Genetic Influence on Efficacy of Pharmacotherapy for Pediatric Attention-Deficit/Hyperactivity Disorder: Overview and Current Status of Research

Abstract

Multiple stimulant and non-stimulant medications are approved for the treatment of Attention-Deficit/Hyperactivity Disorder (ADHD), one of the most prevalent childhood neurodevelopmental disorders. Choosing among the available agents and determining the most effective ADHD medication for a given child can be a time-consuming process due to the high inter-individual variability in treatment efficacy. As a result, there is growing interest in identifying predictors of ADHD medication response in children through the burgeoning field of pharmacogenomics. This article reviews childhood ADHD pharmacogenomic efficacy studies published during the last decade (2009–2019), which have largely focused on pharmacodynamic candidate gene investigations of methylphenidate and atomoxetine response, with a lesser number investigating pharmacokinetic candidate genes and genome-wide approaches. Findings from studies which have advanced the field of ADHD pharmacogenomics through investigation of meta-analytic approaches and gene*gene interactions are also overviewed. Despite this progress, no one genetic variant or currently available pharmacogenomics test has demonstrated clinical utility in pinpointing the optimal ADHD medication for a given individual patient, highlighting the need for further investigation.

1. Introduction

Attention-Deficit/Hyperactivity Disorder (ADHD)--a common childhood neurodevelopmental disorder affecting 3–7% of children worldwide[1, 2]--is associated with marked impairment in academic, social, and family functioning.[3, 4] Medications for ADHD that effectively ameliorate symptoms of inattention, hyperactivity, and impulsivity are available,[5] with stimulant medications (methylphenidate and amphetamine/dextroamphetamine) considered to be first-line treatment and non-stimulants (atomoxetine and alpha-2-agonists) considered to be second-line.[3] Although data showing that ADHD medications can enhance long-term outcomes are sparse, evidence is emerging that treatment may improve academic outcomes[6, 7] and decrease risk for motor vehicle accidents,[8, 9] injuries,[10] depression,[11] and criminality.[12]

Despite the above evidence of ADHD medication efficacy and improved outcomes in groups of individuals with ADHD, finding an effective medication and dosage for a given child with ADHD can be a complex process. Currently, no phenotypic or patient factors have been shown to consistently predict ADHD medication response.[13] As a result, pediatric clinical practice guidelines continue to recommend empirical determination of each child’s ADHD treatment regimen through a process of gradual upward dose titration and a trial-and-error approach to trying different preparations.[3] This trial-and-error method often yields marked delays in achieving optimal symptom remission, as well as premature termination of treatment due to provider or family frustration with the titration process.[14]

In other branches of medicine, progress in predicting medication response has been made through pharmacogenomics (PGx), which utilizes knowledge of an individual’s genes to predict medication efficacy, dosing, and adverse effects. In fact, evidence has accrued to the point where the Clinical Pharmacogenetics Implementation Consortium (CPIC) has established universal guidelines for genotype-based drug prescription[15] for more than 20 drug-gene pairs.[16] Clinical implementation of these guidelines began as early as 2011 with the launch of the Pharmacogenomics Research Network Translational Pharmacogenomics Program at 8 different medical centers, including the University of Maryland. There, an anti-platelet pharmacogenetics initiative has shown much success in prescribing heart catherization patients anti-platelet drugs based on CYP2C19 genotype.[17] Unfortunately, ADHD PGx has not fully realized its promise to facilitate more expeditious identification of the optimal medication regimen for each child with ADHD. Nonetheless, numerous pediatric ADHD pharmacogenomics studies have been conducted in recent years and warrant further examination to guide future research and clinical care efforts. Hence, this article provides an overview of pediatric ADHD PGx studies published from 2009–2019 to provide an update to our prior review.[18]

This review is limited to ADHD PGx studies enrolling participants who were <18 years old which used 1) standardized measures to assess improvements in symptoms or global functioning with medication treatment (e.g., ADHD rating scales or Clinical Global Impression [CGI] ratings) and/or 2) direct observations of inattention or activity level. ADHD PGx investigations of academic or neuropsychological outcomes are beyond the scope of this review. A comprehensive literature search was conducted using the PubMed (http://www.ncbi.nlm.nih.gov/sites) and PsycINFO (http://www.apa.org/psycinfo/) databases to identify relevant English-language articles. Search terms included the following: ‘pharmacogen*’ AND ‘ADHD’ OR ‘attention-deficit’ OR ‘attention’ OR ‘hyperactiv*’ OR ‘methylphenidate’ OR ‘amphetamine’ OR ‘atomoxetine’ OR ‘stimulant’ OR ‘psychostimulant’ OR ‘clonidine’ OR ‘guanfacine’ or ‘α-agonist’. References of identified papers were also screened to pinpoint additional relevant literature.

Due to the sparsity of pharmacogenomics investigations evaluating response to amphetamines, guanfacine, and clonidine, this review focuses primarily on methylphenidate (MPH) and atomoxetine (ATX) studies. The MPH and ATX sections each open with a review of findings for PGx studies of the gene encoding their respective metabolizing enzymes. A summary of findings for pharmacodynamic candidate genes that have been significantly associated with MPH or ATX response in at least one prior study follows in the text as well as in Tables 1–6, with additional information about each study’s methods (medication trial duration, dosage used, and primary outcome measures) summarized in Supplemental Tables 1–6. Minor allele frequencies for relevant polymorphisms (extracted from the Ensembl Genome Browser[19]) are also presented in the text. The section for each genetic variant closes with an integration paragraph in cases where this polymorphism has been the focus of a meta-analysis or significant prior literature. Additional genetic variants that did not show significant effects in MPH or ATX PGx studies from 2009–2019 are included in Tables 1–6 (with further study information available in Supplemental Tables 1–6) rather than being discussed in the text.

Table 1.

Pharmacokinetic Candidate Gene Studies of Methylphenidate Efficacy Published from 2009–2019

Gene	Polymorphism(s)	Authors	Design / Sample	Findings
Carboxylesterase 1 (CES1)	Gly143Glu	Nemoda et al 2009 [23]	POL N=122 Hungary	No effect.
	rs3815589 rs2287194 rs2244613 rs2002577 rs2307244 rs2307240 rs12443580	Johnson et al 2013 [27]	POL N=77 Ireland	No effects.

POL=prospective, open-label

2. Methylphenidate Pharmacogenomic Studies

MPH’s main mechanism of action involves increase in synaptic dopamine and norepinephrine via blockade of the dopamine transporter and norepinephrine transporter.[20] Additional MPH actions which lead to an increase in extracellular dopamine and norepinephrine include its agonist activity at the serotonin 1A receptor and its redistribution of vesicular monoamine transporter 2, which regulates release of monoamines from vesicular storage.[21]

2.1. Methylphenidate Pharmacokinetic Genetic Variants

2.1.1. Carboxylesterase 1 (CES1) [Table 1].

The hepatic enzyme CES1 is the main metabolizer of MPH, de-esterifying it to form inactive ritalinic acid.[22]

Prior to 2009, little was known regarding the impact of CES1 variants on MPH efficacy. Since then, two prospective, open-label studies have investigated the link between CES1 polymorphisms and MPH efficacy. Nemoda et al.[23] explored the association between the CES1 Gly143Glu (rs71647871, minor allele frequency [MAF]=0.01–0.04 [T/glutamate]) functional polymorphism and response to MPH. This polymorphism is of particular interest since the glutamate substitution has been associated with a lower level of CES1 enzyme activity and altered MPH metabolism in several pharmacokinetics studies.[24–26] Nemoda et al.’s study, which included five glutamate allele carriers, did not find a link between the Gly143Glu polymorphism and MPH responder status or reduction in hyperactivity/impulsivity or inattention scores. However, they did observe that MPH responders possessing the 143Glu allele required a lower MPH dose for symptom reduction compared to responders who were Gly/Gly homozygotes. The second study, by Johnson et al.[27], evaluated seven CES1 polymorphisms (but not Gly143Glu) and found no association between any of these single nucleotide polymorphisms (SNPs) and MPH dose response or efficacy.[27]

2.2. Methylphenidate Pharmacodynamic Genetic Variants

2.2.1. Adrenergic α_2A –Receptor (ADRA2A) [Table 2].

Table 2.

