Intra-individual variability in genetic and environmental models of attention-deficit/hyperactivity disorder^†

INTRODUCTION

Attention-deficit/hyperactivity disorder (ADHD) is a serious behavioral disorder affecting 8–12% of adolescents [Faraone et al., 2005]. Besides hyperactivity, inattention and impulsiveness, ADHD-affected individuals frequently have high variability in reaction time during stop-signal, Go/No-Go and continuous performance tasks [Castellanos et al., 2005; Nigg et al., 2005; Klein et al., 2006; Di Martino et al., 2008], termed intra-individual variability (IIV). IIV is common and may itself be an endophenotype of ADHD [Castellanos et al., 2005; Sagvolden et al., 2005a; Johnson et al., 2007; Geurts et al., 2008; Johansen et al., 2009]. Little is known of the physiological/genetic basis for IIV although various works suggest deficiencies in energy supply to active neurons or the poor re-establishment of ionic homeostasis across neuronal membranes [Russell et al., 2006], poor noradrenergic/dopaminergic co-function [Cho et al., 2008], deficits in neurocognitive timing [Castellanos et al., 2005], or poor behavioral mechanisms due to dopamine hypofunction [Aase and Sagvolden, 2005; Sagvolden et al., 2005a; Aase et al., 2006; Johansen et al., 2009].

Various rodent models of ADHD exist. The Spontaneous Hypertensive Rat (SHR), in particular, expresses all major features of ADHD, including hyperactivity, impulsiveness and inattention [Sagvolden et al., 2005b; Kantak et al., 2008], smaller brain volume [Bendel and Eilam, 1992], sex differences in ADHD-like phenotype [Sagvolden and Berger, 1996; Berger and Sagvolden, 1998; Bucci et al., 2008] and differences in the expression of genes for catecholaminergic functioning, neuronal metabolism, and ion channels compared to controls [Mill et al., 2005; Russell et al., 2005; Das-Banerjee et al., 2008]. The Wistar-Kyoto Hyperactive Rat (WKHA) expresses hyperactivity specifically [Moisan et al., 2003]. Environmental contaminants such as polychlorinated biphenols (PCBs) also induce hyperactivity and impulsiveness, the prenatal period and weaning periods being the most sensitive for neurological damage by PCBs [Pantaleoni et al., 1988; Lilienthal and Winneke, 1991; Morse and Brouwer, 1995]. Ortho-substituted congeners are common [McFarland and Clarke, 1989] and have marked effects on neurological behavior [Holene et al., 1995; Schantz et al., 1995]. Prenatal exposure to PCBs leads to deficits in executive function and ADHD in humans [Boucher et al., 2009]. Rats exposed prenatally to PCBs also show ADHD-like symptoms [Daly et al., 1989; Holene et al., 1998; Berger et al., 2001] and dysregulated expression of catecholaminergic genes [Das-Banerjee et al., 2008].

There appears to be a genetic component to IIV: Andreou et al. 2007 found that familial elements of reaction time variability (as much as 70% of phenotypic variance) were shared with familial variance components in ADHD. Wood et al. 2009 found that genetic variance associated with ADHD and with mean reaction time was also highly correlated with variability in reaction time. SHR have higher IIV in correctly locating target holes during nose-poking investigations: SHR located the correct hole a third of the time with initial investigations, while WKY rats nose-poked the correct hole approximately half the time [Johansen et al., 2007]. However, little is known of IIV in other genetic or environmental models of ADHD, or of IIV in traits other than reaction time. Our objective was to test IIV in ADHD-like traits in rodent models of ADHD and their controls. The identification of such differences might permit the mapping and eventual identification of candidate genes or physiological systems responsible for IIV.

MATERIALS AND METHODS

RESULTS

Intra-Individual Variability

In each of the original studies, the SHR had higher hyperactivity, impulsiveness and inattention than the WKY rats [Sagvolden et al., 1992, 1993; Berger and Sagvolden, 1998; Boix et al., 1998; Holene et al., 1998]. PCB-treated rats had higher lever-pressing activity and shorter IRTs [Holene et al., 1998].

SHR rats had significantly higher PD_i for FI hyperactivity than WKY rats in three of the four studies in which they were tested against the latter (Table II, Fig. 1; P < 0.05). PD_i in SHR males was significantly higher than in WKYs of either sex (P < 0.05) after Bonferroni correction. Dispersion for hyperactivity during the EXT phase resembled that during FI; SHRs had higher PD_i for hyperactivity than the WKY in all tests involving both strains, and higher PD_i for hyperactivity than SPRD rats (Table II, Fig. 2; P < 0.05). As for the FI phase, SHR males had higher PD_i than male (P < 0.01) and female WKYs (P < 0.05) after Bonferroni correction, and female SHR higher PD_i than male WKYs (Table II, Fig. 2; P < 0.05). PD_i for hyperactivity was significantly higher in SHR than in WKY across the five studies using contrast analysis (P < 0.0001; Table III).

