity and visceral adiposity. 13. Subar AF, Thompson FE, Kipnis V, et al. the accuracy of the data analysis. Comparative validation of the Block, However, carbohydrate consumption Parts of this study were presented in abstract Willett, and National Cancer Institute was assessed by dietary history, which is form at The Endocrine Society’s 94th Annual food frequency questionnaires : the Eating highly variable, especially in our cannabis Meeting & Expo, Houston, Texas, 23–26 June at America’s Table Study. Am J Epidemiol cohort. Fourth, we used surrogate mea- 2012. 2001;154:1089–1099 sures of insulin sensitivity and secretion. 14. Guenther PM, Reedy J, Krebs-Smith SM. Although these indices and b-cell function Development of the Healthy Eating Index- modeling have been extensively validated References 2005. J Am Diet Assoc 2008;108:1896– and used, the findings from our explor- 1. Pacher P, Batkai S, Kunos G. The endo- 1901 atory study need to be confirmed using system as an emerging - 15. van der Meer RW, Lamb HJ, Smit JW, de gold-standard techniques such as euglyce- get of pharmacotherapy. Pharmacol Rev Roos A. MR imaging evaluation of car- 2006;58:389–462 diovascular risk in metabolic syndrome. mic hyperinsulinemic and hyperglycemic 2. Pagotto U, Marsicano G, Cota D, Lutz B, Radiology 2012;264:21–37 clamp techniques. Fifth, given the high Pasquali R. The emerging role of the en- 16. Ouwerkerk R, Pettigrew RI, Gharib AM. proportion of African Americans in our docannabinoid system in endocrine reg- Liver metabolite concentrations measured cohort, our results may have also been af- ulation and energy balance. Endocr Rev with 1H MR spectroscopy. Radiology fected by the relative resistance of African 2006;27:73–100 2012;265:565–575 Americans to hepatic steatosis (49). Fi- 3. Liu J, Zhou L, Xiong K, et al. Hepatic 17. Muniyappa R, Lee S, Chen H, Quon MJ. nally, confounding effects of varying com- -1 mediates diet- Current approaches for assessing insulin position of cannabis preparations and the induced insulin resistance via inhibition sensitivity and resistance in vivo: advan- of insulin signaling and clearance in mice. tages, limitations, and appropriate usage. relative amounts and activities of non- – THC cannabinoids (e.g., CBD, cannabi- Gastroenterology 2012;142:1218 1228. Am J Physiol Endocrinol Metab 2008; e1211 294:E15–E26 gerol, and ) and 4. Kim W, Lao Q, Shin YK, et al. Cannabi- 18. Mari A, Pacini G, Murphy E, Ludvik B, even noncannabinoids such as terpenoids noids induce pancreatic b-cell death by Nolan JJ. A model-based method for as- on glucose metabolism is an unavoidable directly inhibiting insulin receptor acti- sessing insulin sensitivity from the oral limitation. vation. Sci Signal 2012;5:ra23 glucose tolerance test. Diabetes Care In healthy young individuals, chronic 5. Osei-Hyiaman D, DePetrillo M, Pacher P, 2001;24:539–548 cannabis smoking was associated with et al. Endocannabinoid activation at he- 19. Abdul-Ghani MA, Matsuda M, Balas B, visceral adiposity and adipose tissue in- patic CB1 receptors stimulates fatty acid DeFronzo RA. Muscle and liver insulin sulin resistance but not with hepatic synthesis and contributes to diet-induced resistance indexes derived from the oral obesity. J Clin Invest 2005;115:1298– glucose tolerance test. Diabetes Care steatosis, insulin insensitivity, impaired – pancreatic b-cell function, or glucose in- 1305 2007;30:89 94 20. Katz A, Nambi SS, Mather K, et al. Quan- tolerance. 