Neurotransmitter Receptor Candidate Gene Studies of Methylphenidate Efficacy Published from 2009–2019

Gene	Polymorphism(s)	Authors	Design / Sample	Findings
Adrenergic α2A –Receptor (ADRA2A)	MspI	Cheon et al 2009 [38]	POL N=114 South Korea	Improved MPH response with MspI G allele.
	MspI	Froehlich et al 2011 [33]	DBPC N=89 U.S.A.	No effect after multiple comparison correction.
	MspI DraI	Kim et al 2011 [34] Hong et al 2012 [35] Park et al 2013 [36]	POL N=102 for Kim et al N=103 for Hong et al N=115 for Park et al South Korea	Kim et al, Hong et al, and Park et al: No main effects in whole sample analyses. Park et al: Improved MPH response with DraI C/C homozygosity in analyses limited to combined type participants. Hong et al: Interaction effects between the DraI SNP and the DRD4 VNTR, NET G1287A and NET −3081(A/T) polymorphisms.
	MspI	Unal et al 2016 [40]	POL N=108 Turkey	Decreased MPH response with MspI GG genotype.
	MspI DraI	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effects.
	MspI	Huang et al 2018 [39]	POL N=59 Taiwan	Improved MPH response with MspI GG genotype.
Dopamine Receptor D1 (DRD1)	rs4867798 rs251937 rs11749676 rs835540 rs835616 rs835541 rs863126 rs265977	Pagerols et al 2017 [51]	POL N=107 Spain	No effects.
Dopamine Receptor D2 (DRD2)	rs1800497	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	Faster MPH response with C allele.
	rs4630328 rs7131056 rs4245146 rs17529477 rs2002453 rs12363125 rs2283265 rs2242592 rs1554929 rs2234689	Pagerols et al 2017 [51]	POL N=107 Spain	No effect on MPH efficacy for any variant. Trend for rs2283265 T allele carriers to require a higher MPH dose.
Dopamine Receptor D3 (DRD3)	rs6280 rs9825563 rs1800828 rs10934256 rs167770 rs167771 rs324035 rs9880168 rs2134655 rs3732790	Pagerols et al 2017 [51]	POL N=107 Spain	Decreased MPH response nominally associated with rs2134655 G allele. Decreased MPH response with GG haplotype at rs2134655-rs1800828 in whole sample analyses, with effects accentuated in children with prenatal smoke exposure. No effects for other variants.
	rs6280	Fageera et al 2018 [62]	DBPC N=327 boys Canada	Improved MPH response with C/C homozygosity according to teacher but not parent ratings.
Dopamine Receptor D4 (DRD4)	Exon 3 VNTR 120 BP promoter duplication	McGough et al 2009 [49]	DBPC N=82 U.S.A	No effects.
	Exon 3 VNTR	Froehlich et al 2011 [33]	DBPC N=89 U.S.A.	Decreased response on hyperactive-impulsive symptoms across MPH doses with absence of 4-repeat (short) allele.
	Exon 3 VNTR	Hong et al 2012 [35]	POL N=103 South Korea	No independent effect. Interaction effect with the ADRA2A DraI SNP.
	Exon 3 VNTR	Ji et al 2013 [67]	POL N=114 South Korea	No effect.
	Exon 3 VNTR 120 BP promoter duplication rs3758653	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	Decreased MPH response with homozygous short allele promoter duplication. No effects for other variants.
	Exon 3 VNTR	Naumova et al 2017 [66]	DBPC N=374 Canada	Improved MPH response with 7-repeat (long) allele homozygosity based on parent but not teacher ratings.
	rs936465	Pagerols et al 2017 [51]	POL N=107 Spain	No effect.
Dopamine Receptor D5 (DRD5)	rs10033951	Pagerols et al 2017 [51]	POL N=107 Spain	No effect.
5-Hydroxytryptamine (Serotonin) Receptor 2A (HTR2A)	rs7322347	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effect.
5-Hydroxytryptamine (Serotonin) Receptor 2C (HTR2C)	rs6318	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effect.
Metabotropic Glutamate Receptor Subtype 7 (mGluR7/GRM7)	rs3792452	Park et al 2014b [82]	POL N=175 South Korea	Improved MPH response with G/A genotype.
	rs3792452	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effect.
N-methyl-D-aspartate (NMDA) Receptor Subunit 2A (GRIN2A)	rs2229193	Kim et al 2017 [101]	POL N=75 South Korea	No effect.
N-methyl-D-aspartate (NMDA) Receptor Subunit 2B (GRIN2B).	rs2284411	Kim et al 2017 [101]	POL N=75 South Korea	Improved MPH response with C/C genotype.

BP=base pair

DBPC=double-blind, placebo-controlled

POL=prospective, open-label

SNP=single nucleotide polymorphism

VNTR=Variable nucleotide tandem repeat

ADRA2A encodes a norepinephrine autoreceptor whose activation limits norepinephrine release.[28] Prior animal studies suggest ADRA2A may mediate MPH effects, since α₂-adrenoceptor antagonists have been shown to block MPH’s beneficial effects.[29, 30]

MspI (−1291 C>G; rs1800544)

Since 2009, one placebo-controlled and five open-label MPH PGx cohorts have assessed ADRA2A’s −1291 C>G SNP (rs1800544, MAF=0.46 [C]), which results in an MspI restriction site within a putative transcription factor binding site [31] in the gene promoter region.[32] The randomized controlled trial (RCT) reported a main effect such that G allele homozygosity was linked to higher hyperactive-impulsive scores on placebo and across MPH doses, but MspI genotype did not influence MPH dose response.[33] Among the open-label cohorts, two found no effects[34–37], two linked the G allele with improved MPH response[38, 39] and one [40] linked the G allele to diminished MPH response.

Meta-analysis Findings.

Although the findings described above do not depict a clear impact of the MspI variant on MPH efficacy, a recent meta-analysis by Myer et al.[41] integrated data from four studies, including two described above[33, 42], and identified an association between the MspI G allele and improved MPH response (Odds Ratio [OR]=1.69, 95% Confidence Interval (CI):1.12–2.55, p=0.01; I² =80%, indicating considerable heterogeneity in individual study results).

DraI (rs553668)

An A to G substitution in the ADRA2A 3’ Untranslated Region (UTR) (rs553668, MAF=0.33 [A]) creates a DraI site that has also been the subject of ADHD PGx investigation. From 2009–2019, this polymorphism was assessed in two cohorts, and none identified significant independent associations with MPH response in whole sample analyses.[34–37] However, when Park et al. excluded inattentive type participants and limited analysis to children with ADHD-combined type, they found a link between C/C homozygosity and improved MPH response.[36] Significant pairwise interactions of the ADRA2A DraI SNP with the DRD4 exon 3 VNTR, the DraI SNP with the norepinephrine transporter (NET) G1287A SNP, and the DraI SNP with NET −3081(A/T) SNP on MPH treatment response were also reported within the same cohort.[35]

2.2.2. Brain-derived Neurotrophic Factor (BDNF) [Table 3].

Table 3.

Neuronal/Synaptic Plasticity and Synaptic Effector Candidate Gene Studies of Methylphenidate Efficacy Published from 2009–2018

Gene	Polymorphism(s)	Authors	Design / Sample	Findings
Brain-derived Neurotrophic Factor (BDNF)	Val66Met	Kim et al 2011 [34]	POL N=102 South Korea	Improved MPH response with Val/Val homozygosity.
Cadherin 13 (CDH13)	rs6565113	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effect.
Fatty Acid Desaturase 2 (FADS2)	rs498793	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effect.
Glucose-fructose Oxidoreductase Domain Containing 1 (GFOD1)	rs552655	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effect.
Latrophilin 3 / Adhesion G Protein-coupled Receptor L3 (LPHN3/ADGRL3)	Arcos-Burgos et al: rs6551665 rs6551665, rs1947275, rs9683662 haplotype Jain et al: rs6551665 11q haplotype block1	Arcos-Burgos et al 2010 [75] Jain et al 2012 [80]	POL N=240 for Arcos-Burgos et al N=82 for Jain et al U.S.A.	Arcos-Burgos et al: Improved MPH response on inattentive symptoms with rs6551665 G allele in both single marker and rs6551665, rs1947275, rs9683662 haplotype analysis. Jain et al: Improved stimulant response linked to having rs6551665 G/G homozygosity and two copies of the 11q ADHD susceptibility haplotype.
	rs2122643 rs1868790 rs6551665 rs1947274 rs6858066 rs2345039	Choudhry et al 2012 [79]	DBPC N=380 Canada	Improved MPH response with rs6551665 G allele. Improved MPH response with rs1868790 T allele. Improved MPH response for the rs6551665, rs1947274 and rs6858066 GCA haplotype in children whose mothers did not experience prenatal stress. No effects for other variants.
	rs1868790 rs1947274 rs2122643 rs2345039 rs6551665 rs6858066	Labbe et al 2012 [78]	DBPC N=416 Canada	Decreased MPH response linked to rs1947274 C allele, rs2345039 G allele, rs6551665 G allele, and rs6858066 A allele. No effects for other variants.
	rs1947274 rs2345039 rs6551665 rs6551665 and rs1947274 haplotype	Song et al 2014 [76]	POL N=139 South Korea	No independent or haplotype effects.
	rs6813183, rs1355368, rs734644 haplotype rs6551665 and rs1947275 haplotype	Bruxel et al 2015 [77]	POL N=172 Brazil	Faster MPH treatment response for rs6813183, rs1355368, and rs734644 haplotype CGC homozygotes. No effects for the rs6551665 and rs1947275 GT haplotype after multiple comparison correction.
	rs1397548 rs1868790 rs2305339 rs6551655 rs6813183 rs6858066	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	Decreased MPH response for rs1868790 AA homozygotes. No effects for other variants.
Neurotrophin-3 (NTF-3)	rs6332 rs6489630	Kim et al 2011 [34]	POL N=102 South Korea	No effects.
Steroid Sulfatase (STS)	rs12861247 rs17268988	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effects.

