Differential post-treatment outcomes of methylphenidate for smoking cessation for individuals with ADHD

Abstract

Background and Objectives:

In a multisite, randomized study (CTN-0029), a 3-month course of osmotic-release oral system methylphenidate (OROS-MPH) improved smoking cessation in a group of patients with higher baseline severity in Attention-Deficit/Hyperactivity Disorder (ADHD). This treatment, however, worsened smoking cessation outcome in the group with lower baseline ADHD severity. We want to examine whether this differential treatment effect persisted after OROS-MPH was discontinued.

Method:

We conducted a secondary analysis of the 1-month follow-up data from CTN-0029 after the discontinuation of OROS-MPH (N=134). Nicotine patch was tapered during this month. We tested whether OROS-MPH had an effect on self-reported 7-day abstinence by week, as well as possible treatment by baseline ADHD severity interactions.

Results:

Abstinence diminished overall in time after the end of the treatment. In the high baseline severity group, patients who received OROS-MPH had an advantage in 7-day abstinence at week 15 (40% for OROS-MPH vs. 20% for placebo, OR = 2.63, p = 0.028). In the lower severity group (n=121), no difference was detected (29% for OROS-MPH vs. 32% for placebo, p = 1.00) between the two treatment groups. There was also a significant treatment by baseline ADHD severity interaction (p=0.03).

Conclusions and Scientific Significance:

OROS-MPH promotes abstinence beyond the course of treatment for patients with more severe ADHD, while the paradoxical effects in the lower baseline severity group is not persistent after medication discontinuation. Targeting ADHD in smoking cessation as a co-morbidity therefore can have broader impact with more precise patient selection.

Introduction

Cigarette smoking is commonly associated with Attention-Deficit/Hyperactivity Disorder (ADHD) and both nicotine and dopamine are involved in executive functioning and reward processing¹. ADHD symptom severity is associated with cigarette smoking^2,3. Long-acting psychostimulants, such as Osmotic Release Oral System Methylphenidate (OROS-MPH), are the mainstays of treatment for ADHD^4,5. The effect of psychostimulants on cigarette smoking, however, is complex: acute administration of psychostimulants may increase cigarette smoking^6–8, craving for cigarettes, and smoking impulses⁹. There are also concerns that psychostimulants may potentiate initiation of cigarette smoking and other drug use, though case control studies have not shown this^10,11. In non-ADHD adults, methylphenidate might have little impact on smoking cessation outcome¹².

A multi-center, randomized controlled trial was conducted in the National Drug Abuse Treatment Clinical Trials Network (CTN-0029) to test whether OROS-MPH is useful as part of a smoking cessation treatment program in adults with ADHD¹³. There was no significant difference in prolonged abstinence between placebo and OROS-MPH groups. However, treatment response differed with ethnicity and ADHD subtype^14,15 and with baseline ADHD severity¹⁶: patients with higher baseline ADHD severity achieved prolonged abstinence more frequently with OROS-MPH than with placebo. In contrast, patients with lower baseline ADHD severity achieved prolonged abstinence less frequently with OROS-MPH. Stratifying treatment based on baseline ADHD severity could optimize smoking cessation outcome¹⁷. This divergent effect has been shown to be not mediated by intermediate outcomes such as withdrawal symptoms and craving, as they decreased in both the higher and the lower severity subgroups¹⁸. New approaches that tested multiple moderators simultaneously such as Best Approximating Modeling¹⁹ suggested more complex subgroups of smokers might benefit OROS-MPH.