6. Després JP, Golay A, Sjöström L; Rimo- nabant in Obesity-Lipids Study Group. titative insulin sensitivity check index: Effects of rimonabant on metabolic risk a simple, accurate method for assessing AcknowledgmentsdThis work was sup- factors in overweight patients with dysli- insulin sensitivity in humans. J Clin En- ported by the Intramural Research Programs of pidemia. N Engl J Med 2005;353:2121– docrinol Metab 2000;85:2402–2410 National Institute of Diabetes and Digestive 2134 21. Abdul-Ghani MA, Molina-Carrion M, Jani and Kidney Diseases, National Institute on 7. Triay J, Mundi M, Klein S, et al. Does ri- R, Jenkinson C, Defronzo RA. Adipocytes Drug Abuse, and National Institute on Alcohol monabant independently affect free fatty in subjects with impaired fasting glucose Abuse and Alcoholism, National Institutes of acid and glucose metabolism? J Clin En- and impaired glucose tolerance are re- Health. docrinol Metab 2012;97:819–827 sistant to the anti-lipolytic effect of in- No potential conflicts of interest relevant to 8. Hollister LE, Reaven GM. Delta-9- sulin. Acta Diabetol 2008;45:147–150 this article were reported. tetrahydrocannabinol and glucose toler- 22. Tschritter O, Fritsche A, Thamer C, et al. R.M. participated in the conduct of the ance. Clin Pharmacol Ther 1974;16:297–302 Plasma adiponectin concentrations pre- study, data analysis, and writing of the man- 9. Podolsky S, Pattavina G, Amaral M. Effect dict insulin sensitivity of both glucose and uscript and approved the final manuscript. of marijuana on the glucose-tolerance test. lipid metabolism. Diabetes 2003;52:239– S.S., R.O., A.M.G., M.W., A.C., and G.H. AnnNYAcadSci1971;191:54–60 243 participated in the conduct of the study and 10. Foltin RW, Fischman MW, Byrne MF. 23. Mari A, Schmitz O, Gastaldelli A, collection of data and approved the final Effects of smoked marijuana on food Oestergaard T, Nyholm B, Ferrannini E. care.diabetesjournals.org DIABETES CARE 7 Metabolic effects of cannabis

Meal and oral glucose tests for assess- 32. Le Strat Y, Le Foll B. Obesity and cannabis 41. De Petrocellis L, Ligresti A, Moriello AS, ment of beta -cell function: modeling use: results from 2 representative national et al. Effects of cannabinoids and analysis in normal subjects. Am J Physiol surveys. Am J Epidemiol 2011;174:929– cannabinoid-enriched Cannabis extracts Endocrinol Metab 2002;283:E1159– 933 on TRP channels and endocannabinoid E1166 33. Matias I, Gonthier MP, Orlando P, et al. metabolic enzymes. Br J Pharmacol 2011; 24. Seltzer HS, Allen EW, Herron AL Jr, Regulation, function, and dysregulation 163:1479–1494 Brennan MT. Insulin secretion in re- of endocannabinoids in models of adipose 42. Teixeira-Clerc F, Julien B, Grenard P, et al. sponse to glycemic stimulus: relation of and beta-pancreatic cells and in obesity CB1 cannabinoid receptor antagonism: delayed initial release to carbohydrate in- and hyperglycemia. J Clin Endocrinol a new strategy for the treatment of liver tolerance in mild diabetes mellitus. J Clin Metab 2006;91:3171–3180 fibrosis. Nat Med 2006;12:671–676 Invest 1967;46:323–335 34. Bluher€ M, Engeli S, Klöting N, et al. Dys- 43. Riggs PK, Vaida F, Rossi SS, et al. A pilot 25. Rodondi N, Pletcher MJ, Liu K, Hulley SB, regulation of the peripheral and adipose study of the effects of cannabis on appetite Sidney S; Coronary Artery Risk De- tissue endocannabinoid system in human hormones in HIV-infected adult men. velopment in Young Adults (CARDIA) abdominal obesity. Diabetes 2006;55: Brain Res 2012;1431:46–52 Study. Marijuana use, diet, body mass 3053–3060 44. Pertwee RG. The diverse CB1 and index, and