¹spans all or part of the genes encoding neural cell adhesion molecule 1 (NCAM1), encompasses the tetratricopeptide repeat domain 12 (TTC12) and the ankyrin repeat and kinase domain containing 1 (ANKK1), and is adjacent to the 5’ untranslated region of the dopamine receptor D2 gene (DRD2)

DBPC=double-blind, placebo-controlled

POL=prospective, open-label

BDNF plays an important role in neuronal development and maturation, particularly for dopaminergic neurons.[43]

We identified one pediatric study which evaluated the link between a BDNF polymorphism and MPH efficacy. Kim et al.’s open-label study investigated the BDNF Val⁶⁶Met (rs6265, MAF = 0.20 [T]) polymorphism and found that valine/valine (Val/Val) homozygotes had a better MPH response (in terms of both symptom improvement and a response definition which required both improvement in symptoms and CGI ratings) than methionine (Met) carriers.[34] Several lines of evidence suggest the Met allele may be linked to adverse changes in brain anatomy and function. For example, decreased prefrontal cortex, temporal lobe, and occipital lobe volumes have been documented for Met carriers compared to Val/Val homozygotes.[44, 45] An additional study showed that the neurons of Val/Val homozygotes have increased activity-dependent BDNF secretion compared to Met/Met homozygotes.[46]

2.2.3. Catechol-O-methyltransferase (COMT) [Table 4].

Table 4.

Neurotransmitter Synthesis and Degradation Candidate Gene Studies of Methylphenidate Efficacy Published from 2009–2019

Gene	Polymorphism(s)	Authors	Design / Sample	Findings
Catechol-O-methyltransferase (COMT)	Val158Met	McGough et al 2009 [49]	DBPC N=82 U.S.A	No effect.
	Val158Met	Froehlich et al 2011 [33]	DBPC N=89 U.S.A.	Trend toward greater improvement on hyperactive-impulsive symptoms with increasing MPH dose for Val/Val homozygotes.
	Val158Met	Salatino-Oliveira et al 2011 [53]	POL N=251 Brazil	No independent effect on ADHD symptoms. Met allele was linked to improved MPH effects on oppositional symptoms over time.
	Val158Met	Yatsuga et al 2014 [52]	POL N=50 boys Japan	No effect.
	Val158Met	Park et al 2014a [50]	POL N=120 South Korea	Improved MPH response on hyperactive-impulsive symptoms with Val/Val homozygosity (not significant after multiple comparison correction).
	Val158Met	Unal et al 2016 [40]	POL N=108 Turkey	No effect.
	Val158Met rs4818	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effects.
	Val158Met rs2020917 rs933271 rs1544325 rs740603 rs740601 rs4646316 rs165774 rs9332377	Pagerols et al 2017 [51]	POL N=107 Spain	No effects.
DOPA Decarboxylase (DDC)	rs6592961	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effect.
Dopamine Beta Hydroxylase (DBH)	rs2007153 rs2797851 rs1548364 rs2797855 rs1541332 rs2519154 rs2797853 rs6479643 rs77905 rs2073833 rs1611131 rs129883	Pagerols et al 2017 [51]	POL N=107 Spain	MPH non-response nominally associated with rs2073833 C allele. Decreased MPH response with TC genotype at rs1541332–rs2073833 in whole sample analyses, with effects accentuated in children with prenatal smoke exposure.
Tyrosine Hydroxylase (TH)	rs10770140 rs6356 rs2070762	Pagerols et al 2017 [51]	POL N=107 Spain	No effect for rs10770140 or rs6356. Decreased MPH response with rs2070762 C/C homozygosity, with effects accentuated in children with prenatal smoke exposure.

DBPC=double-blind, placebo-controlled

POL=prospective, open-label

COMT degrades dopamine and norepinephrine in the synaptic cleft and plays an important role in the catabolism of dopamine in the prefrontal cortex due to the relative paucity of dopamine transporters in this region.[47]

A SNP at COMT’s codon 158 (Val158Met, rs4680, MAF=0.37 [A]) results in a substitution of methionine for valine which changes enzyme activity levels: Met/Met homozygotes exhibit four times lower enzymatic activity than those with wild type Val/Val homozygosity.[48] From 2009 onward, we identified two RCTs and six naturalistic studies which evaluated effects of the Val158Met polymorphism on MPH efficacy. Both RCTs evaluated genotype*dose effects: McGough et al. did not find significant COMT*MPH dose effects on ADHD symptom response,[49] while Froehlich et al. identified a trend (p=0.09) toward COMT*MPH dose effects suggesting improved response on hyperactive-impulsive symptoms with increasing dose for Val/Val homozygotes.[33] Park et al.’s open-label study also suggested a link between Val/Val homozygosity and an improved MPH response, but this finding was not significant after multiple comparison correction.[50] An additional five open-label studies reported null effects of the Val158Met genotype on MPH treatment of ADHD symptoms.[37, 40, 51–53]

Meta-analysis Findings.

Although studies from the past decade do not conclusively show a role for this variant in MPH response, aggregation of the available COMT PGx data in a recent meta-analysis by Myer et al.[41] does shed light on the impact of the Val158Met polymorphism. This meta-analysis, which included data from five of the studies detailed above,[33, 49, 50, 52, 53] found that the Val/Val genotype was associated with improved MPH treatment response compared with Met carriers (OR=1.4, 95%CI:1.04–1.87; p=0.02; I² =70%, may represent substantial heterogeneity in individual study findings).

2.2.4. Dopamine Beta Hydroxylase (DBH) [Table 4].

The enzyme DBH converts dopamine to norepinephrine.[54]

We identified only one study which evaluated the link between DBH variants and MPH response in children. This naturalistic study by Pagerols et al.[51] found a nominal association between DBH rs2073833 (MAF=0.43 [C]) and MPH treatment response that did not meet the Bonferroni-corrected threshold for statistical significance. However, a DBH two-marker haplotype was identified as a significant predictor of MPH response: the TC allelic combination of a haplotype involving rs1541332 (MAF=0.36 [G]) and rs2073833 was overrepresented among treatment-resistant patients (OR=3.43, 95%CI 1.59–7.4; p_adjusted=0.049), with risk for treatment failure further accentuated in children whose mothers smoked during pregnancy.[51] The functional significances of the rs1541332 and rs2073833 variants are currently unclear, although both SNPs are intronic.[55]

2.2.5. Dopamine Receptor D2 (DRD2) [Table 2].

DRD2 encodes the D2 dopamine autoreceptor, a feedback regulator critical for dopamine transmission and release, modulation of working memory, and regulation of the reward pathway.[56]

From 2009–2019, two naturalistic studies focusing on different polymorphisms assessed DRD2’s role in MPH efficacy. Gomez-Sanchez et al.[37] found an association between faster MPH response and the C allele of rs1800497 (MAF=0.33 [A]), which has previously been linked to increased DRD2 expression and availability.[57, 58] Pagerols et al. reported a trend for T allele carriers of rs2283265 (MAF=0.23 [T]) to require a higher MPH dose than G/G homozygotes (p=0.06), but did not find differences between the rs2283265 genotypes in terms of treatment response as assessed by CGI ratings.[51] One prior study of the intronic rs2283265 SNP found that T allele carriers expressed lower levels of the short isoform DRD2 (presynaptic) receptor and were more likely to have impairments on working memory and attentional tasks despite higher prefrontal cortex activity levels on functional magnetic resonance imaging (fMRI).[59]

2.2.6. Dopamine Receptor D3 (DRD3) [Table 2].

DRD3 receptors are involved in the modulation of dopamine synthesis and release, and are known to play a role in reward processing, addictive behaviors, incentive-based learning, and inhibition of motor responses.[60]

rs6280

The C allele of rs6280 (MAF=0.49 [C]) has been shown to confer higher dopamine binding affinity.[61] One placebo-controlled and one naturalistic MPH PGx study have investigated effects of rs6280 variants. In their RCT, Fageera et al. observed greater behavioral improvement with MPH treatment for rs6280 C allele homozygotes according to teacher but not parent ratings.[62] However, Pagerols et al.’s naturalistic study did not observe a significant impact of rs6280 on MPH efficacy.[51]

Additional Polymorphisms and Haplotypes

After correcting for multiple comparisons, Pagerols et al. did not find a significant main effect of nine other DRD3 SNPs (see Table 1) on MPH efficacy, although they did find that children with the GG haplotype at rs2134655-rs1800828 were more likely to be MPH non-responders.[51] Both rs2134655 (MAF=0.20 [T]) and rs1800828 (MAF=0.22 [G]) are intronic variants whose exact functions remain unknown.[55]

