Importance of construct validity in functional neuroimaging of craving.

Abstract

Sayette & Wilson 1 raise a number of important considerations for functional neuroimaging of drug craving, particularly whether existing studies accurately capture construct validity of craving as a strong desire or urge to use. Intensity of craving may be a useful marker of severity of substance-related problems. An urge so intense that the individual has difficulty thinking about anything else could approximate loss of control over substance use 2. Working within cue–reactivity studies among cigarette smokers, the authors conducted a meta-analysis exploring differences in reported craving and brain activation among smokers who had been deprived before scanning and those who had smoked a cigarette just before participation. They observed significantly lower urge ratings among non-deprived participants, and found a region of the anterior cingulate cortex (ACC) extending into the ventromedial prefrontal cortex (PFC) that was more likely to show increased activation during cigarette cues in deprived than in non-deprived smokers. The authors use this finding to highlight the importance of carefully considering craving intensity when interpreting cue–reactivity studies. The ACC has been implicated as a node between regions related to reward (i.e. subcortical regions project to the ACC) and cognitive control (i.e. the ACC projects to the PFC). The likely mechanism is that reward response activates the ACC which, in turn, activates PFC to initiate planning and appetitive behavior 3. Overall, the increased likelihood of greater activation in ACC/PFC observed by the authors is consistent with increased craving among deprived smokers. However, a major potential confound identified by Sayette & Wilson 1 should be considered in more detail. In the included studies, deprived smokers had abstained for 2–16 hours, and non-deprived smokers had been instructed to smoke as they desired. The effects of nicotine may last for hours after the last smoked cigarette, and nicotine is notably associated with vasoconstriction 4. It is therefore still difficult to interpret blood oxygen level-dependent signals in deprived smokers, given the variance in time since the last cigarette was smoked. Time since last cigarette and/or a measure of urgency of craving should, arguably, be included as a covariate in functional analyses. Studies among non-deprived smokers may still be useful in understanding relapse prevention. Decreased ACC activation during a modest desire to smoke, or when a craving has presumably been satisfied, provides additional support for reduced activation of the ACC as a biomarker for decreased craving and a possible predictor of treatment efficacy. Further treatment development could approach smoking cessation by reducing reward or by increasing cognitive control. Given that the authors excluded any studies where participants were instructed to resist craving or to attempt to reduce urges to use, additional studies could compare activation in those attempting to control their craving, those who recently satisfied their craving and those who were experiencing intense craving. Such a study could inform clinical conceptualization of craving (e.g. linear or nonlinear intensity with varying cognitive processes). It could also elucidate whether intense craving is an entirely distinct construct involving recruitment of additional functional regions, or any regions of decreased activation (e.g. in additional cognitive control circuitry), compared to more modest desire or satisfied urge. Despite potential limitations, imaging measures provide an important link between behavioral approaches to treatment and changes in the brain that may underlie treatment gains. Future studies will probably incorporate multi-modal imaging or multiple analytical techniques. Further consideration of construct validity in functional neuroimaging studies will be important in increasing translational impact in improving treatment efficacy, as well as increasing the ability of researchers to compare results meaningfully across studies. None.