Abuse of Amphetamines and Structural Abnormalities in Brain

Abstract

We review evidence that structural brain abnormalities are associated with abuse of amphetamines. A brief history of amphetamine use/abuse, and evidence for toxicity is followed by a summary of findings from structural magnetic resonance imaging (MRI) studies of human subjects who had abused amphetamines and children who were exposed to amphetamines in utero. Evidence comes from studies that used a variety of techniques that include manual tracing, pattern matching, voxel-based, tensor-based, or cortical thickness mapping, quantification of white matter signal hyperintensities, and diffusion tensor imaging. Ten studies compared controls to individuals who were exposed to methamphetamine. Three studies assessed individuals exposed to 3-4-methylenedioxymethamphetamine (MDMA). Brain structural abnormalities were consistently reported in amphetamine abusers, as compared to control subjects. These included lower cortical gray matter volume and higher striatal volume than control subjects. These differences might reflect brain features that could predispose to substance dependence. High striatal volumes might also reflect compensation for toxicity in the dopamine-rich basal ganglia. Prenatal exposure was associated with striatal volume that was below control values, suggesting that such compensation might not occur in utero. Several forms of white matter abnormality are also common, and may involve gliosis. Many of the limitations and inconsistencies in the literature relate to techniques and cross-sectional designs, which cannot infer causality. Potential confounding influences include effects of pre-existing risk/protective factors, development, gender, severity of amphetamine abuse, abuse of other drugs, abstinence, and differences in lifestyle. Longitudinal designs in which multimodal datasets are acquired and are subjected to multivariate analyses would enhance our ability to provide general conclusions regarding the associations between amphetamine abuse and brain structure.

Amphetamines and their use

Synthesized in 1887, amphetamine (1-methyl-2-phenethylamine) was the first member of a group of compounds that have similar structures and biological properties and are collectively called “amphetamines”. The group also includes methamphetamine, synthesized six years later, and 3-4-methylenedioxymethamphetamine (MDMA), patented in 1914. Amphetamines produce their principal effects by increasing synaptic levels of the biogenic amines, dopamine, norepinephrine and serotonin, through multiple mechanisms^1-3.

Amphetamine is FDA-approved for treatment of attention deficit-hyperactivity disorder (ADHD) and narcolepsy. Methamphetamine is approved for treatment of ADHD and obesity. Both drugs are classified by the Drug Enforcement Agency (DEA) as belonging in Schedule II, having accepted medical uses but being tightly controlled because of their potential for abuse that can lead to severe psychological and physiological dependence. Although once unregulated and used as an adjunct in psychotherapy, MDMA has been added to DEA Schedule I, indicating a high potential for abuse but no accepted medical use.

Amphetamines have been used illegally since the FDA limited them to prescription use in 1965. From 1971, when 30% of US college students surveyed reported using amphetamines illegally⁴, illicit amphetamine use declined in the US during the 1980s. However, use subsequently increased in the 1990s and has continued to rise in young adults. Although data from the 2006 US National Survey indicated that illicit use of the combined licit and illicitly-manufactured amphetamine may have peaked, illicit use of prescription medications is currently at its highest level in decades and amphetamines are the most abused prescription medications⁵.

In addition to abuse of pharmaceutical amphetamines, there has been substantial illicit manufacture and abuse of methamphetamine and MDMA. Substantial levels of illicit manufacture of methamphetamine began in the 1960s in the United States and elsewhere, and has accelerated in both amount and geographic distribution since the 1980s. Methamphetamine has thus become one of the world’s fastest growing illicit drug problems⁶. Since its criminalization in the 1980s, MDMA has been manufactured and consumed as a popular recreational drug, commonly known as “ecstasy”. Legislative changes have not effectively discouraged its use⁷.

In sum, abuse of amphetamines constitutes a serious public health concern. Amphetamines are the prescription drugs most commonly abused by adolescents and young adults, and illicit amphetamines are second only to marijuana as a form of illicit drug abuse in young adults⁵. Prevalence of problematic use of amphetamines by older adults has also been rising. Emergency department mentions of amphetamines among patients 55 years and older have increased 700% from 1995 to 2002⁸, for example.