One question that has not been examined is whether this counterproductive effect of OROS-MPH on smoking cessation persists after medication discontinuation, and if it does, for how long. Some evidence suggests existing smoking-cessation treatments, such as varenicline and nicotine replacements would be more efficacious as a maintenance treatment beyond the standard 3 months^20,21. Medications might have a residual beneficial effect, though such effects were typically dwarfed by the amount of relapse overall: in large Phase III studies, both varenicline and bupropion treatment for only 3 months were still advantageous over placebo at 1 year, but the majority relapsed at that point^22,23. In the CTN-0029 trial, after the end of the trial and OROS-MPH discontinuation, patients were followed for an additional 4 weeks while their nicotine patches were tapered. We wish to examine whether treating smokers with OROS-MPH would also have a residual beneficial effect on smoking after the discontinuation of OROS-MPH. In general, the effect of stimulants on ADHD symptoms requires maintenance treatment²⁴, and we expect that its effect on smoking cessation would therefore also not be long-lasting. Of secondary interest, previous studies²⁵ of combination strategies such as nicotine replacement and bupropion generally removed both treatment at the same time—this study, which discontinued OROS-MPH first, provides some results specific to this treatment. We therefore hypothesize that whatever effect the medication might have on smoking cessation would tend to wash out as patients discontinued the medication. We also hypothesize that if there was a residual treatment effect, it would similarly be divergent between the higher and lower baseline ADHD groups.

Methods

Design of the Parent Trial

Details of the parent trial, including full inclusion and exclusion criteria, were described elsewhere¹³. Briefly, this study was conducted in the National Drug Treatment Clinical Trials Network (CTN) between December 2005 and January 2008. Data were collected from six study sites across the country: two community substance abuse treatment programs, two ADHD clinics, two smoking cessation clinics. The trial enrolled adult smokers with ADHD (n=255) into an 11-week treatment program with a four-week pre-quit phase and a seven-week post-quit phase. The main relevant inclusion and exclusion criteria are smokers between 18 and 55 in good health and meeting criteria for ADHD as assessed by ACDS v1.2²⁶, and ADHD Rating Scale (ADHD-RS)²⁷ total score > 22, and to smoke at least 10 cigarettes per day for at least 3 months; candidates were excluded if they were a significant suicide/homicide risk, used tobacco products other than cigarettes in the past week, or had a positive urine screen for an illicit drug or met DSM-IV criteria for a current abuse or dependence for any psychoactive substance other than nicotine, as well as previous treatment with stimulants for ADHD. Participants received a fixed dose of OROS-MPH (titrated to 72mg/d or the highest dose tolerated) or placebo, randomized at 1:1 ratio, stratified by site. They received medications during weeks 1–11, used 21 mg nicotine patches daily beginning on the target quit date (i.e., the fifth day of week 4), and received smoking cessation counseling at each weekly clinic visit. At the week 11 visit, OROS-MPH was discontinued and the participants were provided with 14mg/d patches for the study weeks 12 and 13, and 7mg/d patches for study week 14 as part of standard clinical treatment during a post-treatment follow-up period. Patients returned to the clinic on Week 15 for one final follow-up appointment.

Outcomes and Measures

Main outcomes including prolonged and complete abstinence and selected secondary outcomes up to week 11 have already been reported in previous studies^{13,14,16,18,28}. We analyzed post-trial smoking cessation outcome measures after the primary intervention (OROS-MPH vs. placebo) of the study was discontinued at week 11. 7-day self-reported abstinence (defined as no cigarette use, not even a puff) was assessed with the standard Time Line Follow Back in weeks 12, 13, 14 and 15 reported at the week 15 follow-up visit. ADHD symptom severity was assessed at baseline (week 0) with the Adult ADHD Rating Scale (ADHD-RS), a standard 18-item scale a rating for the severity of the symptoms—scoring is based on a 0–4 Likert-type scale with symptoms such as carelessness, difficulty sustaining attention and difficulty listening to others²⁷. We analyzed ADHD-RS in two different ways: 1) using it as a continuous covariate; 2) for exploratory purposes, dividing the patients into higher versus lower baseline severity group as in our previous studies based on a median-split of the sample^17,29: ADHD-RS ≥ 36 defined the higher baseline-severity group.

