Potent Dopamine D2 Antagonists Block the Reward-Enhancing Effects of Nicotine in Smokers With Schizophrenia

Alexis E. Whitton; Alan I. Green; Diego A. Pizzagalli; Robert M. Roth; Jill M. Williams; Mary F. Brunette


Schizophr Bull. 2019;45(6):1300-1308. 

In This Article



Sample characteristics are shown in Table 1. Half of the sample were African American (n = 93, 50.5%), most were non-Hispanic, (n = 158, 85.9%) and unemployed (n = 166, 90.2%). The mean (±SEM) age was 45.73 ± 11.35 and range = 20–70, and the sample was two-thirds male (n = 122, 66.3%). Symptoms were moderate (BPRS score = 41.32 ± 11.45, mean lifetime hospitalizations = 11.65 ± 15.0). They reported smoking 15.57 ± 13.34 cigarettes/day in the past 3 months.

Of those who completed the PRT (n = 174), 114 (65.5%) had valid data both prior to and post-smoking. Those who were excluded (81.7% male; mean age = 46.7, SD = 12.8) did not differ from those who were retained in terms of symptom severity (BPRS) or nicotine dependence and craving (FTND and QSU; all Ps > .20).

Smoking and Clinical Characteristics of Antipsychotic Medication Groups

Of those with valid PRT data, 98 could be classified as taking or not taking potent DA D2 receptor antagonists (taking potent D2 antagonists [D2 antagonism+], n = 71; not taking potent D2 antagonists [D2 antagonism–], n = 27). Table 2 shows how the medication profiles were grouped. Independent samples t tests revealed no group differences in the average number of cigarettes smoked daily (P = .85), pre-assessment expired CO (P = .71), post-smoking expired CO (P = .82), self-reported nicotine dependence on the FTND (P = .67), craving on the QSU (P = .65), or symptom severity on the BPRS (total score and subscales, all Ps > .15). However, the groups did differ with respect to gender and ethnicity, where the D2 antagonism + group contained a higher proportion of males, χ 2 = 0.43, P = .04, and Hispanic individuals, χ 2 = 6.74, P = .01.

Interactive Effects of Antipsychotic Type and Smoking on Reward Processing

A significant 3-way interaction emerged from the medication (D2 antagonism+, D2 antagonism–) × smoking × block ANOVA, F(1,96) = 8.17, P = .005 ηp 2 = .08 (figure 1). This interaction was followed up by considering all 2-way interactions, and Bonferroni-corrected post hoc tests of simple effects were conducted in cases where the 2-way interaction was significant. For brevity, only main effects and interactions involving smoking or medication are reported.

Figure 1.

Response bias in block 1 (B1) and block 2 (B2), prior to smoking and post-smoking, for the D2 antagonism+ and D2 antagonism– groups. Bars show mean (±SEM). *P < .05; ***P < .001.

When examining the smoking × block interaction separately within each medication group, we observed a significant smoking × block interaction for the D2 antagonism– group, F(1,26) = 8.56, P = .007, ηp 2 = 0.25, but not for the D2 antagonism+ group (P = .64). In the D2 antagonism– group, an increase in response bias from block 1 to block 2 of the task was observed following smoking (P < .001) but not prior to smoking (P = .72), indicating that smoking enhanced reward learning in this group.

When examining the medication × block interaction separately for each smoking condition, at pre-smoking, the medication × block interaction was not significant (P = .28) nor was there a main effect of medication (P = .14). For the post-smoking condition, however, a significant medication × block interaction emerged, F(1,96) = 9.07, P = .003, ηp 2 = 0.09. Specifically, response bias increased from block 1 to block 2 in both the D2 antagonism– (block 1 = 0.05, block 2 = 0.20, P < .001) and the D2 antagonism+ group (block 1 = 0.11, block 2 = 0.15, P = 0.03); however, this increase was greater in the D2 antagonism– group.

Figure 2 summarizes the interactive effects of medication and smoking on reward learning, conceptualized as the increase in response bias from block 1 to block 2. The increase in response bias was greater in the D2 antagonism– group, indicating greater reward learning in these patients.

Figure 2.

Interactive effects of potent D2 antagonist and smoking on reward learning (defined as response bias in block 2 – response bias in block 1). Bars show mean (±SEM). **P < .01.

Independent samples t tests confirmed that the D2 antagonism+ and the D2 antagonism– groups did not differ significantly in terms of the ratio of rich to lean rewards that they received (summed across blocks) at pre- or post-smoking (both Ps > 0.10), indicating that effects were not due to differential rates of exposure to the asymmetrical reinforcement schedule.

Because the 2 medication groups differed in terms of gender and ethnic makeup, further analyses were conducted to examine these variables as potential confounds. To maximize statistical power, we performed 2 medication × smoking ANCOVAs containing reward learning as the dependent variable and gender/ethnicity as the covariate. For the model containing gender as a covariate, a medication × smoking interaction emerged, F(1,95) = 7.23, P = .008, ηp 2 = 0.07. Bonferroni-corrected post hoc tests showed that after controlling for gender, reward learning increased from pre- to post-smoking in the D2 antagonism– group (P = .004), but not in the D2 antagonism+ group (P = .69). Furthermore, in this model, the main effect of gender was not significant, F(1,95) = 0.06, P = .80, ηp 2 = 0.001, nor was the gender × smoking interaction, F(1,95) = 0.01, P = .64, ηp 2 = 0.002. For the model containing ethnicity as a covariate, the medication × smoking interaction emerged, F(1,95) = 8.29, P = .005, ηp 2 = 0.08. Post hoc tests showed that after controlling for ethnicity, reward learning increased from pre- to post-smoking in the D2 antagonism– group (P = .006), but not in the D2 antagonism+ group (P = .19). Similarly, the main effect of ethnicity was not significant in this model, F(1,95) = 1.11, P = .30, ηp 2 = 0.01, and ethnicity did not interact with smoking, F(1,95) = 0.25, P = .62, ηp 2 = 0.003. These results suggest that differences in the gender and ethnic makeup of the medication groups did not drive our primary findings.

Effects of Potent D2 Antagonists and Smoking on Discriminability

Mean discriminability at block 1 and block 2, pre- and post-smoking, across the 2 medication groups is shown in supplementary figure S1. No main effects or interactions involving medication or smoking emerged from the medication × smoking × block ANOVA on discriminability scores (all Ps > .10). This indicates that our primary findings were specific to reward processing, and not due to the effect of D2 antagonists or smoking on more general task performance.

Figure S1.

Discriminability in block 1 ("B1") and block 2 ("B2"), prior to smoking and post-smoking, for the D2 antagonism + and D2 antagonism – groups. Bars show mean (±SEM).

Associations Between Smoking-related Increases in Reward Processing and Smoking Severity

Reward learning was not correlated with scores on the FTND or QSU (both Ps > .05).