Amphetamine neurotoxicity

It is well known that exposure of experimental animals to acute, high doses of amphetamine and methamphetamine alters dopaminergic neurons that innervate the striatum (caudate-putamen)^9,10. Exposure of experimental animals to acute, high doses of MDMA alters serotonergic neurons^11,12. Dopaminergic toxicity is inferred from deficits in phenotypic markers for dopaminergic nerve terminals, including dopamine itself, its biosynthetic enzymes tyrosine hydroxylase and L-aromatic amino acid decarboxylase, and both the plasma membrane dopamine transporter (DAT) and the vesicular monoamine transporter (VMAT)¹³. In rats given MDMA, protracted depletion of forebrain 5-HT (serotonin), reductions in evoked 5-HT release, changes in hormone secretion, and persistent anxiety-like behaviors have also been interpreted as evidence for neurotoxicity, although this interpretation is not conclusive¹².

High doses of amphetamines have produced hyperemia, hemorrhage and glial proliferation in monkeys¹⁴, and enlarged chromatolytic medullary neurons in cats¹⁵. Parenteral dosing in rodents can also produce swelling or reduction of dopaminergic axons, reduced dopaminergic terminals, and serotonin deficits. Deficits in dopaminergic nerve terminals are not always accompanied by apparent damage to the dopamine-containing cell bodies within the substantia nigra, but can persist during years of abstinence from drug exposure. The mechanisms responsible for amphetamine-induced neurotoxicity have not been fully identified. However, available evidence suggests that the high levels of cytoplasmic dopamine associated with amphetamine-mediated disruption of vesicular storage leads to accumulation of reactive oxygen species and severe oxidative stress^16,17. This may be due in part to methamphetamine-associated increases in brain iron¹⁸, since iron is a well-known catalyst for oxidative reactions¹⁹. Because similar neurotoxicity has not been apparent after high-dose exposure to the non-amphetamine stimulant methylphenidate^20,21, it has been speculated that damage may reflect disruption of vesicular storage of dopamine, an action of amphetamine, but not of methylphenidate.

Neurotoxicity of amphetamine used therapeutically?

It is not known whether long-term administration of prescription amphetamines at doses abused by humans produces similar alterations in the dopaminergic system²². Because evidence for amphetamine-mediated neurotoxicity in rodents derives primarily from studies in which high doses were parenterally administered^23,24, and because repeated exposure of rodents to lower doses does not produce such evidence²⁵, the relevance of these data for therapeutic use of amphetamines in humans can be questioned. However, in the most relevant animal model, adult baboons and squirrel monkeys were treated for 4 weeks with a 3:1 mixture of dextroamphetamine to levoamphetamine, a ratio similar to that found in the pharmaceutical product Adderal, at doses that mimic those used in human clinical treatment²⁶. Plasma concentrations of amphetamine (136 +/- 21 ng/ml) matched the levels reported in human ADHD patients (120 to 140 ng/ml) after amphetamine treatments that lasted for 3 weeks²⁷ or 6 weeks²⁸. Both monkey species treated in this way developed 30-50% reductions in striatal dopamine, its major metabolite dihydroxyphenylacetic acid, its rate-limiting enzyme (tyrosine hydroxylase), DAT, and VMAT. Other evidence suggests that non-human primates may be more vulnerable than rodents to stimulant-induced neurotoxicity²⁹. However, a large clinical study of clinically treated individuals with ADHD showed no impact on the developmental trajectories of brain volumes³⁰.

The potential for amphetamine-mediated neurotoxicity may vary with age, with cumulative exposure to the drug, or even with the behavioral context of drug administration. Brain amphetamine levels at both 20 and 65 min after IP administration of 2.5 mg/kg of amphetamine were twice as high in the brains of old as compared to young rats, suggesting that aging might increase the risks for toxicity³¹. In fact, older rats, mice and gerbils all experienced greater methamphetamine neurotoxicity than younger animals, as manifested by striatal dopamine reduction, structural deficits and increased levels of glial fibrillary acid protein^32-34. Stefanski and associates³⁵ observed regional downregulation of dopamine D2 and D1 receptors in rats who had learned to self-administer methamphetamine. At the relatively low doses delivered, no further pathology was seen. This observation suggests that doses of methamphetamine that may be too low to be neurotoxic can induce dependence. Each of the above factors suggests that amphetamine-abusing patients examined in neuroimaging studies may represent a heterogeneous mix of individual with a combination of brain factors that might predispose to methamphetamine dependence and neurotoxic sequelae of methamphetamine use. This heterogeneity may manifest as variability in neuroimaging endpoints such as volumes of macroscopic structures within the brain.

