Craving dynamics and related cerebral substrates predict timing of use in alcohol, tobacco, and cannabis use disorders

Abstract

Background: Patients treated for Substance Use Disorders exhibit highly fluctuating patterns of craving that could reveal novel prognostic markers of use. Accordingly, we 1) measured fluctuations within intensively repeated measures of craving and 2) linked fluctuations of craving to connectivity indices within resting-state (rs) brain regions to assess their relation to use among patients undergoing treatment for Alcohol, Tobacco and Cannabis Use Disorders.

Method: Participants -64 individuals with SUD for tobacco, alcohol, or cannabis and 35 healthy controls-completed a week of Ecological Momentary Assessment (EMA) during which they reported craving intensity and substance use five times daily. Before EMA, a subsample of 50 patients, and 34 healthy controls also completed resting-state (rs)-MRI acquisitions. Craving temporal dynamics within each day were characterized using Standard Deviation (SD), Auto-Correlation Factor (ACF), and Mean Successive Square Difference (MSSD). Absolute Difference (AD) in craving between assessments was a prospective prediction measure.

Results: Within-day, higher MSSD predicted greater substance use while controlling for mean craving. Prospectively higher AD predicted later increased substance use independently of previous use or craving level. Moreover, MSSD was linked to strength in five functional neural connections, most involving frontotemporal systems. Cerebello-thalamic and thalamo-frontal connectivity were also linked to substance use and distinguished the SUD from the controls.

Conclusion: To the best of our knowledge, this is the first study to indicate that instability in craving may be a trigger for use in several SUD types, beyond the known effect of craving intensity.

Keywords: Craving; Dynamic; EMA; Markers; Resting state; Substance use.

Fig. 1.. Overall methodology for the EMA sample and MRI subsample.

Legend: All participants underwent an interview to assess inclusion criteria, including the Mini International Neuropsychiatric Interview and the Addiction Severity Index for the patients. All participants underwent a week of EMA assessments. Usual “wake” and “sleep” times were selected by each participant to ensure that they would be able to answer each test signal. Typical day of EMA is represented above and included signals occurring randomly within 5-time intervals periods. At each signal, craving was assessed on a scale from 1 to 7, and Substance use was coded in a dichotomous manner (“Yes” or “No”) for the substance they are treated for and for any substance. These within day assessments were then extracted to summarize craving variability (Standard Deviation), inertia (Auto Correlation Factor), instability (Mean Successive Square Difference) and number of Substance Use occasions for each day. A subsample of these participants also underwent MRI acquisition, including a resting state session, prior to the EMA phase.

Fig. 2.. Resting state connections correlating negatively (in blue) and positively (in red) with craving instability

L: Left; R: Right; Oper: Opercular; Inf: Inferior; Orb: Orbital; ACC: Anterior Cingulate Cortex, Re: Reuniens, PuL: Pulvinar Lateral, Rs: resting state Legend: Regression analyses were conducted within the CONN toolbox between the individual connectomes and mean MSSD craving from each patient with SUD in the MRI subsample correcting for false discovery rate (FDR) with alpha level of 0.05. Resulting connections are highlighted in the above figure and colored depending on their association with MSSD craving, negatively (highlighted in blue) or positively (highlighted in red). The strength of connectivity within these connections was then extracted for both patients with SUD and controls to be compared across both groups and conduct correlations analysis with the number of substance use occasions during the week.