2.2.7. Dopamine Receptor D4 (DRD4) [Table 2].

DRD4 encodes dopamine receptor D4, a G-protein coupled receptor whose roles include inhibition of adenylyl cyclase.[63, 64] DRD4 has many important functions in the central nervous system, including regulation of corticostriatal neurotransmission by controlling glutamate receptor activity and by influencing dopamine synthesis and release.[60, 64]

Exon 3 VNTR

The 48 base pair DRD4 exon 3 VNTR has been widely evaluated in ADHD pharmacogenomics, with evidence suggesting the seven-repeat (long) allele confers decreased DRD4 activity.[65] From 2009–2019, three placebo-controlled and three naturalistic MPH PGx studies have focused on this polymorphism. The three RCTs produced varying results. McGough et al. failed to find an association between the exon 3 VNTR and MPH effects on ADHD symptoms.[49] Froehlich et al. observed that those lacking the four-repeat (short) allele among 89 American children had a decreased treatment response on hyperactive-impulsive symptoms.[33] Naumova et al. reported an association between the seven-repeat allele and improved treatment response according to parent but not teacher ratings.[66] All three naturalistic studies failed to identify an association between the exon 3 VNTR polymorphism and MPH efficacy.[35, 37, 67] However, Hong et al. identified an interaction between the DRD4 exon 3 VNTR and ADRA2A MspI genotype and MPH response.[35]

Meta-analysis Findings.

In a meta-analysis of six studies evaluating the 4-repeat VNTR (including four discussed above[33, 35, 49, 67]) and five studies evaluating the 7-repeat VNTR, Myer et al. documented a significant association between 4-repeat homozygosity and improved MPH response (OR=1.66, 95%CI:1.16–2.37, p=0.005; I² =39%, may represent moderate heterogeneity in study findings), and a trend toward a link between the 7-repeat allele and diminished MPH response (OR=0.68, 95%CI:0.47–1.00, p=0.05; I² =65%, may represent substantial heterogeneity in study findings).[41]

Promoter Region Polymorphism

One RCT and one naturalistic MPH PGx study have investigated the role of a 120-base pair repeat sequence located in DRD4’s promoter region which may be involved in regulating DRD4 gene expression.[68] Only Gomez-Sanchez et al.’s naturalistic study reported an association between short allele promoter duplication homozygosity and decreased MPH response,[37] while the RCT documented null effects on the outcome of ADHD symptomatic improvement.[49]

2.2.8. Dopamine Transporter (DAT/SLC6A3) [Table 5].

Table 5.

Neurotransmitter Reuptake and Release Candidate Gene Studies of Methylphenidate Efficacy Published from 2009–2019

Gene	Polymorphism(s)	Authors	Design / Sample	Findings
Dopamine Transporter (DAT/SLC6A3)	3’UTR VNTR	McGough et al 2009 [49]	DBPC N=82 U.S.A	No effect.
	3’UTR VNTR	Froehlich et al 2011 [33]	DBPC N=89 U.S.A.	Diminished MPH response on hyperactive-impulsive symptoms with increasing MPH dose for 10-repeat carriers.
	3’UTR VNTR	Hong et al 2012 [35]	POL N=103 South Korea	No independent effect. Trend toward an interaction with NET −3081 (A/T).
	3’UTR VNTR	Stein et al 2014 [71]	DBPC N=56 U.S.A	Diminished MPH response over time for 9-repeat homozygotes.
	3’UTR VNTR Intron 8 VNTR rs2550948 rs2652511 rs11564750	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	Improved MPH response with rs2550948 G allele. Faster MPH response over time with intron 8 VNTR 6/− or −/− genotypes. No effects for other variants.
	rs460700 rs37020 rs13161905 rs27048 rs6347 rs11133767 rs40184	Pagerols et al 2017 [51]	POL N=107 Spain	No effects.
Norepinephrine Transporter (NET/SLC6A2)	G1287A	McGough et al 2009 [49]	DBPC N=82 U.S.A	No effect.
	G1287A −3081(A/T)	Kim et al 2010 [91]	POL N=112 South Korea	Trend toward improved MPH response for T allele of −3081(A/T). No effect for G1287A.
	G1287A −3081(A/T)	Kim et al 2011 [34] Hong et al 2012 [35]	POL N=102 for Kim et al N=103 for Hong et al South Korea	Kim et al: No effects. Hong et al: No independent effects. NET G1287A*ADRA2A DraI interaction effect (improved response with G allele of G1287A). NET −3081(A/T)*ADRA2A DraI interaction effect (improved response with T allele of −3081(A/T)).
	G1287A −3081(A/T) rs5568 rs998424 rs1616905 rs2242446 rs2242446 and rs1610905 haplotype rs5569 and rs998424 haplotype	Lee et al 2011 [92]	POL N=137 South Korea	No single marker or haplotype effects.
	G1287A	Song et al 2011 [94]	POL N=114 South Korea	Improved MPH response with G1287A G/G homozygosity.
	G1287A −3081(A/T)	Park et al 2012 [95]	POL N=37 South Korea	Improved MPH response with G1287A G/G homozygosity based on clinician but not parent ratings. No effect for −3081(A/T).
	30 tag SNPs G1287A	Thakur et al 2012 [97]	DBPC N=475 Canada	Improved MPH response with rs36021 T allele in children with prenatal smoke exposure. Improved MPH response with rs3785152 C allele in children who lacked prenatal smoke exposure.
	rs192303 rs3785143	Song et al 2014 [76]	POL N=139 South Korea	Improved MPH response with rs192303 C/C homozygosity (not significant after multiple comparison correction). No effect for rs3785143.
	G1287A −3081(A/T)	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effects.
	G1287A −3081(A/T) rs2242446 rs3785143 rs3785157 rs7194256	Angyal et al 2018 [93]	POL N=122 Hungary	Improved MPH response on hyperactive-impulsive symptoms for −3081(A/T) T allele carriers. No independent effects for other variants. Improved MPH response on inattention symptoms for the TCT haplotype of −3081(A/T), rs2242446, and rs3785143 (compared to the ATC haplotype).
Serotonin Transporter (SLC6A4/5-HTT)	5HTTLPR Intron 2 VNTR	McGough et al 2009 [49]	DBPC N=82 U.S.A	Decreased MPH response with Intron 2 VNTR −12/−12 homozygosity. No effect for 5HTTLPR.
	5HTTLPR	Thakur et al 2010 [105]	DBPC N=157 Canada	Improved MPH response with long allele homozygosity.
	5HTTLPR	Park et al 2015 [106]	POL N=114 South Korea	No effect.
	5HTTLPR Intron 2 VNTR	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effects.
Solute Carrier Family 9 (SLC9A9)	rs9810857	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	No effect.
Synaptosomal-Associated Protein 25 (SNAP25)	T1065G T1069C	McGough et al 2009 [49]	DBPC N=82 U.S.A	No effects.
	T1065G	Song et al 2014 [76]	POL N=139 South Korea	Improved MPH response with T allele (not significant after multiple comparison adjustment).
	T1065G	Gomez-Sanchez et al 2017 [37]	POL N=208 Spain	Improved MPH response with G allele (no adjustment for multiple comparisons).

DBPC=double-blind, placebo-controlled

POL=prospective, open-label

SNP=single nucleotide polymorphism

VNTR=Variable nucleotide tandem repeat

DAT encodes a presynaptic protein responsible for reuptake of synaptic dopamine. One of MPH’s main mechanisms of action is to increase the concentration of dopamine in the synapse by inhibiting DAT.[69] Early ADHD PGx studies focused on DAT,[18] making it among the most well-studied candidate genes, although the rate of investigation has slowed over the past decade.

3’ UTR VNTR

A VNTR in DAT’s 3’ UTR has been linked to functional differences, as neuroimaging studies have demonstrated that individuals with the 10-repeat allele exhibit ~50% greater DAT densities compared to other genotypes.[70] From 2009–2019, three RCTs and two naturalistic studies have evaluated the effects of DAT 3’ UTR VNTR variants on MPH efficacy. The RCTs found inconsistent results: Froehlich et al.[33] observed a greater reduction in hyperactive-impulsive symptoms with increasing MPH dose for children who were not 10-repeat carriers, Stein et al.[71] found an improved MPH response in those who were 10-repeat carriers, and McGough et al.[49] did not observe a significant impact of variation in the 3’ UTR VNTR on MPH response. On the other hand, the prospective open-label MPH PGx studies were uniform in their documentation of null effects for the DAT 3’ UTR VNTR.[35, 37]

Meta-analyses Findings.