Data Analysis

We conducted longitudinal analysis with smoking outcomes for follow-up weeks 12 to 15 using linear mixed effect models, with time and treatment as fixed effects and patient as random effects. Individuals dropped out prior to the final visit (18% missing values) were treated as smoking throughout the follow-up period to maintain consistency with previous studies^13,16,30. Site was treated as fixed effect and removed from the model if not significant (p>0.1), similar to previous study¹³. The following covariates were added in an adjusted model: baseline ADHD severity, age, gender, ethnicity (white or not white), baseline total Fagerstrom score. There were otherwise no missing values for the baseline measures or data collected throughout Time Line Follow Back, and the mixed effect models were fitted using maximum likelihood. Statistical tests were conducted at a 5% Type I error rate (two-sided) for all measures. All analyses were conducted in R Version 3.1.1 using lme4 package.

Results

Participant Characteristics

We included all of the participants in the original trial. In the study, 255 individuals were randomized to OROS-MPH (N=127) or placebo (N=128), with a mean age of 38.1 (SD =10.38) in the OROS-MPH group and age = 37.5 (SD = 9.57) in the placebo group. 60.6% in the OROS-MPH group are male (vs. 52.3% in the placebo group). 86.4% in the OROS-MPH were white ethnicity (vs. 78.9% in the placebo group). There was no difference in the overall baseline ADHD-RS between the OROS-MPH group (mean = 36 SD = 7.1) and the placebo group (mean = 36.7, SD = 7.5). There was also no difference in baseline history of psychiatric conditions or smoking history between the two groups¹³. Approximately 18% (47) of the participants did not attend the week 15 clinic visit. No statistically significant effect of treatment group assignment (logistic regression, p = 0.90) or ADHD baseline severity (p=0.73) was detected on whether an individual missed the final visit. No significant effect was detected for any of the covariates that were included on whether a participant missed the final visit.

Overall Treatment Effect on Abstinence During the Follow-up Period

Overall abstinence rate decreased in time in the follow-up month from week 12 to week 15 in both the OROS-MPH group and the placebo group (Fig. 1A). In the OROS-MPH group, 42% participants were abstinent at week 11 and 26% were abstinent at week 15; in the placebo group, 43% participants were abstinent at week 11 and 35% were abstinent at week 15. To assess for segregated effects of time and treatment, we first built a mixed effect model with only main effects of time and treatment, with site and patient as random effects. The model detected an effect overall effect for time (adjusted beta = −0.05, p < 0.001), but not for treatment (adjusted beta = 0.046, p = 0.39). We also did not detect a treatment by time interaction (beta = 0.02, p = 0.06). No significant effect was detected for any of the covariates that were included.

Figure 1. Effect of OROS-MPH on Smoking Abstinence Rates After Medication Discontinuation.

A) Overall, at the end of the study (Week 11), patients stopped receiving either OROS-MPH or placebo. Abstinence rates decreased in the following month for both OROS-MPH and Placebo groups. B) For the higher baseline severity group, the OROS-MPH group showed a significantly improved abstinence rate over placebo group, which was maintained throughout the 1-month follow-up period. (B) For the lower baseline severity group, the OROS-MPH group initially had a lower abstinence rate, however, the differences between the two groups decreased and no difference was detected by the end of the follow-up period between OROS-MPH and placebo groups.

Testing Treatment by Baseline ADHD Severity Interaction

Detecting no significant overall effect of treatment in our first model, we then focused on the hypothesis that there might be an interaction between treatment and baseline ADHD severity, similar to the interaction observed for the primary outcome in the parent study¹⁶. We first treated baseline ADHD symptom severity as a continuous covariate and built a mixed effect model including two additional terms: time by treatment and treatment by ADHD severity interactions. This model detected significant fixed effects in time, treatment, and a significant treatment by baseline ADHD severity interaction (Table 1A), but not in time by treatment interactions. No significant effect of any of the included covariates were detected.

Table 1 -. Summary of Results of Longitudinal Analyses.

