Biochemical and genetic predictors and correlates of response to lamotrigine and folic acid in bipolar depression: Analysis of the CEQUEL clinical trial

Objectives CEQUEL (Comparative Evaluation of QUEtiapine plus Lamotrigine combination versus quetiapine monotherapy [and folic acid versus placebo] in bipolar depression) was a double‐blind, randomized, placebo‐controlled, parallel group, 2×2 factorial trial that examined the effect of adding lamotrigine and/or folic acid (FA) to quetiapine in bipolar depression. Lamotrigine improved depression, but its effectiveness was reduced by FA. We investigated the baseline predictors and correlates of clinical response, and the possible basis of the interaction. Methods The main outcome was change in depressive symptoms at 12 weeks, measured using the Quick Inventory for Depressive Symptoms—self report version 16 (QIDS‐SR16). We examined the relationship between symptoms and lamotrigine levels, and biochemical measures of one‐carbon metabolism and functional polymorphisms in catechol‐O‐methyltransferase (COMT), methylene tetrahydrofolate reductase (MTHFR) and folate hydrolase 1 (FOLH1). Results Lamotrigine levels were unaffected by FA and did not differ between those participants who achieved remission and those with persisting symptoms. When participants with subtherapeutic serum levels were excluded, there was a main effect of lamotrigine on the main outcome, although this remained limited to those randomized to FA placebo. None of the biochemical measures correlated with clinical outcome. The negative impact of FA on lamotrigine response was limited to COMT Met carriers. FOLH1 and MTHFR had no effect. Conclusions Our results clarify that FA's inhibition of lamotrigine's efficacy is not a pharmacokinetic effect, and that low serum lamotrigine levels contributed to lamotrigine's lack of a main effect at 12 weeks. We were unable to explain the lamotrigine−FA interaction, but our finding that it is modulated by the COMT genotype provides a starting point for follow‐on neurobiological investigations. More broadly, our results highlight the value of including biochemical and genetic indices in randomized clinical trials.