During the past decade, three meta-analyses have examined the impact of DAT 3’ UTR VNTR polymorphisms on MPH efficacy. Kambeitz et al.[72] aggregated data from 13 pediatric (none of which appear above due to their publication prior to 2009) and three adult studies, and reported no effect on ADHD total symptom scores or the specific dimensions of hyperactivity-impulsivity or inattention. Effects remained non-significant when analyses were limited to pediatric studies. However, when analyses were limited to naturalistic studies, 10-repeat homozygosity was associated with a poor response to MPH (d=−0.63, 95%CI:−1.1 to −0.15, p=0.009, I²=91%, indicating considerable heterogeneity in study findings).[72] Echoing Kambeitz et al.’s findings for naturalistic studies, Myer et al.[41] analyzed data from 16 pediatric studies (three of which were reviewed above[33, 35, 49]) and documented reduced MPH efficacy for 10-repeat homozygotes (OR=0.74, 95%CI:0.60–0.90, p=0.004; I² =68%, may represent substantial heterogeneity in study findings). Similarly, Soleimani et al.[73] linked improved MPH treatment response with lack of 10-repeat homozygosity (10/R/10R) in a pooled analysis of eight naturalistic pediatric studies (pooled Cohen’s d=0.44; 95%CI: 0.12–0.75, p<0.01, I² statistic not available), all of which were published prior to 2009, but effects of the 10R/10R genotype were non-significant in pooled analyses of seven clinical trials, including one reviewed above.[35]

Additional Polymorphisms

A recent study by Gomez-Sanchez et al.[37] investigated additional DAT loci (see Table 1). They observed only two significant associations: a link between improved MPH response and the G allele of rs2550948 (MAF=0.33 [T]) and an association between faster MPH response over time and lack of 6-repeat homozygosity at the intron 8 VNTR. The functional significance of the intronic rs2550948 SNP is currently unknown, while intron 8 VNTR increased repeat number has been correlated with increased DAT expression.[74]

2.2.9. Latrophilin 3/Adhesion G Protein-coupled Receptor L3 (LPHN3/ADGRL3) [Table 3].

LPHN3 encodes a brain-specific G-protein coupled receptor (Adhesion G protein-coupled Receptor L3) that is important for synaptic function and neurotransmitter exocytosis.[75]

rs1947274

From 2009–2019, four MPH PGx studies evaluated the rs1947274 (MAF=0.40 [C]) intronic SNP and reported inconsistent results. Two naturalistic studies by Song et al.[76] and Bruxel et al.[77] found no independent effects of rs1947274 variants on MPH efficacy. Song et al.[76] also reported that a haplotype block containing rs1947274 and rs6551665 (MAF=0.40 [G]) had no effect on treatment response. In contrast, in their RCT, Labbe et al. found that the rs1947274 C allele predicted MPH non-response, as assessed using an outcome measure combining direct observation of child behavior and actigraphy data.[78] On the other hand, an RCT by Choudhry et al. observed no significant main effects of rs1947274 variants; however, a haplotype analysis involving rs6551665, rs1947274, and rs6858066 (MAF=0.50 [G]) found that the GCA haplotype was associated with improved response to MPH (as rated on the CGI), but only in children whose mothers did not experience stress during pregnancy.[79] Of note, effect modification by prenatal stress may help to explain why Labbe et al. (who did not test for this interaction) and Choudhry et al. (who did) found that different rs1947274 alleles portended poor MPH response.

Meta-analysis Findings.

When Myer et al. performed a meta-analysis which included three of the studies discussed above[76, 78, 79], they did not observe a significant impact of rs1947274 variants on MPH response (OR=0.95, 95%CI:0.71–1.26, p= 0.70; I² =81%, indicating considerable heterogeneity in study findings).[41]

rs6551665

The six MPH PGx studies evaluating the rs6551665 (MAF=0.40 [G]) intronic SNP documented mixed effects. Three naturalistic studies by Song et al.[76], Bruxel et al.[77], and Gomez-Sanchez et al.[37] reported null effects. However, Labbe et al.’s RCT linked the rs6551665 G allele to MPH non-response,[78] while a naturalistic study by Arcos-Burgos et al.[75] and an RCT by Choudhry et al.[79] observed improved MPH response with the G allele in single marker analyses. As above, Choudhry et al. also found a link between improved MPH response and the rs6551665-rs1947274-rs6858066 GCA haplotype, but only in children whose mothers did not experience prenatal stress.[79] Furthermore, in addition to the independent effects reported above, Arcos-Burgos evaluated an LPHN3 haplotype block containing rs6551665, rs1947275, and rs9683662 (MAF=0.18 [T]) and found that the haplotype containing the rs6551665 G allele conferred better treatment response. [75] An analysis by Jain et al.[80] which utilized a subset of Arcos-Burgos et al.’s[75] sample investigated the interaction between LPHN3’s rs6551665 and a haplotype block on chromosome 11q containing all or part of the genes encoding neural cell adhesion molecule 1 (NCAM1), the tetratricopeptide repeat domain 12 (TTC12), the ankyrin repeat and kinase domain containing 1 (ANKK1), and situated adjacent to the 5’ UTR of DRD2. They found that being a G/G homozygote at rs6551665 and having two copies of the 11q ADHD susceptibility haplotype to be correlated with good response to stimulant medication treatment.[80]

Meta-analysis Findings.

In Myer et al.’s meta-analysis, which analyzed data from four of studies detailed above[76–79], effects of the rs6551665 variant on MPH response were non-significant (OR=1.07, 95%CI:0.84–1.37, p=0.59; I² =77%, indicating considerable heterogeneity in study findings).[41]

rs1868790

Effects of rs1868790 (MAF=0.48 [A]) on MPH efficacy were investigated in two RCT and one naturalistic study: two reported null effects[37, 78], while Choudhry et al.’s RCT linked the T allele to improved MPH response.[79]

rs2345039

Effects of rs2345039 (MAF=0.38 [C]) on MPH efficacy were investigated in three studies. In their RCT, Labbe et al.[78] found that the G allele predicted MPH non-response, while a naturalistic study by Song et al.[76] and an RCT by Choudhry et al.[79] reported null effects.

rs6858066

Three studies evaluated the impact of rs6858066 (MAF=0.50 [G]) variants on MPH efficacy. Labbe et al.’s RCT linked the rs6858066 A allele to diminished MPH response,[78] while Gomez-Sanchez et al.’s naturalistic study[37] and Choudhry et al.’s RCT[79] found no significant main effects for rs6858066. However, Choudhry et al. reported a link between improved MPH response and the rs6551665-rs1947274-rs6858066 GCA haplotype in children whose mothers did not experience prenatal stress.[79] Of note, Choudhry et al.’s haplotype findings and Labbe et al.’s main effect findings suggest opposite effects of the A allele on MPH efficacy.

Additional Polymorphisms and Haplotypes

Although additional LPHN3 SNPs (see Table 1) have been investigated in one MPH PGx study each, none had significant main effects on MPH efficacy. However, a haplotype analysis by Bruxel et al. documented a faster MPH treatment response for CGC homozygotes of a haplotype involving rs6813183 (MAF=0.39 [G]), rs1355368 (MAF=0.48 [A]), and rs734644 (MAF=0.32 [T]).[77]

2.2.10. Metabotropic Glutamate Receptor Subtype 7 (mGluR7/GRM7) [Table 2].

The glutamate receptor subtype 7 (mGluR7/GRM7) is activated by the excitatory neurotransmitter L-glutamate and is an important presynaptic neurotransmission regulator.[81]

rs3792452

Since 2009, rs3792452’s (MAF=0.14 [A]) association with MPH efficacy has been evaluated in two open-label studies. Park et al. found that children with the G/A genotype had a better response to MPH than G/G homozygotes,[82] while Gomez-Sanchez et al. reported null effects.[37]

Genome-wide Association Study (GWAS) Findings and Integration with the Prior Literature.

Consistent with Park et al.’s findings, a 2008 GWAS by Mick et al. reported an association between rs3792452’s G allele and MPH treatment response.[83] A 2018 GWAS by Pagerols et al. did not replicate this rs3792452 finding, although they did identify a nominal association between GRM7 rs17047590 (MAF=0.16 [T]) and treatment response.[84] The functional significance has not been elucidated for either rs3792452 or rs17047590, which are both intronic.[85]

2.2.11. Norepinephrine Transporter (NET/SLC6A2) [Table 5].

NET plays a role in presynaptic reuptake of both norepinephrine and dopamine.[86, 87] It is thought to be more important for dopamine reuptake in the prefrontal cortex than the dopamine transporter due to the low density of dopamine transporters in this region and dopamine’s higher affinity for the NET compared to DAT.[88] Inhibition of NET is a key MPH mechanism of action.[89]

G1287A (rs5569)

The most well-studied NET variant in MPH PGx investigations is G1287A (rs5569, MAF=0.23 [A]). While it is routinely regarded as a silent polymorphism, G allele homozygosity has been associated with elevated levels of 3-methoxy-4-hydroxyphenylglycol, the major brain metabolite of norepinephrine.[90] From 2009–2019, G1287A was assessed in one RCT and seven prospective open-label cohorts. In their RCT, McGough did not observe significant effects of G1287A variants on MPH efficacy.[49] The open-label G1287A MPH studies also had null findings by and large, with five studies finding no significant effects,[34, 37, 91–93] and only three[35, 94, 95] suggesting a role for this SNP. Both Song et al. and Park et al. documented an improved MPH response for G1287A G/G homozygotes.[94, 95] Hong et al. (who examined the same cohort as Kim et al.[34]) did not observe significant independent effects of G1287A variants, but did document significant NET G1287A*ADRA2A DraI interaction effects on MPH response, such that having the G1287 G allele portended an improved MPH response.[35]

Meta-analyses Findings:

Although the results of individual studies failed to provide conclusive evidence of G1287A’s impact on MPH efficacy, aggregation of data from seven studies (including five discussed above)[35, 49, 91, 92, 94] in Myer et al.’s meta-analysis[41] provides evidence of a link between G/G homozygosity and improved MPH response (OR versus A carriers=1.73, 95%CI:1.26–2.37, p=0.0007; I²=0%, indicating little heterogeneity in study findings). On the other hand, a meta-analysis by Angyal et al.[93] which combined data from five studies[35, 91–94] did not show significant effects of the G1287A polymorphism on MPH response (MPH positive response OR for A carriers versus G homozygotes=0.92, 95%CI:0.51–1.74, p=0.76; heterogeneity statistic not provided) despite the substantial overlap between studies included in their and Myer et al.’s[41] meta-analyses.