Beta represents coefficients in logistic regression and linear mixed effect model. Bolded terms were statistically significant with a p-value threshold of 0.05. Parenthesis indicates 95% Confidence Interval.

A. Longitudinal Analysis of Treatment by ADHD Severity Interaction: Adjusted coefficients
	Time	Treatment	Baseline ADHD Severity	Time by Treatment Interaction	Treatment by Baseline ADHD Severity Interaction
Beta	−0.06	−0.59	−0.01	−0.02	0.02
Std Err	0.01	0.28	0.01	0.01	0.01
Odds Ratio	0.94 (0.92–0.95)	0.55 (0.32–0.97)	0.99 (0.98–1.00)	1.02 (1.00–1.05)	1.02 (1.00–1.03)
P-value	<0.001	0.04	0.22	0.07	0.04
B. Effect of the Treatment on Self-Reported Abstinence In Higher Baseline Severity Group (Linear Mixed Effect Model Week 12–15):
	Time	Treatment	Time by Treatment Interaction
Beta	−0.05	0.18	0.02
Std Err	0.01	0.07	0.02
Odds Ratio	0.95 (0.93–0.96)	1.19 (1.03–1.38)	1.02 (0.87–1.19)
P-value	<0.001	0.02	0.37
C. Effect of the Treatment on Self-Reported Abstinence In Lower Baseline Severity Group (Linear Mixed Effect Model at Week 12–15):
	Time	Treatment	Time by Treatment Interaction
Beta	−0.07	−0.16	0.03
Std Err	0.01	0.09	0.02
Odds Ratio	0.92 (0.91–0.96)	0.84 (0.71–1.02)	1.03 (0.99–1.07)
P-value	0.001	0.06	0.07

Visualizing Trajectories of Individuals with Higher vs. Lower Baseline ADHD

To better appreciate the nature of the treatment by baseline ADHD severity interaction, we explored the stratified trajectories based on higher and lower baseline severity subgroups^17,18,29. We first visualized the trajectories of abstinence rates in the two groups (Figure 1B, C). In the higher baseline severity group, abstinence rates were consistently higher for the OROS-MPH group compared to the placebo group (40% vs. 20% at week 15, OR = 2.63, p = 0.028, Fisher’s Exact Test). In the lower baseline severity group, the higher abstinence rates in the placebo group at week 11 gradually decreased and eventually converged to the abstinent rates in the OROS-MPH group (29% vs. 32% at week 15, p = 0.98). We conducted the same linear mixed effect model with treatment, time, and treatment by time interaction and revealed a consistent pattern of findings: in the higher baseline severity group, there was a significant treatment effect and a time effect, though no treatment by time interaction. In the lower baseline severity group there was only a time effect, and no evidence for treatment effect or treatment by time interaction. No significant treatment by time interaction was detected in either group.

Discussion

In this secondary analysis of the Adult Smokers with ADHD Trial (CTN-0029), we examined the post-trial follow-up data to see whether OROS-MPH had an effect on whether patients maintained smoking cessation after OROS-MPH was discontinued. We expected that there would be relapses over time, both related to discontinuation of nicotine patch and to discontinuation of OROS-MPH, which improved abstinence rates during the trial for the higher baseline severity group. This was confirmed by our results. We hypothesized that since OROS-MPH had divergent effects for patients with higher vs. lower baseline severity, it would also have similar effects in the follow-up period. We observed a general increase of relapse in time, consistent with a loss of the effect of nicotine patch in both subgroups. Surprisingly, while OROS-MPH continued to exert a more durable positive effect on maintaining abstinence in patients with higher baseline ADHD severity, the possible counterproductive effect in individuals with lower baseline ADHD severity disappeared after medication discontinuation. This pattern of findings suggests that placebo patients with low ADHD lose the benefit (presumably of patch), while in the high ADHD group the benefit of OROS-MPH seems more durable even when the patch was tapered.