| INTRODUCTION
Depression rather than mania accounts for the majority of the burden of increased mortality and long-term disability in bipolar disorder. 1 Treatment of bipolar depression remains a major clinical challenge with few treatment trials, particularly beyond the acute phase of illness, nor have many examined combination therapies, which are the norm in clinical practice. 2,3 We recently reported the results of the CEQUEL trial (Comparative Evaluation of QUEtiapine plus Lamotrigine combination versus quetiapine monotherapy [and folic acid versus placebo] in bipolar depression). 4 CEQUEL was a multi-centre, double-blind, randomized clinical trial, with a 2×2 factorial design. It examined the efficacy of adding lamotrigine and/or folic acid (FA) to quetiapine monotherapy for depressive symptoms in participants with bipolar disorder. Participants were followed up at 12, 22 and 52 weeks, with changes in selfreported depressive symptoms at 12 weeks as the prespecified outcome. The rationale for CEQUEL was that adding lamotrigine, which has a lengthy (6-week) titration period, to quetiapine, which is relatively quick-acting and is thought to act via different pharmacological mechanisms, might lead to synergistic long-term outcomes. 4 Similarly, FA, which is widely used as an over-the-counter vitamin supplement, has shown some clinical benefit in major depression, 5 consistent with reports of a cerebral folate deficiency in patients with treatmentrefractory depression. 6 Thus, we hypothesized that adding lamotrigine to quetiapine would be superior to quetiapine monotherapy, and that FA might be of additional benefit, independently of lamotrigine.
The findings of CEQUEL were intriguing: instead of a clear benefit of lamotrigine over placebo at 12 weeks (P=.066), there was an interaction between FA and lamotrigine (P=.028). Specifically, FA appeared to block the therapeutic effect of lamotrigine at 12 weeks, although this interaction disappeared at later time points and lamotrigine was superior to placebo at 52 weeks. 4 FA's actions are generally ascribed to its effects on one-carbon metabolism. FA is reduced to folates, which are essential cofactors and co-substrates in this pathway ( Figure 1). There are several single nucleotide polymorphisms which influence one-carbon metabolism ( Figure 1). The folate hydrolase 1 (FOLH1) C 484 T (rs202676) polymorphism may influence the binding potential of FOLH1. 7 Dietary folates (but, notably, not FA) must be hydrolyzed by FOLH1 before they can be absorbed. 8 Despite not directly influencing FA uptake, the FOLH1 C 484 T genotype has been implicated in the beneficial effect of FA on negative symptoms in patients with schizophrenia (Roffman et al., 2013). The methylene tetrahydrofolate reductase (MTHFR) C 677 T (rs1801133) polymorphism alters the activity of the encoded MTHFR (the T allele results in an enzyme with lower activity than that encoded by the wild-type C allele), which is a key regulator of the folate cycle. 9 Similarly, the catechol-O-methyltransferase (COMT) Val 158 Met (rs4680) polymorphism affects COMT's activity (the enzyme encoded by the ancestral Val allele has higher activity than that encoded by the Met allele). 10 Because COMT converts its methyl donor into a homocysteine precursor, its activity indirectly affects one-carbon metabolism ( Figure 1). Notably, we showed interactive effects of these MTHFR and COMT polymorphisms on total homocysteine levels, such that levels were elevated in individuals carrying both MTHFR TT and COMT Val genotypes compared with other genotype groups, suggesting that the effects of these functional polymorphisms may be synergistic. 11 In contrast, lamotrigine has multiple pharmacological actions and, although its anticonvulsant actions may result from its blockade of voltage-gated sodium channels, it is unclear which of these are relevant for its efficacy in bipolar disorder. 12 Interestingly, cellular and animal models suggest that lamotrigine has neuroprotective effects, mediated by reductions in oxidative stress. [13][14][15] These findings are notable given evidence for increased oxidative stress in patients with bipolar disorder. 16,17 Given that the lamotrigine−FA interaction was unexpected, this aspect of the results of CEQUEL should be treated with caution.
However, if correct, it would be of clinical significance given the widespread use of FA as a dietary supplement, particularly by women in the periconceptual and prenatal period, and its increasingly widespread fortification in flour and other grains around the world.
Here we present secondary analyses of the CEQUEL data, conducted with two goals in mind: first, to identify potential baseline F I G U R E 1 The one-carbon cycle. The actions of folate hydrolase 1 (FOLH1), methylene tetrahydrofolate reductase (MTHFR) and catechol-Omethyltransferase (COMT) are highlighted in blue. Note that the activity of all S-adenosylmethionine (SAM)-dependent methyltransferases results in SAM's conversion to S-adenosylhomocysteine (indicated by the conversion of "R" to "R-CH 3 "). THF, tetrahydrofolate; MTHF, 5,10-methylene tetrahydrofolate [Colour figure can be viewed at wileyonlinelibrary. com] predictors and correlates of the response to lamotrigine; second, to investigate the possible mechanistic basis of the observed lamotrigine−FA interaction. To address these issues, we conducted protocoldefined analyses of biochemical and genetic measures, particularly related to one-carbon metabolism, as well as exploratory assessments of lamotrigine and folate levels in the blood.

| Study design and participants
CEQUEL was a double-blind, randomized, placebo-controlled, parallel group, 2×2 factorial trial that was conducted across 27 UK sites.
CEQUEL is described in detail in our earlier publication 4 and the trial protocol is available from JRG. Briefly, following a 7−14-day run-in on quetiapine, patients with bipolar disorder type I or II (according to DSM-IV 18 criteria based on clinician interview) were randomly assigned to added lamotrigine or placebo and, separately, to FA or placebo. Lamotrigine was commenced at 25 mg daily and increased gradually to 200 mg (100 mg/day with concurrent valproate and 400 mg/day with concurrent combined oral contraceptives).
Participants who were not already taking FA, and who had no contraindications to it, were separately randomized to FA (500 μg/day, a dose close to that of many over-the-counter supplements) or placebo. Participants were followed up at 12, 22 and 52 weeks; the main outcome was the change in depressive symptoms at 12 weeks, meas-

| Biochemical measures and genotyping
Plasma concentrations of folate, vitamin B 12 , total transcobalamin, holo-transcobalamin and total homocysteine were determined as described previously 19 in individuals who provided samples for biochemical analysis. Total homocysteine was not determined if the plasma sample was processed outside of 48 hours post-collection; therefore, this information is missing for five individuals (see below for group allocations) for whom other biochemical measures were available.