Neurotoxicity from abuse

Most of the evidence for amphetamine-induced human brain damage comes from examination of current and former amphetamine and methamphetamine abusers. Evidence suggesting that the neurotoxicity reported in animals also occurs in methamphetamine-abusing humans has accrued from neuroimaging findings of reduced availability of dopamine D2 receptors, and transporters for dopamine, serotonin, and vesicular monoamines³⁶. Autopsy data consistent with dopaminergic neuronal damage consist of deficits in dopamine, the dopamine transporter, and tyrosine hydroxylase^15,37,38. However, autopsy data has revealed little reduction in levels of the vesicular monoamine transporter (VMAT₂). These findings contrast with the profound decrements in VMAT₂ expression that follow administration of high stimulant doses to rodents or nonhuman primates^39-42. Because a decrease in VMAT₂ is thought by some to indicate reduction of intact monoamine nerve terminals^39,43-45, it has been suggested that decrements in dopamine transporter in the absence of decrements in VMAT₂ may represent compensatory changes rather than degeneration of dopamine terminals³⁸.

Proton magnetic resonance spectroscopy of cortex and basal ganglia metabolites has consistently reported reduced markers of neuronal integrity, and increased markers of glial content, suggesting that glial proliferation may follow neural damage³⁶. Using glucose metabolism as an index of functional activity in brain, abnormalities have been observed in research subjects who were in early abstinence from chronic abuse of methamphetamine. Abnormally high relative activity was observed in amygdala, ventral striatum and lateral orbitofrontal cortex. Abnormally low activity was noted in medial prefrontal and, especially, cingulate cortex⁴⁶. Continued abstinence from the drug accompanies abnormally high global and cortical glucose metabolism, particularly in the parietal lobe^47,48, and relatively lower activity, when scaled to global mean activity, in striatal and thalamic regions^48,49.

Some degree of recovery after protracted abstinence has been noted in studies of perfusion of the cingulate cortex⁵⁰, striatal dopamine transporters⁵¹, and glucose metabolism in the thalamus but not the striatum⁴⁹. When we retested 12 healthy control subjects and 10 hospitalized methamphetamine abusers one month after an initial test when the methamphetamine abusers had been abstinent for 5-9 days, glucose metabolism did not change in subcortical regions of the methamphetamine abusers but it increased in their neocortical regions ⁴⁷. The increase exceeded 20% in the parietal lobes. Increased cortical activity was interpreted as reflecting effects of compensatory processes, or unmasking of damage that might be obscured by suppression of cortical glucose metabolism for at least 5 days after cessation of drug use. Alternatively, new damage might conceivably be sustained during abstinence.

The possibility of additional damage during early abstinence from amphetamine is consistent with observations that three daily exposures of rats to methamphetamine was sufficient to induce reactive gliosis that continued for over two weeks after the final exposure⁵². In addition, the P300 event-related potential recorded from the human scalp, which is modulated by catecholaminergic neurotransmission^53,54, is reduced in amplitude during early abstinence from chronic methamphetamine abuse, and was also reduced in a rat model by 15 days of methamphetamine followed by over a week of abstinence⁵⁵.

For MDMA, the evidence for human neurotoxicity is less definite than for methamphetamine, partially because of methodological problems and potential confounds in the extant studies, which will be discussed in the last half of this chapter. However, the bulk of the evidence suggests there are residual alterations of serotonergic transmission in MDMA users. There appears to be partial restitution with abstinence, but also persistent functional sequelae that persist even after long periods of abstinence⁵⁶.

Structural brain abnormalities and methamphetamine

Most of the evidence for structural brain damage in human substance abusers derives from recently developed applications of magnetic resonance imaging (MRI), which can be used for quantitative morphometrics and for assessment of structural change over time. Many of the relevant studies have involved automated or semi-automated segmentation of the brain into gray and white matter components, for restricted or separate analyses. Techniques that have been used include a wide range from labor-intensive methods, such as manual slice-by-slice tracing and volume drawing of individual structures, as well as more computationally intensive approaches that include pattern matching as well as voxel-based, tensor-based and cortical thickness mapping⁵⁷. This review also includes studies that assess white matter signal hyperintensities and diffusion tensor imaging (DTI) indices, since these measures help us understand the state of white matter microstructure in methamphetamine abusing individuals.