Table II. Differences Among Strains and Sexes for Individual Phenotypic Dispersion (PD_i) of ADHD-Like Behavior During Operant Behavior Challenges (PROC NPAR1WAY, SAS [1999]) by Study and Trait

Behavior/period	Study	n	χ2	P
Hyperactivity—FI	Boix et al. 1998	17	7.25	0.0071
	Berger and Sagvolden 1998	32	17.6	0.0005
	Sagvolden et al. 1992	32	11.3	0.0103
Hyperactivity—EXT	Boix et al. 1998	17	10.1	0.0015
	Berger and Sagvolden 1998	32	18.6	0.0003
	Sagvolden et al. 1992	32	8.49	0.0369
	Sagvolden et al. 1993	48	10.7	0.0302
Impulsiveness—FI	Boix et al. 1998	17	4.48	0.0343
	Berger and Sagvolden 1998	32	14.5	0.0023
	Sagvolden et al. 1992	32	7.88	0.0486
Impulsiveness—EXT	Boix et al. 1998	17	8.34	0.0039
	Berger and Sagvolden 1998	32	13.7	0.0033
	Sagvolden et al. 1992	32	12.0	0.0072
	Sagvolden et al. 1993	48	13.1	0.0110
Attention	Boix et al. 1998	17	5.79	0.0161
	Berger and Sagvolden 1998	32	10.1	0.0177
	Sagvolden et al. 1993	48	9.50	0.0498

• χ² and P values are derived from Kruskal–Wallis tests of differences for the entire data set.

Table III. Multi-Study Estimates of Differences in Individual Phenotypic Dispersion (PD_i) Intra-Individual ADHD-Like Behavior Between the Spontaneous Hypertensive Rat (SHR) and Wistar-Kyoto (WKY) Rat

Trait	F	P	(95% CI)
Impulsiveness, fixed-interval	17.6	<0.0001	−3.56 (1.83)
Impulsiveness, extinction interval	3.21	0.0756	−0.821 (0.893)
Hyperactivity, fixed-interval	36.3	<0.0001	−17.1 (5.57)
Hyperactivity, extinction interval	ns	—	—
Inattention	ns	—	—

• Calculated from Boix et al. 1998, Berger and Sagvolden 1998, and Sagvolden et al. 1992, 1993. The estimate and associated 95% confidence interval (CI) refer to the estimated difference in (µ_PDi, _SHR-µ_PDi,WKY) over the complete array of all four studies involving SHR and WKY rats.

SHR rats also had higher PD_i for impulsiveness in the FI phase than WKY rats in two of the four studies in which they were tested. SHR males and females both had significantly (P < 0.05) higher PD_i for FI impulsiveness than female WKY rats in Boix et al. 1998 (Table II; Fig. 3). The SHR had higher a priori PD_i for FI impulsiveness than the WKY in the data of Sagvolden et al. 1992, but these differences were not significant post-Bonferroni correction (Table II; Fig. 3). As in the FI phase, SHR had higher PD_i for EXT impulsiveness than the WKY in two of the four studies comparing their dispersion. SHR males had higher PD_i for impulsiveness than WKY females (P < 0.05) in the study of Berger and Sagvolden 1998; unlike the FI phase, SHR also had higher PD_i than WKHT for EXT impulsiveness (P < 0.05) in the data of Sagvolden et al. 1992, and higher PD_i than PVG (P < 0.05) and SPRD rats (P < 0.01) from Sagvolden et al. 1993( Fig. 4). As for hyperactivity, contrast analysis indicated that PD_i for FI and EXT impulsiveness was significantly higher in the SHR across all four studies overall than the WKY (P < 0.0001; Table III).

SHR had significantly higher PD_i for inattention than WKY rats (P < 0.05) from the data of Boix et al. 1998 and male SHR significantly higher PD_i than WKY females from Berger and Sagvolden 1998 (Table II; Fig. 5). SHR also had higher PD_i than SPRD rats (P < 0.05) (from Sagvolden et al. 1993). As hyperactivity and impulsiveness, contrast analysis indicated that PD_i for inattention was significantly higher in the SHR than the WKY across all four studies in which they were represented (P < 0.0001) (Table III).

DISCUSSION

Our results indicate genetic effects on IIV (measured here as PD_i) and illustrate that the SHR has higher IIV for ADHD-related characters than other strains, including WKY. As in children with ADHD [Castellanos et al., 2005; Klein et al., 2006; Di Martino et al., 2008; Geurts et al., 2008], the SHR strain, and male SHR in particular, were more variable in their expression of ADHD-like phenotype than WKY controls. This recommends both that SHR as a model of IIV and the WKY strain as a control line in mapping genes for IIV. Our interpretation is partially cautioned by the use of dispersion values uncorrected for the mean, raising the possibility of heterscedasticity. However, the extremely high values of operant phenotype expressed by the SHR in the four source studies that include that strain [Sagvolden et al., 1992, 1993; Berger and Sagvolden, 1998; Boix et al., 1998] make such a transform highly susceptible to overcorrection for SHR phenotype. Mean-transformed values for dispersion (not shown) resulted in a range of results, with mean PD_i in the SHR ranging everywhere from significantly higher to significantly lower PD_i than our central control strain, the WKY. This suggests that such an approach may be overly conservative. We expect to be able to separate means and variances for operant behavioral phenotype from that of the SHR and WKY strains using F₂ SHR × WKY intercrosses, permitting us to simultaneously search for classically acting genes and those responsible for IIV.