| Data analysis
Statistical analyses were conducted in IBM SPSS Statistics version 24 (IBM, Armonk, NY, US). With the exception of defining "remission" and "non-remission" groups, described below, the only clinical variable examined was the CEQUEL primary outcome: the change in QIDS-SR16 score from baseline to 12 weeks (with baseline QIDS score included as a covariate). With the exception of the comparison of serum lamotrigine levels, detailed further below, analyses were limited to individuals randomized to both lamotrigine and FA. Biochemical measures were highly skewed and so were log 10 -transformed (with a constant added prior to log transformation to ensure all values were positive).
Genotyping results were pooled into two groups to increase statistical power. Specifically, COMT Val homozygotes were compared to Met carriers, and MTHFR C carriers to T homozygotes, as in our earlier study. 11 For FOLH1, C carriers were compared to TT homozygotes, since C is the minor allele. Effects of COMT and MTHFR were examined within the same analyses, given that we have previously shown interactive effects between them on one-carbon metabolism. FOLH1 was examined in separate analyses, since there was no a priori reason to anticipate a direct interaction with either COMT or MTHFR.
Lamotrigine levels were compared between FA positive (n=20) and FA negative (n=23) groups, and between those in remission from depression at 12 weeks (n=13; defined as 12-week QIDS-SR16 score ≤5) and a non-remission group (n=30; defined as 12-week QIDS-SR16 score ≥6), using one-way analysis of variance (ANOVA). The effects of genotype(s) and/or treatment (lamotrigine vs lamotrigine placebo; FA vs FA placebo) were examined using ANOVA in those randomized to The relationship between lamotrigine levels and clinical outcome was explored using partial correlation, controlling for FA group.
Spearman's correlations were used to investigate the relationship between biochemical variables (baseline and 12-week change scores) and clinical outcome.

| The effect of lamotrigine may be limited by low serum levels in some individuals
We compared lamotrigine levels between FA-positive and FA-negative individuals to investigate whether the lamotrigine×FA interaction ob- Notably, a number of individuals assigned to lamotrigine had low or undetectable serum lamotrigine levels ( Figure 2A). Therefore, we

| The effect of FA randomization is influenced by COMT genotype
We investigated whether functional polymorphisms in genes related to one-carbon metabolism might influence clinical response, given that FA's effects are generally ascribed to its modulation of this bi-

| FA supplementation increased folate and decreased total homocysteine
We explored whether biochemical indices of one-carbon metabolism (either indices at baseline or their change over time) related to clinical outcome, as well as testing whether folate levels increased (and total homocysteine decreased) over time in those randomized to FA, as would be expected.
There were no differences in baseline biochemical measures ( Baseline total TC (log 10 ) Spearman's rho Homocysteine levels decreased from baseline to 12 weeks in those randomized to FA (Figure 4). There was a main effect of time

| DISCUSSION
Our findings clarify and extend the primary analysis of CEQUEL in several ways. Firstly, they argue against a pharmacokinetic explanation for the observed clinical lamotrigine−FA interaction, since serum lamotrigine levels did not differ between those who were randomized to FA and those who were not. Secondly, the lack of a main effect of lamotrigine on depressive symptoms at 12 weeks appeared to be due to low or absent lamotrigine levels in some participants. Thirdly, we demonstrate that the interactive effect of FA and lamotrigine on clinical outcome may be modulated by the COMT genotype. These analyses highlight the value of biochemical and genetic measures to help explain and understand clinical trial outcomes.

| Lamotrigine levels and clinical response
A notable aspect of our current findings is that a proportion of participants randomized to lamotrigine had low or undetectable

| One-carbon metabolism and clinical response
We explored a number of biochemical measures of one-carbon metabolism in order to investigate the possible biological basis of the lamotrigine−FA interaction observed in our primary analysis. In the event, none shed light on this question (for example, clinical improvement did not correlate with serum folate or total homocysteine levels, although it remains possible that correlations may have been observed with other markers of one-carbon metabolism, e.g. 5-methyltetrahydrofolate 5 ). However, the biochemical data confirmed that FA augmentation robustly both increased folate and decreased total homocysteine, and these changes were negatively correlated, in line with expectations ( Figure 1).
We also studied the relationship between functional polymor-

| CONCLUSIONS
Our secondary analyses help clarify the findings from the CEQUEL trial and have implications for future studies of lamotrigine and FA.
First, they suggest that the lack of a main effect of lamotrigine on the primary outcome (change in depressive symptoms from baseline to 12 weeks) may be due in part to low lamotrigine levels in some participants. Therefore, where possible, we recommend that future ran-