The strongest evidence for stimulant-associated differences, and possible stimulant-mediated neurotoxicity, has come from studies that relate to the use of methamphetamine (see Table 1, in which reports are presented in the approximate order of publication). We will first review studies of methamphetamine abusers and then discuss studies that assess MDMA abusers.

Table 1.

Imaging evidence for structural abnormalities in amphetamine abusers

	Groups	Male	Amphetamine use variables (MEAN ± SD)	Measures	Tesla	Regions	Assessed?*	Results
Bartzokis et al. 200058	9 MA	9	MA use > 2 years (mean 6.7 ± 3.9 years)	Manual tracing of MRI image	1.5	Whole brain	WM,GM	↓ temporal lobe volume
			abstinence >2 weeks	(total, gray & white matter volumes)		frontal lobe		↓ temporal lobe GM volume
	16 C	16				temporal lobe
Cowan et al. 200384	31 polydrug with MDMA	17	MDMA use: 26%<11x, 48%=11-39x, 26%>40x	VBM (voxel-based morphometry)	1.5	Whole brain	WM,GM,	↓ GM concentration in occipital (BA18),
			Polydrug users, 55% used other amphetamines	gray & white matter				L. temporal (BA21), L. frontal lobe (BA45),
	29 polydrug without MDMA	18	abstinence >3 weeks				S	R/L. cerebellum, medial brainstem
Thompson et al. 200461	22 MA	15	MA mean use 10.5±1.1 years	T1 segmentation/volume drawing	3	Cortical sulci & gyri	WM,GM	↑ R. temporal occipital lobe WM & frontal horn of lateral ventricle
			cumMA(g)=1878±126	cortex & hippocampal pattern matching		hippocampus		↓ R. mid-posterior cingulate GM, ↓hippocampal volume
	21 C	10	used 18.9±1.8 of prior 30 days	mood, word/picture recall & recognition				↓hippocampus correlates with ↓ word recall
Chang et al. 200462	13 MA	4	MA exposure > 2/3 of gestation	T1 segmentation/pattern matching	1.5	Caudate, putamen, globus pallidus	None	↓ striatum & hippocampus
			age = 6.9± 3.5	neuropsychological battery		thalamus, midbrain,	(unsegmented volume)	↓attention, memory, visual motor integration
	15 C	6				hippocampus, cerebellar vermis		↓striatum correlated with ↓ attention & delayed verbal memory
Chang et al. 200563	50 MA	24	MA use>2 years (mean 110± 68 months)	T1 Drawn ROIs	1.5	Caudate, putamen, globus pallidus	WM,GM,	↑ striatal volumes
			cumMA(g)=4519±5730	neuropsychological battery		thalamus, midbrain, cerebellar vermis	A,	↓ striatal volume correlated with ↓ cognitive performance & ↑ lifetime MA use
	50 C	24	abstinence>1 week (mean 4±6 months)			Corpus Callosum (4 ROIs)	S	female MA users ↑ callosal posterior midbody
Oh et al. 200566	27 MA	23	cumMA(g)= 334±506	MRI-modeled callosal extraction	3	Corpus Callosum (7 ROIs)	WM	↑ Genu curvature
			mean abstinence 21±35 months	automated shape analysis				↓ width posterior midbody & isthmus
	18 C	14		callosal width in each region				(ie - frontal & parietal WM abnormalities)
Jernigan et al. 200568	21 MA	17	MA use 6-20 years (mean 12.1±4.1 years)	T1 semi-automated morphometry	NG	Striatum, thalamus	GM,	↑striatum & parietal cortex
	22 MA+HIV	21	cumMA(g)= 4930±94	neuropsychological exam. (unspecified)		hippocampus, amygdala		younger MA users = ↑nucleus accumbens volume
	30 C	17	abstinence > 10 days (mean 94±89 days)			cortex (all lobes)	S	↑parietal lobe correlated with cognitive impairment
Bae et al. 200669	33 MA	22	MA mean use 59±36 months	White matter hyperintensity (WMH)	3	Deep (all lobes + insular),	WM,	↑ WMH (all, deep & periventricular)
			cumMA(g)=292±247	rated/graded on T2 MRI	periventricular (frontal and nonfrontal)		A,	Severity of deep WMH correlated with lifetime MA use
	32 C	21	mean abstinence 18±29 months				S	M > Fem (No abnormality in Fem)
Schlaepfer et al. 200674	16 polydrug abusers	16	Most (9/16) abused MA, but also abused	Semi-automated segmentation	1.5	Whole brain	WM,GM	↓ frontal WM volume (both methods)
			other drugs for 7-12 years. All used	stereological volume estimates		frontal lobe
	16 nonabusers	16	cannabis, 14 cocaine, 15 heroin.	total, gray matter, white matter		lateral ventricles
Kim et al. 200680	11 short-abstinent MA	11	MA mean use 64±44 months	VBM	3	Whole Brain	WM,GM,	↓ GM Density R. Midfrontal Cortex
	18 long-abstinent MA	16	cumMA(g)= 276±236	WCST, Trailmaking test, STROOP		gray and white matter	A	↓WCST was correlated with R. Midfrontal Cortex ↓ GM
	20 C	15	mean abstinence 20±34 months, long=>6 months					Both effects partially recover at > 6 months abstinence
Moeller et al. 200785	12 polydrug with MDMA	10	MDMA use > 1 year (mean 6.2±5.0 years)	DTI (FA, diffusivity, longitudinal & transverse)	1.5	Corpus Callosum	WM	↓ Longitudinal eigenvalues rostral corpus collosum body
			abstinent from 0-101 days (mean 40)	Iowa Gambling Task (IGT)		(6 ROIs)		↓IGT was correlated with rostral body longitudinal eigenvalue
	20 C (no drugs)	13
De Win et al. 200790	30 MDMA users	12	mean MDMA use 1.8 ± 1.3 tablets	DTI (FA, apparent diffusion)	1.5	Thalamus,	WM,	↑ FA centrum semiovale (trend)
	pre + post early MDMA use		0.9 - 29.5 months between tests (mean 8.1± 6.5)	psychological questionnaires		3 basal ganglia ROIs,		↑ centrum semiovale FA was correlated with number of MDMA tablets
			abstinence >3 weeks at test 2 (mean 7.7± 4.4)	MRS, PWI		centrum semiovale	S
Chung et al. 200781	32 MA	23	MA mean use M= 75±51, Fem= 47±50 months	DTI FA	3	Frontal white matter ROIs	WM,	↓ Frontal FA (WM integrity)
			cumMA(g) M=412±543, Fem=133±150	WCST				↓WCST was correlated with FA
	30 C	20	>1month abstinence, mean M=24±38, Fem=43±66				S	M >Fem