There are a number of mechanisms by which IIV in ADHD symptoms might occur. Russell et al. [Russell et al., 2006] proposed that IIV resulted from deficiencies in astrocyte function. Under this hypothesis, insufficient production of ATP to active neurons and inadequate supply of lactate for energy by neighboring astrocytes [Sanchez et al., 2001; Kasischke et al., 2004], resulting short-term IIV [Russell et al., 2006]. A deficiency in ATP production might also limit the activity of Na⁺/K⁺ATPase and Ca⁺ATPase pumps, disrupting transmembrane ionic gradients and affecting the ability of neuronal cells to release neurotransmitters [Russell et al., 2006]. Astrocytes do appear to affect neurotransmission [Grimaldi et al., 1997; Moldrich et al., 2002; Miyazaki et al., 2004]. The state regulation hypothesis of Johnson et al. 2009 strongly resembles this case; in this hypothesis, task efficiency (and thus variability) is a function of short-term energetic status and the arousal/activation state of a stream of task response elements including stimulus recognition, the search of memory, cognitive decision and finally motor activity. IIV occurs generally in several other human behavioral disorders such as schizophrenia [Schwartz et al., 1989], cognition after head injury [Stuss et al., 1989; Walker et al., 2000; Murtha, 2002] and Tourette's syndrome [Peterson and Leckman, 1998; Geurts et al., 2008], which jointly suggests dysfunction in basal physiology such as energetic metabolism. This is also suggested by IIV in Tourette's Syndrome, which has been linked to individual variability in motor neuron activity [Walters et al., 2001].

IIV, as an essentially random process, might result from poor regulation of normal periodicity in neural physiology, such as that associated with autonomous functions like heart rate and respiration [Castellanos et al., 2005]. This could result from poor co-ordination of catecholaminergic physiology, which appears to modulate executive function and attention [Faraone, 2004, 2006; Faraone et al., 2005; Staller and Faraone, 2007; Cho et al., 2008; Mick and Faraone, 2008]. This presumably occurs via variability in noradrenergic output by the locus coeruleus, causing variability in prefrontal cortex regulation and increased noise in neuron function and attention [Arnsten et al., 1996; Cho et al., 2008]. Notably, the expression of genes involved in dopaminergic/catecholaminergic physiology in the SHR differs to that of control rats [Mill et al., 2005; Russell et al., 2006; Das-Banerjee et al., 2008].

The dynamic developmental theory of ADHD [Sagvolden et al., 2005a] suggests that reduced dopamine activity changes fundamental behavioral selection mechanisms by causing deficient reinforcement of successful behavior combined with deficient EXT of unsuccessful behavior. Such deficient selection mechanisms will slow the association (“chunking”) of simple response units into longer, more elaborate chains of adaptive behavioral elements that function as higher-order response units [Aase and Sagvolden, 2005; Aase et al., 2006]. When response units are chunked together into a chain, one response unit reliably precedes the next with high predictability within the chain. Deficient or slowed chunking of behavior means that the reliable and predictable pattern of responses is absent, resulting in the increased IIV observed in ADHD [Johansen et al., 2009].

It has been argued that the phenomenon of IIV in ADHD represents either a nuisance element, or an underlying genetic liability intrinsically associated with ADHD [Castellanos et al., 2005; Russell et al., 2006]. The latter seems more likely; the 10-repeat allele at the dopamine transporter (DAT), previously associated with ADHD [Faraone et al., 2005], has also been associated with IIV in specific behavioral [Bellgrove et al., 2005] and neurological [Loo et al., 2003] phenotypes. The patterns of differences in IIV among strains (especially for SHR-WKY contrasts) in this study obliquely support our hypothesis that IIV is under genetic control rather than representing random error. Notably, the total and the posterior area of the corpus callosum has been negatively associated with IIV in reaction time [Anstey et al., 2007]; as ADHD-affected humans [Paule et al., 2000; Castellanos et al., 2002] SHR have smaller brain size compared to controls [Bendel and Eilam, 1992].

Our findings suggest a genetic component to IIV in ADHD. However, clinical implications of IIV on ADHD and its genetics are not entirely clear. IIV could create difficulties in the diagnosis of borderline ADHD-affected individuals with high variance in their symptoms, or require the restructuring of cognitive tasks used for the diagnosis of ADHD, such as the division of longer tasks into shorter ones with more rapid reinforcement [Sagvolden et al., 2005a; Russell et al., 2006; Cho et al., 2008]. In the broader sense, a genetic basis for IIV in ADHD suggests the existence of genes causing IIV in other traits and other model organisms, for which there is already some evidence [SanCristobal-Gaudy et al., 1998; Perry et al., 2003; Sorensen and Waagepetersen, 2003; Mackay and Lyman, 2005]. The genetic control of IIV would require a major reevaluation of the fundamental Fisherian theorem [Fisher, 1918] to include estimates of the genetic control of residual variability, radically altering the entire continuum of all quantitative genetic research.