Alcohol use disorder (AUD) impacts a significant number of Americans and has an impact on both the economy and healthcare. Evidence indicates that glutamate neurotransmission plays a critical role in alcohol and other substance addiction. It is believed that the glutamate system is disturbed by acute and chronic alcohol exposure and as a result leads to cue induced and alcohol induced reinstatement of drinking during withdrawal. The majority of the research investigating the relationship between brain glutamate and addiction has been conducted with animal models. Although few studies reported higher glutamate levels in alcohol-dependent patients compared with normal controls¹, how the glutamate system interacts with craving effects over the course of alcohol dependence remains unknown in humans. The aim of this study is to explore dynamic changes of glutamate associated with the craving effect for early developed alcohol use disorder. 1H-MR spectroscopy (MRS) was employed to quantify glutamate level changes in response to alcohol cues in individuals with AUD and social drinkers. 13 individuals with AUD (4 females, 23.1±2.5y) and 15 controls (social drinkers; 6 females, 22.5±3.0y) participated in the study. AUD subjects met criteria for moderate or severe AUD based on DSM-V criteria using the Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA)²; control subjects had no history of any AUD. Lifetime alcohol and marijuana problem counts were calculated as the sum of positive responses pertaining to alcohol or marijuana use, respectively, as reported on the SSAGA. All subjects were scanned on a 3 Tesla Siemens TIM Trio scanner. All subjects were instructed not to drink alcohol 12 hours prior to the scan which was ensured by a breath alcohol test. A T1-weighted high-resolution anatomical image was acquired with the MP-RAGE sequence, followed by two single-voxel MRS scans 15 minutes apart. The PRESS sequence was used for MRS on a voxel (15 mm x 30 mm x 20 mm) in the anterior cingulate cortex (ACC): TR/TE = 2000/30 ms, bandwidth = 2000 Hz, 2048 data points, number of measurements = 120, scan time = 4 min, followed by a water reference scan (8 averages). During the first MRS scan, a total of 60 furniture pictures (Fig. 1A) were shown to the subjects in a random order These pictures were changed to pictures of alcoholic beverages from University of Arkansas³ (Fig. 1B) six minutes before the second MRS scan until the scan ended. After each MRS scan session, the subjects were asked to rate their craving on a scale of 1 to 5 with 1 meaning little to no craving and 5 meaning extreme craving. The spectra data were processed with LCModel to quantify glutamate concentrations along with other metabolites including total NAA (tNAA), Choline (Cho), and total creatine (tCr). The baseline metabolite levels were compared between AUD and controls, and between furniture cues and alcohol cues, using two-sample t-tests. We also examined the correlation between baseline glutamate level and total alcohol problem counts after excluding AUD subjects with other drug problems (9 AUD subjects left).AUD subjects reported more craving scores than controls (p = 0.021), indicating the alcoholic beverage pictures did induce a craving effect. Both the absolute value of glutamate concentration and its ratio to total creatine (Glu/tCr) measured decreased significantly for AUD subjects during the second MRS scan, i.e. while they viewed the pictures of alcoholic beverages. This was not the case for controls. No significant change of other metabolites was found, as shown in Fig. 2. Fig. 3 displays the glutamate level as well as Glu/tCr for different cohorts and conditions. There is no significant difference of baseline glutamate level between AUD and controls. However, a high correlation (r = -0.90) between baseline glutamate level and alcohol problem counts was observed for AUD subjects, as shown in Fig. 4. This is the first in vivo study seeing glutamate decrease in alcohol depedent individuals induced by alcohol cues. The subjects typically developed AUD within 2-3 years before the scan. This observation suggests that the homeostasis of the glutamatergic system can be affected by the craving effect in early alcohol dependence in humans, which is consistent with the negative correlation between baseline glutamate level and alcohol problem counts. To our knowledge, the cue induced decrease of glutamate is not a direct translation from animal studies. In contrast, animal studies have shown elevated glutamate level in ACC and nucleus accumbens associated with craving⁴. The discrepancy could be due to the difference in the stage of alcohol dependence and the underlying neurobiological cause of craving.