−3081(A/T) (rs28386840)

An additional NET polymorphism of particular interest is the promoter region −3081 A to T SNP (rs28386840, MAF=0.39 [T]) whose T-allele has been linked to lower NET expression.[96] From 2009–2019, this polymorphism was investigated in six open-label cohorts. Although four of these studies failed to document significant effects of the −3081(A/T) SNP,[34, 37, 92, 95] Kim et al.[91] reported a trend toward improved MPH response with the T allele and Angyal et al.[93] found improved MPH effects on hyperactive-impulsive symptoms for T allele carriers. While Hong et al.[35] (who evaluated the same sample as Kim et al. [34]) did not observe independent effects of −3081(A/T) variants, they did find significant NET −3081(A/T)*ADRA2A DraI interaction effects on MPH response such that having the T allele of −3081(A/T) predicted improved MPH effects.

Meta-analyses Findings:

Two meta-analyses evaluated effects of the −3081(A/T) SNP on MPH efficacy. Myer et al.[41] combined data from three studies, including two which were reviewed above[35, 91] and documented improved MPH response for T allele carriers (OR versus A/A homozygotes=2.93, 95%CI:1.76–4.90, p<0.0001; I²=8%, indicating little heterogeneity in study findings). An additional meta-analysis by Angyal et al.[93], which aggregated data from the same studies as Myer et al. also showed a link between the T allele and improved MPH response (OR versus A/A homozygotes=2.70, 95%CI:1.27–5.73, p=0.01; heterogeneity statistic not provided).

Additional Polymorphisms and Haplotypes.

Additional NET polymorphisms and haplotypes (see Table 1) have been evaluated in one RCT[97] and three open-label studies[76, 92, 93]. Angyal et al.[93] observed improved MPH effects on inattention symptoms for the TCT haplotype of −3081(A/T), rs2242446 (MAF=0.25 [C]), and rs3785143 (MAF=0.10 [T]) (compared to that of the most frequent haplotype [ATC]). In an RCT, Thakur et al. found the effects of two NET SNPs were moderated by exposure to prenatal smoking: the T allele of rs36021 (MAF=0.40 [T]) was linked to improved MPH response in children whose mothers smoked during pregnancy, while the C allele of rs3785152 (MAF=0.15 [T]) was associated with improved MPH response in children whose mothers did not smoke during pregnancy.[97] The functional significance of the intronic SNPs rs36021 and rs3785152 has not been fully elucidated.[97]

2.2.12. N-methyl-D-aspartate (NMDA) Receptor Subunit 2B (GRIN2B) [Table 2].

GRIN2B encodes subunit 2B of a glutamate receptor known as the NMDA receptor.[98] Since the glutamatergic system is regulated by the dopamine and norepinephrine systems and inhibition of NMDA receptors has been shown to interfere with MPH response in animal models, it is plausible that GRIN2B polymorphisms may impact ADHD pharmacological response.[99]

We found one study which investigated the link between variants in rs2284411 (MAF=0.28 [T]), an intronic GRIN2B SNP [100] with unknown functional significance, and MPH efficacy. In an open-label trial, Kim et al. found that rs2284411 C/C homozygotes showed improved treatment effects compared to T carriers in terms of inattentive symptoms scores and overall CGI ratings, while effects on hyperactive and total symptom scores were nominally significant.[101]

2.2.13. Serotonin Transporter (SLC6A4/5-HTT) [Table 5].

SLC6A4 encodes the serotonin reuptake transporter 5-HTT. Previous evidence has linked 5-HTT dysfunction with hyperactivity and inattentiveness,[102] indicating that it may play a role in progression of ADHD and ADHD treatment response.

5HTTLPR

Over the past decade, two MPH PGx RCT and two prospective, open-label studies investigated effects of a 44 base pair insertion/deletion region in SLC6A4’s promoter region (5HTTLPR) which results in long and short variant alleles. The 5HTTLPR short allele is associated with reduced SLC6A4 expression and serotonin reuptake, while the long allele is linked to higher 5-HTT expression and more rapid serotonin uptake.[103, 104] In their RCT, Thakur et al. reported a link between long-allele homozygosity and improved MPH response.[105] McGough et al.’s RCT did not find an effect of 5HTTLPR variants on MPH response in terms of ADHD symptomatic improvement, although long-allele carriers were significantly less responsive to higher doses of MPH on a math accuracy measure.[49] Neither open-label trial identified significant effects of 5HTTLPR variants on MPH efficacy.[37, 106]

Integration with the prior literature.

Similar to findings over the past decade, the literature prior to 2009 contains scant evidence that 5HTTLPR variants impact MPH efficacy: two prior naturalistic studies found no association between 5HTTLPR variants and MPH response[107, 108], while a naturalistic study by Seeger et al. found a decreased response to MPH for children who were both 5HTTLPR long-allele homozygotes and DRD4 7-repeat carriers.[109]

Intron 2 VNTR

One placebo-controlled and one open-label study evaluated effects of a VNTR in SLC6A4’s intron 2. McGough et al.’s RCT documented a decreased MPH response (in terms of symptomatic improvement) for −12/−12 homozygotes at SLC6A4 intron 2,[49] while effects of this VNTR on MPH response were not observed in Gomez-Sanchez et al.’s naturalistic study.[37]

2.2.14. Synaptosomal-associated protein 25 (SNAP25) [Table 5].

SNAP-25 is a highly localized presynaptic neuronal protein that plays a key role in exocytosis of catecholamine-containing vesicles.[110]

T1065G (rs3746544)

During the 2009–2019 period, we identified one RCT and two naturalistic studies evaluating effects of T1065G (rs3746544, MAF=0.28 [G]) variants on MPH efficacy. McGough et al.’s RCT[49] found no association between T1065G variants and MPH response,[49] while the open-label studies documented opposing effects. Song et al. found a link between the T allele of T1065G and improved MPH response (on an outcome which combined CGI and ADHD symptom ratings) that was no longer significant when adjusted for multiple comparisons.[76] In contrast, Gomez-Sanchez et al. reported that carriers for the G allele of T1065G showed a better MPH response than T/T homozygotes, although this finding was not adjusted for multiple comparisons.[37]

Integration with the Prior Literature:

Only one prior study by McGough et al.[111] investigated effects of T1065G variants on MPH response: they found that the T allele of T1065G was linked to better MPH dose response. Given the sparsity of the prior literature and recent null or mixed effects, there does not appear to be conclusive evidence that T1065G variants impact MPH efficacy.

2.2.15. Tyrosine Hydroxylase (TH) [Table 4].

TH is the rate-limiting enzyme in the synthesis of catecholamines, including dopamine, noradrenaline and adrenaline, as it converts tyrosine into l-dopa.[112, 113]

During the past decade, one naturalistic pediatric study by Pagerols et al. investigated effects of three TH variants on MPH efficacy (see Table 1).[51] They reported that C/C homozygotes for rs2070762 (MAF=0.42 [C]) were less responsive to MPH than T allele carriers in whole sample analyses, and that this adverse impact of C/C homozygosity on MPH response was accentuated in children who had in utero tobacco exposure.[51] In vitro assays indicate that the rs2070762 C allele might be a functional enhancer element regulating TH gene expression.[114]

2.3. Methylphenidate Genome-Wide Approaches

One pediatric MPH PGx GWAS was identified by our 2009–2019 literature search. This GWAS did not identify any variants which had a significant association with MPH efficacy (as assessed via CGI ratings) after multiple comparison correction in a discovery cohort of 173 Spanish children.[84] However, a meta-analysis of this pediatric cohort and a second cohort of 189 adults utilized a polygenic risk score model and found a significant association between MPH response and variants in the PEBP4 (rs17685420, MAF=0.16 [T]) gene, as well as suggestive effects of markers within the ALDH1L1 (rs2886059, MAF=0.24 [A]), CDH23 (rs17712523, MAF=0.13 [A]), and ARSA (rs2071421, MAF=0.22 [C]) genes.[84] PEBP4, a serine protease important for cellular development, has been implicated in Alzheimer’s disease.[115] Both ALDH1L1, which encodes an aldehyde dehydrogenase, and CDH23, which encodes an important cell adhesion molecule, have previously been associated with ADHD.[116, 117] Additionally, deficiencies in ARSA, a sulfatase, have been associated with poor school performance and cerebellar ataxia.[118]

3. Amphetamine Pharmacogenomic Studies

Like MPH, amphetamines act by blocking reuptake of dopamine and norepinephrine through dopamine transporter and NET inhibition.[21] In addition, amphetamines increase synaptic dopamine by inhibiting the vesicular monoamine transporter 2, which releases dopamine from vesicular storage, as well as by reverse transport at DAT.[21, 119]. The hepatic enzyme cytochrome P450 2D6 (CYP2D6), which also breaks down ATX, has been implicated in amphetamine metabolism.[120] However, our literature search did not identify any amphetamine PGx studies investigating the link between ADHD symptom response after amphetamine treatment and variants in pharmacokinetic genes.