MRI=magnetic resonance imaging, MRS=magnetic resonance spectroscopy, DTI=diffusion tensor imaging, FA=fractional anisotropy, PWI= perfusion -weighted imaging, MA=methamphetamine, MDMA=methylenedioxymethamphetamine, C=control subjects, cum=cumulative, ROI=region of interest, R=right, L=left, BA= Brodmann area, NG=not given, M=male, Fem=female

^*WM, GM, A, & S represent assessment of white matter, gray matter, abstinence effects, and sex effects.

The first controlled study to measure brain volumes in amphetamine users compared 9 methamphetamine–dependent young men (27.9 [SD=3.9years]) with 10 cocaine-dependent men and 16 men who did not use drugs⁵⁸. It involved semiautomated, histogram-driven measurements of frontal and temporal lobes on 1.5 Tesla (1.5T) T2-weighted MRI scans. Methamphetamine-dependent subjects had used the drug for more than 2 years (mean = 6.7 +/- 3.9 years), although the length of time between most recent consumption and imaging was not available. Smaller volumes of the temporal lobe, but not of the frontal lobe, were found in both methamphetamine and cocaine users, in comparisons with control men. The stimulant using groups did not differ from each other overall. However, only the cocaine group demonstrated an age-related decline in temporal lobe volume. In both groups, reductions in temporal lobe volume were localized primarily to gray matter. Smaller gray matter volume was invariably accompanied by larger white matter volume, consistent with the prolongation of complete myelination into adulthood⁵⁹. The trend for increased white matter volume was especially prominent in the methamphetamine group, providing a reduced group difference for total volume. The authors speculated that age-related loss of cortical gray matter in stimulant abusers might be associated with reduced capacity to experience euphoriant effects of psychostimulants such as cocaine⁶⁰, the only group to show age-related atrophic reductions. They suggested that this phenomenon might explain the well-established age-related reduction in psychostimulant addiction. However, no interpretation was made for finding stronger temporal than frontal effects. In addition, volumes of only parts of the frontal and temporal lobes were quantified.

