Combination Therapy of Varenicline With Nicotine Replacement Therapy Is Better Than Varenicline Alone

A Systematic Review and Meta-analysis of Randomized Controlled Trials

Ping-Hsun Chang; Chien-Hsieh Chiang; Wei-Che Ho; Pei-Zu Wu; Jaw-Shiun Tsai; Fei-Ran Guo


BMC Public Health. 2015;15(689) 

In This Article


Our research identified three smoking cessation trials comparing varenicline combined with nicotine patch versus varenicline combined with placebo patch. No other types of nicotine product were combined with varenicline in these trials. We found no trials comparing combination therapy with NRT alone that met our inclusion and exclusion criteria. One RCT that compared combination therapy and NRT was not included because it combined varenicline and counseling to get long-term NRT users to quit NRT.[17] After meta-analysis, our results demonstrated favorable effects of combination therapy in both early and late outcomes. To our knowledge, this is the first systematic review and meta-analysis on this issue.

In our main analysis, the early outcomes in varenicline combined with placebo patch group were consistent with previous studies using varenicline mono-therapy (abstinence rate 44.4 % vs 43.9 %).[18] The late outcomes were also similar (abstinence rate 32.4 % vs 29.7 %). In sensitivity analysis, we identified the largest RCT[9] which markedly influenced the results. After eliminating this RCT, the favorable effects of both early and late outcomes became insignificant. We identified two distinguishing factors that might have caused this difference. First, there were more female subjects in this RCT. Previous studies have revealed that the effect of varenicline did not differ among genders.[19–21] On the other hand, some studies yielded inconsistent reports on whether male subjects using NRT had higher abstinence rates than females.[22–25] If nicotine patch use had been less effective in females, the higher percentage of females in this RCT would bring about less effective outcomes, which was not the case. Therefore, gender difference did not seem to contribute to the treatment effect of this RCT.

Second, this RCT[9] used a pre-cessation nicotine patch. A meta-analysis conducted by Shiffman S et al. showed that pre-cessation nicotine patch significantly increased abstinence rates at 6 weeks (OR = 1.96, 95 % CI 1.31 to 2.93) and at 6 months (OR = 2.17, 95 % CI 1.46 to 3.22).[26] However, this positive effect was not consistent in the following meta-analyses.[6,27,28] These meta-analyses showed that pre-cessation nicotine patch and gum had a moderate but insignificant increase in abstinence rates. Pre-cessation nicotine patch appeared to be more effective than pre-cessation nicotine gum.[27] Therefore, we favored that the effect of pre-cessation nicotine patch contributed to the better outcomes in this RCT.

The rationale of combination therapy of varenicline with NRT resides in the hypotheses that 1) varenicline does not fully saturate α4β2 nicotinic acetylcholine receptors; 2) varenicline does not completely replace the dopaminergic effect of smoking.[29] A standard-dose of varenicline (1.0 mg) could achieve higher abstinence rates than low-dose varenicline (0.5 mg).[3] Further saturation of the receptors seemed to explain the additive effect of NRT. However, a neuropharmacological study utilizing positron emission tomography revealed that a single dose of 0.5 mg varenicline could saturate α4β2 receptors in the human brain.[30] It deserves a debate whether the combination to α4β2 nicotinic acetylcholine receptors, and subsequent mesolimbic dopamine release is the only pathway that causes reward in smoking. Nicotine addiction develops from complex pathways, and individual genotypes influence both smoking behavior and treatment effects.[31,32] The modulation of α4β2 receptors might not be the only pathway to release dopamine, and dopamine might not be the only neurotransmitter involved in smoking behavior. More studies are required to explore the mechanism of combining varenicline with NRT.

The event rates of adverse effects were similar in the two groups. The only significant increase in adverse events was skin reactions in one RCT.[9] The event rate of skin reactions was comparable to those in nicotine patch mono-therapy.[28] Other adverse events of combination therapy were not higher than findings from previous studies of varenicline mono-therapy.[3] The birth of an infant with Down syndrome (trisomy 21) in one study[9] was considered relevant to varenicline, classified as a Pregnancy Category C drug. Koegelenberg et al. considered that the causality was less likely because the association was not observed in post-marketing researches.[33,34] Varenicline combined with nicotine patch appeared to be safe and tolerable in smoking cessation.

The Question of Small Study Number

Critics might argue that it was too early to perform a meta-analysis when there were only three RCTs available. Borenstein et al. suggested that it makes sense to perform a meta-analysis as soon as we have two studies, since a summary based on two studies yields a more precise estimate of the true effect than either study alone.[35] However, the estimate of the pooled effects would have poor precision. Three solutions are suggested under such circumstance.[36] One option is to report the separate effects and not report a summary effect. Readers are expected to understand that authors cannot draw conclusions about the effect size. The problem is that some readers may revert to 'vote counting' and possibly reach an erroneous conclusion. Another option is to perform a fixed effect analysis. This approach yields a descriptive analysis of the included studies, but does not allow us to make inferences about a wider population. A third option is to take a Bayesian approach, where the estimate of variance is based on data from outside of the current studies. We selected the second approach and used fixed effect model in our analysis.

Strengths and Limitations

In our study, we searched the major databases with rigorous strategies. There were duplicated authors who selected the articles independently, allowing for a low probability that an important study was missed. The included studies had high quality and the Jadad score ranged from 4 to 5. The heterogeneity between the selected studies was low, and there were no significant publication biases. We evaluated both early and late outcomes which was more comprehensive than short-term evaluations. However, there were some limitations. We did not search grey literature or un-published data. Trials that were less known might have been missed. The strength of our research was compromised by the small number of trials. The largest RCT which had the greatest influence to our results was different from the other RCTs in demographic characteristics and treatment design. The impact was that our conclusions could not be generalized to other populations. Also, the funnel plot and tests of publication bias had low power to detect a potential bias. In our review, the adverse events of depression and skin reactions were only reported in one study. There was no report of cardiovascular or suicidal events. The safety of combination therapy requires further investigations.