3.1. Amphetamine Pharmacodynamic Genetic Variants

3.1.1. Dopamine Transporter (DAT/SLC6A3).

We identified only one PGx study from 2009–2019 that evaluated pharmacodynamics genetic predictors of amphetamine effects on ADHD symptoms. This RCT by Stein et al.[71] investigated whether polymorphisms in DAT’s 3’ UTR VNTR moderated the dose-response effects of both dexmethylphenidate (d-MPH) and mixed amphetamine salts (MAS). They found a reduced response to both medications for 9-repeat homozygotes: 10–20 mg doses of either d-MPH or MAS had little to no effect on hyperactivity-impulsivity or ADHD total symptom scores in children with the 9/9 genotype, but low to moderate stimulant doses did lead to improvement in 10-repeat carriers.

4. Atomoxetine Pharmacogenomic Studies

ATX was the first non-stimulant medication approved for use in ADHD. ATX acts to increase availability of synaptic norepinephrine through selective inhibition of NET.[121, 122] In addition, since NET is thought to play a large role in prefrontal cortex dopamine reuptake given the low density of DAT in this region,[88] ATX blockade of NET may also increase synaptic dopamine in the prefrontal cortex .

4.1. Atomoxetine Pharmacokinetic Genetic Variants

4.1.1. Cytochrome P450 2D6 (CYP2D6) [Table 6].

Table 6.

Candidate Gene Studies of Atomoxetine Efficacy Published from 2009–2019

Gene	Polymorphism(s)	Authors	Design / Sample	Findings
Adrenergic α1A–Receptor (ADRA1A)	rs17426222 rs573514 rs3808585	Yang et al 2013 [132]	POL N=111 China	No effects.
Adrenergic α2A–Receptor (ADRA2A)	MspI DraI	Yang et al 2013 [132]	POL N=111 China	Non-remission status linked to GG haplotype of MspI and DraI (not significant after correction for multiple comparisons).
Cytochrome P-450 2D6 (CYP2D6)	*3, *4, *5, *6, *7, *8 alleles1	Ramoz et al 2009 [126]	Included two cohorts: Multinational POL N=160 U.S.A. DB crossover with methylphenidate N=105	No effects.
Dopamine Beta Hydroxylase (DBH)	rs1076150 rs1611115 rs1108580 rs2873804 rs1548364 rs2519154 rs2073837 rs129882	Fang et al 2015 [130]	POL N=87 China	Decreased ATX response with C allele of rs2519154. No effects for other SNPs or for haplotypes from 2 linkage disequilibrium blocks.
Norepinephrine Transporter (NET/SLC6A2)	108 SNPs	Ramoz et al 2009 [126]	Included two cohorts: Multinational POL N=160 U.S.A. DB crossover with methylphenidate N=105	No effects for any SNP after multiple comparison correction. ATX response linked to a linkage disequilibrium (LD) block covering exons 4–9, including rs3785143.
	rs3785143 rs3785152 rs2279805 rs5569 rs36009 rs2242447	Yang et al 2013 [132]	POL N=111 China	ATX non-response linked to rs3785143 T allele. No effects for other SNPs after multiple comparison correction.

¹Participants who were homozygous or heterozygous for alleles *3, *4, *5, *6, *7, and/or *8 were considered poor metabolizers.

ATX=atomoxetine

DB=double-blind

POL=prospective, open-label

SNP=single nucleotide polymorphism

ATX catabolism is primarily mediated by the liver enzyme CYP2D6[123], which converts ATX to 4-hydroxyatomoxetine, an active metabolite.[124] CYP2D6 is a highly polymorphic enzyme; the link between several functional CYP2D6 polymorphisms and altered metabolizer status has been well-documented. Poor metabolizers (PM) possess one or more mutations in CYP2D6 conferring reduced enzyme activity while extensive metabolizers (EM) exhibit normal/wild type enzyme activity.[125]

Since 2009, the impact of CYP2D6 variants on ATX efficacy was investigated in two samples (one from the U.S. and one multinational) via an RCT design. Findings from both samples were reported by Ramoz et al.[126] They did not observe differences between individuals with CYP2D6 PM versus EM genotypes in ATX efficacy in either cohort, when analyses combined both cohorts, or when analyses were limited to Caucasian participants.

Integration with the Prior Literature:

We previously reported[18] on a 2007 pooled analysis conducted by Michelson et al.[127] which found that CYP2D6 PMs had greater overall symptom reduction scores compared to EMs. Of note, half of the Ramoz et al. cohort [126] was included in the Michelson et al. pooled analysis; therefore, the lack of effect for CYP2D6 metabolizer status observed by Ramoz et al. may be due to small sample size and attendant loss of power. Indeed, in a separate pooled analysis of 1326 children conducted by Trzepacz et al. in 2008,[128] CYP2D6 PMs were found to have significantly greater improvement in inattention symptom scores than EMs. Ultimately, these efficacy findings, combined with evidence that PMs experience increased adverse effects (including greater changes in heart rate and blood pressure) compared to EMs[127, 128] as well repeated demonstration of CYP2D6 genotype effects on ATX pharmacokinetics (reviewed in Brown et al.’s Supplement[129] and by Yu et al.[125]), were deemed strong enough to form the basis for a newly issued CPIC guideline providing therapeutic recommendations for ATX based on CYP2D6 genotype.[129]

4.2. Atomxetine Pharmacodynamic Genetic Variants

Since ATX selectively impacts NET, several studies have explored the effect of variability in noradrenergic-related genes on ATX response in children with ADHD.

4.2.1. Dopamine Beta Hydroxylase (DBH) [Table 6].

DBH plays a key role in the synthesis of norepinephrine from dopamine.[54]

Only one ATX PGx efficacy study[130] during the 2009–2019 period evaluated DBH variants. In their naturalistic study, Fang et al. investigated the link between ATX response and eight DBH SNPs as well as haplotypes from two linkage disequilibrium blocks. Although nominal associations with ATX efficacy were found for multiple DBH SNPs and haplotypes, only one SNP, rs2519154 (MAF=0.43 [T]), was significantly linked to ATX response after multiple comparison correction, with the C allele being associated with non-responder status. Rs2519154 is intronic, and is thought to play a role in DBH gene regulation.[130]

4.2.2. Norepinephrine Transporter (NET/SLC6A2) [Table 6].

Given that ATX’s mechanism of action involves selective inhibition of NET, which leads to increased synaptic norepinephrine in multiple brain regions as well as increased synaptic dopamine in the prefrontal cortex,[131] it is not surprising that ATX PGx studies have also focused on NET variants.

From 2009–2019, two naturalistic studies investigated the link between NET variants and ATX efficacy. Ramoz et al.[126] investigated the association between NET polymorphisms and ATX response in two independent cohorts (one from the U.S.A and one multinational). Of 108 sequenced NET polymorphisms, 20 SNPs were nominally associated with ATX response in at least one of the two cohorts, although none were statistically significant after multiple comparison correction. However, Ramoz et al. did identify a linkage disequilibrium (LD) block covering NET’s exons 4–9 (where 36 SNPs were genotyped) which was significantly associated with treatment response such that a carrier of this allele sequence would have twice the chance of responding to ATX compared to a non-carrier (OR=1.83, p<0.01 in the combined cohort). Yang et al.[132] investigated the link between six NET SNPs and ATX efficacy. They only found one SNP to be linked to ATX efficacy after correction for multiple comparisons: the rs3785143 T allele was significantly associated with ATX non-responder status and the C allele with being an ATX responder. Yang et al. considered their results to be consistent with prior findings: when Ramoz et al.[126] evaluated the LD block covering NET’s exons 4–9, they found that the haplotype containing the rs3785143 C allele was associated with ATX responder status. Although the exact function of rs3785143 has not been fully elucidated, it is located in intron 1, a region important for NET transcriptional regulation.[133]

5. Alpha-agonist Pharmacogenomic Studies

In recent years, the non-stimulant medications guanfacine and clonidine have been FDA-approved for the treatment of ADHD. Guanfacine and clonidine both selectively stimulate α2-adrenergic receptors in the prefrontal cortex, modulating norepinephrine activity and decreasing impulsivity and hyperactivity.[134] Guanfacine is metabolized by CYP3A4.[134] While clonidine metabolism is not well-understood, evidence indicates that its breakdown is primarily mediated by CYP2D6 and to a lesser extent by CYP3A4/5 and CYP1A1/2.[135] Unfortunately, our search of the 2009–2019 literature did not identify any guanfacine or clonidine PGx pharmacodynamic or pharmacokinetic studies.