In 2004, high-resolution surface-based computational image analysis of structural MRI (performed at 3 T) was used to map regional abnormalities in the cortex, hippocampus, white matter, and ventricles of 22 chronic methamphetamine abusers (mean use = 10.5 years) who used methamphetamine on most of the 30 days prior to entering the study, as compared with 21 age-matched control subjects who did not use amphetamines⁶¹. The methamphetamine abusers had less gray matter, averaging 11.3% below levels found in comparison subjects in the cingulate, limbic, and paralimbic cortices. The abusers displayed 7% white-matter hypertrophy and 7.8% smaller hippocampal volumes. Remarkable MRI-assessed gray matter deficits in regions of the anterior and posterior cingulate cortex of the right hemisphere overlapped regions of abnormal glucose metabolism in another sample of methamphetamine abusers that contained some of the same subjects⁴⁶ (see Fig. 1). During a vigilance task, these subjects exhibited high glucose metabolism in posterior cingulate and low glucose metabolism in anterior cingulate. Hippocampal volumes, which displayed deficits in the methamphetamine abuse group, were correlated with performance on a word-recall test. The overall results were consistent with the conclusion that chronic methamphetamine abuse may produce or be associated with a selective pattern of cerebral deterioration, with prominent effects in the medial temporal lobe and limbic cortices. These data are consistent with earlier suggestions of temporal lobe deficits⁵⁸ and hippocampal deficits that were postulated to contribute to impaired memory performance in methamphetamine abusers. The authors speculated that neuronal deficits and or toxicities might lead to the white-matter hypertrophy noted in both studies through developmental differences, neuroadaptation, neuropil reduction, cell death, adaptive glial proliferation or altered myelination⁶¹.

Figure 1.

Gray matter differences and abnormalities in glucose metabolism on the medial surface of the brain (see ⁴⁶,⁶¹). Group difference map (A) shows mean percentage differences in gray matter volumes in the methamphetamine (MA) group compared with the control (CTL) group, according to the color bar, and the significance of these differences (B) plotted as a map of p values. The cingulate gyrus shows gray matter deficits (red; p = 0.034, corrected), whereas other brain regions are comparatively spared (blue/green). Illustrations (C) and (D) show the locations of differences in regional cerebral glucose metabolism (relative values) assessed with PET in MA-dependent (n = 17) compared to control subjects (n = 18). The samples that produced the MRI data in (A) and (B) partially overlapped with those that produced the data in (C) and (D). In (C) and (D), statistical parametric maps reveal regions in which the MA group had higher (red) or lower (blue) relative values of glucose metabolism. Colors superimposed on a gray-scale MRI template indicate voxels where the t-test for group difference exceeds t > 1.69 (P < 0.049). A region of remarkable gray matter deficit (B) in the right hemisphere posterior cingulate cortex (pCING) also showed an apparent increase in glucose metabolism in the methamphetamine group (D). The abbreviations igACC and pgACC denote the inferior and perigenual anterior cingulate cortex, respectively.

A single study assessed effects of prenatal methamphetamine exposure on brain structure and cognitive performance⁶². Volumes for whole brains, cerebellum, thalamus and midbrain were similar to those of controls. However, thirteen children (mean age 6.9 years, 4 male) who were exposed to methamphetamine in utero displayed bilaterally smaller volumes of putamen (17.7%), globus pallidus (28.5%), hippocampus (19.5%), and caudate (13%), as compared to non-exposed children. Moreover, subcortical volumes in hippocampus, putamen and globus pallidus were correlated with deficits in sustained attention and delayed verbal memory, suggesting that methamphetamine exposure during gestation might be neurotoxic to these parts of the developing brain and/or that the predisposition to methamphetamine abuse inherited from the parents might also provide brain substrates for attention and mnemonic deficits.