Our results, in combination with our previous results¹⁸ that OROS-MPH did not directly affect more proximal secondary outcomes such as withdrawal and craving, suggest that smokers with ADHD may have benefitted in other ways from OROS-MPH, such as better organization to allow for the initiation and leverage of coping skills for abstinence, and this benefit may sustain when OROS-MPH is discontinued. In particular, some treatments in substance use disorder show this type of characteristics: psychotherapy programs such as cognitive behavioral therapy may also improve abstinence even after the completion of treatment (Carroll et al., 2009). The psychotherapy program “Smoke Free and Living It” (Croghan et al., 2012) delivered in our trial was manually driven and behaviorally focused, following basic cognitive behavioral principles flexibly. One might speculate OROS-MPH improved focus and executive functioning, allowing patients to learn new skills to better deal with cravings and relapses, and potentiated the positive effect of nicotine patch in relieving and coping with these symptoms cognitively especially as it was tapered. While our participants were not directly assessed in terms of cognitive outcomes and responses to psychotherapy either during the OROS-MPH or during the follow-up period when OROS-MPH was discontinued, these hypotheses could be tested in future studies that might incorporate cognitive performance evaluations.

In individuals with lower baseline ADHD severity, our results show that the apparent counterproductive effect on smoking cessation was short-lived. This is a reassuring result. Human laboratory studies suggest that in the short term, psychostimulant administration potentiates responses to immediate reward^7–9. Our results suggest, however, that the apparent discrepancy between possible harmful effects of psychostimulants identified in human laboratory findings and some positive findings in clinical trials in treating ADHD as a co-morbidity in substance use disorder, may be related to complex effects relating to dosing and baseline ADHD severity—i.e. too high of a dose of psychostimulant is harmful, especially for patients with lower baseline severity. This hypothesis may be tested using more sophisticated designs, such as adaptive dose-finding designs³¹.

Limitations to the study include the short follow-up time, the lack of measure of secondary outcomes such as ADHD symptom severity, craving and withdrawal, other psychiatric symptoms during the post-ORO-MPH follow-up period, as well as limitations inherent in the parent trial. In particular, it was observed that stopping smoking reduced anxiety and depression in this study²⁸, but these outcomes were not measured during the follow-up period, and hence we could not demonstrate whether recurrence of these symptoms were associated with relapse. In pragmatic settings, some individuals cannot tolerate psychostimulants, and it is possible that non-stimulant treatments of ADHD may also improve smoking cessation. A valid pragmatic clinical strategy is a trial of OROS-MPH as an add-on to nicotine patch, and if a paradoxical effect occurs, stop the medication, or reduce the dose, or start with lower dose in patients with lower ADHD severity. These possibilities should to be considered in future research. The lack of biochemical verification during the follow-up period is also a major limitation, the reported patterns of relapse may be imprecise or influenced by recall bias. An important limitation in our study is the lack of query of participants taking other medications (in particular other smoking cessation products or psychostimulants) during this post-trial follow-up period—we expect this to be relatively infrequent as the study covers the period immediately after the trial period. However, any interventions of this type would influence the validity of our results (i.e. perhaps patients feel that they are more likely to relapse would incidentally treat their ADHD more frequently). Future studies may be improved if assessments were more frequently conducted, for example, through mobile technology³². Another limitation in this trial is that everyone received the same standard smoking cessation counseling, and some participants may have responded to this program better than others--an additional factor that might be evaluated in a future study. It would also be interesting in a future study of longer duration to see whether OROS-MPH treatment effect is lost over time, and whether that is related to the re-emergence of ADHD symptoms on an individual basis.

In conclusion, our results continue to suggest that OROS-MPH might be a useful treatment for smoking cessation for patients who have co-morbid ADHD of higher severity. Our results also suggest that longer-term follow-up data might be helpful to clarify whether a sustained course of treatment of OROS-MPH might be of further benefit in smokers with ADHD.