6. Current Challenges and Future Directions

The past decade of pediatric ADHD medication efficacy PGx research has been marked by further accumulation of candidate gene studies focusing mostly on associations with MPH response. A lesser number of ADHD PGx studies have addressed ATX and little to none have evaluated amphetamine, guanfacine, and clonidine. Unfortunately, even when studies of a given genetic variant and ADHD medication efficacy are available, our understanding of their findings is complicated by numerous factors.

Specifically, interpretation of ADHD PGx study results and comparison of findings across studies continues to be constrained considerably by heterogeneity and limitations in study design.[18] For example, many studies are open-label: their inability to account for placebo effects makes it difficult to identify true medication response and its association with genetic variability. Indeed, Polanczyk et al. found that study design (open studies versus RCTs) was associated with heterogeneity of study results in ADHD pharmacogenetics studies.[136] In addition, variability in dosing regimens also presents a barrier to interpretation, as the use of only lower doses in some studies may bias against finding significant treatment effects. Current studies are also limited by their outcome measures, as dichotomous measures (e.g., medication responder versus non-responder status) have less power to detect effects than quantitative measures. The most commonly used measures, including ADHD symptom rating scales and Clinician Global Impression ratings, are also subjective, and relationships between genetic variants and medication response have sometimes varied in the same sample depending on who completed the ratings (e.g., parents versus teachers).[66] In addition, ADHD PGx studies have focused on predictors of short-term effects, with few assessing long-term outcomes, even though responder status to the same medication and dose has been shown to change over time.[37, 49]

The heterogeneity of ADHD PGx study samples--in terms of racial/ethnic and clinical characteristics—also presents a barrier to understanding study findings. Genetic factors predicting medication response may vary by racial/ethnic group, suggesting that the polymorphisms being studied may not be themselves influencing medication response but may be in linkage disequilibrium with the actual functional genetic variants.[137] Unique risk and protective factors which influence medication efficacy may also be present in different ethnic groups.[41] Given prior evidence that some polymorphisms may influence MPH response on inattentive or hyperactive-impulsive symptoms but not both domains[33, 101], the ADHD presentation (e.g., inattentive, hyperactive-impulsive, or combined) profile of the sample may impact PGx study findings.[36] Differences in the comorbidity profile of study samples may also influence study results: children with varied coexisting conditions may have unique underlying neurophysiologies such that different genetic variants may be operative in ADHD medication response. In addition, there is emerging evidence that the effect of a given genetic variant on medication response may be moderated by other genetic polymorphisms,[35, 37] yet few ADHD PGx investigations have considered these interactions. Care must be taken, however, to ensure that studies of interactions or specific subgroups are independent and hypothesis-driven, rather than being conducted for the sake of documenting “statistically significant” findings obtained by chance after performing multiple non-adjusted comparisons.

ADHD PGx studies have also been limited by their generally small sample sizes, which have limited power to detect modest genetic effects. Therefore, the ADHD PGx knowledge base has been considerably advanced by recent meta-analyses[41, 72, 93] which aggregated data from multiple studies to identify a significant impact on MPH efficacy for variants in DAT, DRD4, ADRA2A, COMT, and NET. However, in discussing their meta-analysis findings, Myer et al. note that the OR for each variant is low; therefore, knowing an individual’s genotype for any one variant is unlikely to reliably predict ADHD treatment choice.[41] Rather, multivariable models that incorporate the impact of multiple genes and/or investigation of polygenic risk scores may one day yield clinically actionable MPH efficacy information. Furthermore, to more fully capture the inputs which drive family choice of treatment, ADHD PGx studies and multivariable prediction models need to incorporate not just measures of medication efficacy, but also side effects. Of note, there is a burgeoning ADHD PGx adverse effects literature[138], although its review is beyond the scope of the present article.

Our ability to determine the most important ADHD pharmacogenomic predictors is also likely limited by the field’s focus on pharmacodynamic genes and its dependence on the candidate gene approach. In many branches of medicine where PGx tools are gaining traction, the approach has centered on elucidating the role of pharmacokinetic gene polymorphisms.[139] Indeed, we have made progress in understanding how CYP2D6 polymorphisms impact ATX response such that modifications to ATX dosing based on CYP2D6 genotyping are included in ATX’s FDA labeling. In addition, an ATX-CYP2D6 CPIC guideline has recently been published.[129] However, few prior publications have examined the association between pharmacokinetic genes and MPH, amphetamine, guanfacine, or clonidine efficacy. Furthermore, given our imperfect understanding of the neurobiology underlying ADHD medication response, it is unlikely that we can develop candidate polymorphism hypotheses that cover all the salient pharmacodynamic inputs. To address this limitation, it is likely that more hypothesis-free efforts, such as genome-wide approaches, are needed. However, the need for large sample sizes, which entail considerable expense, presents a significant barrier to conducting PGx GWAS. Investigators targeting other diseases/disorders have addressed this barrier by creating large databases/registries which bank standardized clinical data and biological samples (including DNA) academic institution-wide or across multiple collaborating institutions in a study consortium (e.g., the Children’s Oncology Group).[140] Collection of DNA samples as a standard for all ADHD medication randomized controlled trials would also considerably increase the available data for future ADHD PGx efforts. The resultant sizable databases could be used not just for GWAS, but could also serve as a convenient repository facilitating replication of candidate gene findings from discovery samples (thereby acting as a check on the release of false positive findings).

Increasingly, personalized predictive efforts in other branches of medicine are also expanding beyond standard pharmacogenomic approaches focusing on the code of individual genes. Indeed, there is growing recognition that regulation of gene expression through epigenetic mechanisms may play an important role in drug response, and there is increasing evidence that epigenetic patterns in blood samples can predict and/or reflect response to psychiatric medications, including antidepressants[141–148] and antipsychotics.[146, 149–153] It is also being proposed that pharmacometabolomic approaches may be able to capture the impact of environmental, gene*environment, gene*gene, and epigenetic influences. The metabolome is the repertoire of small molecules–including amine neurotransmitters, lipid signaling molecules, and many other classes of compounds--found within a biological system and representing the final product of interactions between gene expression, protein expression, and the cellular environment.[154] Intriguingly, pharmacometabolomic studies of schizophrenia[155] and major depression[156] have furthered understanding of drug response and provided groundwork for identifying predictive metabolic markers for psychiatric medications.

7. Conclusions

Despite the proliferation of numerous additional PGx publications over the past decade, the era of personalized predictive ADHD clinical care has not yet dawned. To make progress, adequately powered studies which pay careful attention to the study design and sample considerations discussed above are needed. ADHD pharmacogenomic efforts which incorporate pharmacokinetic as well as pharmacodynamic gene investigations, genome-wide approaches, and gene*gene interactions are also likely necessary to drive the field forward. Efforts which go beyond investigation of gene coding sequences, possibly by incorporating epigenetic and metabolomic approaches, may also be crucial to the development of clinically salient ADHD medication response prediction tools.

Key Points.

• Over the past decade, pediatric ADHD medication efficacy pharmacogenomics research has largely focused on pharmacodynamic candidate gene studies of methylphenidate response. Although results of individual studies have often lacked consistency, recent meta-analyses have reported a small but significant impact on methylphenidate efficacy of variants in the dopamine transporter (DAT/SLC6A3), dopamine receptor D4 (DRD4), adrenergic α_2A-receptor (ADRA2A), catechol-O-methyltrasferase (COMT), and norepinephrine transporter (NET/SLC6A2) genes.

• Progress has also been made in understanding how genetic polymorphisms in the atomoxetine-metabolizing enzyme CYP2D6 impact atomoxetine response and pharmacokinetics. In fact, a recently published Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline outlines some recommended variations in atomoxetine dosing based on CYP2D6 genotype.

• Notwithstanding these advances, the field to date has been unable to translate the plethora of available ADHD pharmacogenomics study findings into a tool that can predict with precision which ADHD medication will be most effective for each individual patient. There are myriad factors underlying this impasse--including the small effects of individual genetic variants, study design weaknesses and inconsistencies, lack of confirmatory studies in different racial/ethnic populations, and failure to consider gene-gene interactions--which should be addressed when undertaking future ADHD pharmacogenomics research.