In contrast to the smaller striatal structures in methamphetamine-exposed children, 50 adults (24 male) who were abstinent from chronic (duration > 2 years) methamphetamine use for an average of four months had larger volumes of the putamen (10%), and the globus pallidus (8%) than healthy comparison subjects (see Figure 2, reproduced from⁶³). Female methamphetamine users also had 9.7% larger volumes of the mid-posterior corpus callosum compared to control subjects. Because those methamphetamine abusers with smaller striata had greater lifetime methamphetamine use and more impaired cognitive performance on verbal fluency and speeded motor tasks, the authors proposed that enlargement of the striatum may reflect a compensatory response to methamphetamine toxicity. They propose that the adaptation fails to maintain both function and structural integrity of the striatum after prolonged abuse. Enlargement of the striatum could be produced through inflammation and reactive gliosis, possibly abetted by glia-mediated neurotrophic effects to increase striatal sprouting, as has been shown in response to dopaminergic lesions in experimental animals⁶⁴. The larger volume of the mid-posterior corpus callosum in methamphetamine-abusing women may be related to the increased glucose metabolism and perfusion that has been shown to be maximal in the parietal lobes of abstinent methamphetamine abusers^47,48,65, since the parietal fibers cross in this portion of the corpus callosum.

Figure 2.

Left: Axial MRI slices showing measured brain regions. CA - caudate, PU - putamen, GP - globus pallidus, TH - thalamus, CB - cerebellum, CV - cerebellar vermis, MB - midbrain. Right: Bar graphs showing larger volumes of lentiform nuclei (putamen and globus pallidus) in methamphetamine (MA) abusers compared to healthy volunteers. Putamen MA effects: right – F_1,96 = 11.74, p = .0009; left – F_1,96 = 12.55, p = .0006. Globus Pallidus MA effects: right – F_1,96 = 6.22, p = .01; left – F_1,96 = 10.32, p = .002. *= p<.05; **= p<.005; MRI, magnetic resonance imaging ⁶³.

Structural abnormalities in the corpus callosum were also observed in methamphetamine abusers who were abstinent for a longer period of time (mean = 21 months)⁶⁶. Automated shape analysis indicated greater curvature of the genu and smaller width in both the posterior midbody and isthmus of the corpus callosum of 27 methamphetamine abusers (24 male) relative to comparison subjects. Moreover, both the genu and posterior midbody abnormalities were correlated with lifetime dose of methamphetamine. There were no group differences in total volume of the corpus callosum or of the subregional areas defined by Witelson⁶⁷. Since 89% of the sample was composed of men, this observation does not directly conflict with the earlier reports of larger posterior callosal volume in female methamphetamine abusers ⁶³. The authors suggested that the anterior and posterior shape differences in the interhemispheric white matter tracts may be related to the functional abnormalities that had been observed in frontal and parietal cortices of methamphetamine abusers.

Abnormally large volumes of the striata of former methamphetamine abusers was confirmed in a study that compared healthy individuals to subjects who were HIV+, chronic methamphetamine abusers (abstinent for an average of three months), or both⁶⁸. HIV+ status without methamphetamine abuse was associated with smaller volumes than healthy control subjects in cortical and subcortical structures, including the hippocampus. In contrast, methamphetamine dependence in HIV− individuals was associated with larger volumes than control subjects within the caudate nucleus, lenticular nucleus and nucleus accumbens. Volumetric differences in the nucleus accumbens were larger for younger individuals. Because the age at which subjects were studied was highly correlated with the age of the first methamphetamine abuse, it is unclear which factor was most important. Since greater plasticity is associated with younger as compared to older adults, either finding is consistent with interpreting enlargement of the striatum as an adaptive response to amphetamine-mediated toxicity, or even as a predisposing factor to amphetamine abuse.

Although measures of neurocognitive impairment were not generally correlated with subcortical volumes, in the HIV+ methamphetamine abusers, smaller hippocampal volume was associated with more cognitive impairment⁶⁸. This relationship suggests that if the lower hippocampal volumes associated with being HIV+ were caused by the HIV virus, that the functional consequences of hippocampal reduction might be exacerbated by methamphetamine abuse. Cortical volume was lower in HIV+ individuals who did not use methamphetamine, but tended to be larger in HIV- methamphetamine abusers, attaining significance in the parietal lobe. Moreover, the opposite cortical abnormalities in methamphetamine abusers and HIV+ individuals were both associated with neurocognitive impairment. This makes it less likely that the larger parietal volume in methamphetamine abusers represents an adaptive compensatory response to amphetamine-mediated toxicity, as suggested for the striatal volume increase.