cardiovascular risk factors 35. Côté M, Matias I, Lemieux I, et al. Circu- CB2 receptor pharmacology of three (from the CARDIA study). Am J Cardiol – lating endocannabinoid levels, abdominal plant cannabinoids: delta9-tetrahydro- 2006;98:478 484 adiposity and related cardiometabolic risk , cannabidiol and delta9- 26. Rajavashisth TB, Shaheen M, Norris KC, factors in obese men. Int J Obes (Lond) tetrahydrocannabivarin. Br J Pharmacol et al. Decreased prevalence of diabetes in 2007;31:692–699 2008;153:199–215 marijuana users: cross-sectional data from 36. Di Marzo V, Côté M, Matias I, et al. 45. Hirvonen J, Goodwin RS, Li CT, et al. the National Health and Nutrition Ex- Changes in plasma endocannabinoid lev- Reversible and regionally selective down- amination Survey (NHANES) III. BMJ els in viscerally obese men following a 1 regulation of brain cannabinoid CB1 Open 2012;2:e000494 year lifestyle modification programme receptors in chronic daily cannabis 27. Smit E, Crespo CJ. Dietary intake and and waist circumference reduction: asso- smokers. Mol Psychiatry 2012;17:642– nutritional status of US adult marijuana ciations with changes in metabolic risk 649 users: results from the Third National – Health and Nutrition Examination Sur- factors. Diabetologia 2009;52:213 217 46. Donovan DM, Bigelow GE, Brigham GS, vey. Public Health Nutr 2001;4:781–786 37. Annuzzi G, Piscitelli F, Di Marino L, et al. et al. Primary outcome indices in illicit 28. Abel EL. Effects of marihuana on the so- Differential alterations of the concen- drug dependence treatment research: sys- lution of anagrams, memory and appetite. trations of endocannabinoids and related tematic approach to selection and mea- Nature 1971;231:260–261 lipids in the subcutaneous adipose tissue surement of drug use end-points in clinical – 29. Johansson E, Norén K, Sjövall J, Halldin of obese diabetic patients. Lipids Health trials. 2012;107:694 708 MM. Determination of delta 1-tetrahy- Dis 2010;9:43 47. Henkin L, Zaccaro D, Haffner S, et al. drocannabinol in human fat biopsies from 38. Bennetzen MF, Wellner N, Ahmed SS, et al. smoking, environmental tobacco marihuana users by gas chromatography- Investigations of the human endocanna- smoke exposure and insulin sensitivity: mass spectrometry. Biomed Chromatogr binoid system in two subcutaneous adi- the Insulin Resistance Atherosclero- 1989;3:35–38 pose tissue depots in subjects and in sis Study. Ann Epidemiol 1999;9:290– 30. Wong A, Gunasekaran N, Hancock DP, obese subjects before and after weight loss. 296 et al. The major plant-derived cannabi- Int J Obes (Lond) 2011;35:1377–1384 48. Attvall S, Fowelin J, Lager I, Von Schenck noid D(9)-tetrahydrocannabinol promotes 39. Di Marzo V. The endocannabinoid system H, Smith U. Smoking induces insulin re- hypertrophy and macrophage infiltration in obesity and type 2 diabetes. Dia- sistanceda potential link with the insulin in adipose tissue. Horm Metab Res 2012; betologia 2008;51:1356–1367 resistance syndrome. J Intern Med 1993; 44:105–113 40. Kola B, Hubina E, Tucci SA, et al. Can- 233:327–332 31. Borini P, Guimaraes RC, Borini SB. Pos- nabinoids and ghrelin have both central 49. Guerrero R, Vega GL, Grundy SM, sible hepatotoxicity of chronic marijuana and peripheral metabolic and cardiac ef- Browning JD. Ethnic differences in he- usage. Sao Paulo Med J 2004;122:110– fects via AMP-activated protein kinase. patic steatosis: an insulin resistance para- 116 J Biol Chem 2005;280:25196–25201 dox? Hepatology 2009;49:791–801

8 DIABETES CARE care.diabetesjournals.org