Bae and associates⁶⁹ assessed the prevalence, severity and location of T2-weighted MRI white matter signal hyperintensities (WMH) in 33 methamphetamine abusers who were abstinent for an average of 18 months. WMH represent patchy or diffuse changes associated with structural abnormalities that included dilated perivascular spaces, demyelination, astrocytic gliosis and arteriosclerosis⁷⁰. Such findings have also been found more frequently in both cocaine abusers and opiate abusers relative to control subjects^71-73. Methamphetamine abusers had greater prevalence of WMH (33%) than control subjects (3%), and greater severity of both periventricular and deep WMH, primarily in the frontal lobes. Unlike cocaine addicts⁷¹, methamphetamine abusers showed no difference in the insular region. The severity of deep WMH was correlated both with higher lifetime dose and the average daily dose of methamphetamine during abuse, but was not related to the duration of abstinence. Male methamphetamine abusers had more prevalent WMH than female abusers. While male methamphetamine abusers also had greater severity of WMH than healthy males (odds ratio = 18.9), female methamphetamine abusers did not have significantly greater severity of WMH than healthy females (odds ratio = 1.2).

One study used both semi-automated segmentation and a stereological method to assess MRI volume abnormalities in 16 males currently abusing drugs through intravenous administration. The majority (9) abused methamphetamine along with other drugs⁷⁴. Intravenous drugs had been abused for between seven and twelve years. The specificity of structural abnormalities for amphetamine abuse is unclear because all intravenous drug users, but none of the control group, ingested cannabis. In addition, most drug abusers self-administered cocaine and heroin as well as methamphetamine by intravenous routes. Although these potential confounding factors complicate interpretation, the two volume assessment methods did produce highly correlated estimates (r=0.65, p<0.001). Both methods found that substance abusers had a significantly lower proportion of white matter in the frontal lobe than control subjects. In contrast, there were no group differences in the proportion of whole brain white or gray matter, frontal gray matter, or lateral ventricular volume.

Although an abnormally low proportion of frontal white matter was reported by analyses using both methods, this result is inconsistent with other studies that measured cortical gray and white matter volumes (Table 1). These contrasting reports document either no difference in white matter volumes, increased white matter volumes, or decreased gray matter volumes. The abnormally low proportion of frontal white matter may be associated with features specific to this study, such as the methods used to determine the proportion of white matter, or the confounding effects of other addictive substances. Abnormally small white matter volumes have been associated with alcohol abuse⁷⁵. Cocaine abuse has been associated with impaired myelination^76,77. While no subjects reported histories of or current alcohol abuse, 88% of the drug-abusing sample reported intravenous use of cocaine, as compared to the 56% who reported methamphetamine abuse. Therefore, differences from the control group are at least as likely to be related to use of cocaine as to use of methamphetamine.

Voxel-based morphometry (VBM) is a whole-brain method for assessing volume differences between tissue segmented brain images which has been developed and popularized within the Statistical Parametric Mapping (SPM) software package^78,79. VBM was used to compare gray matter and white matter between healthy control subjects and former methamphetamine users who were abstinent for either less than six months (n=11) or more (n=18) than six months⁸⁰. Lower “gray matter density” was reported in the right middle frontal gyrus (BA 10) of methamphetamine abusers, as compared to control subjects. Methamphetamine abusers also scored lower on a neurospsychological test of executive functioning, the Wisconsin Card Sort Test (WCST). Total WCST errors of methamphetamine abusers, but not control subjects, were correlated with the right middle frontal gray matter density within the area that provided the largest group difference. Both the gray matter and WCST scores were closer to control values in methamphetamine abusers who reported long-term abstinence, relative to those who reported only short-term abstinence. There were no significant white- matter abnormalities. Results could not be separately assessed for women.

A recent report of structural brain abnormalities in methamphetamine abusers came from authors who used diffusion tensor imaging (DTI) to assess white matter integrity⁸¹. DTI measures changes in microstructural white matter organization by providing measures of functional anisotropy (FA) and diffusivity of water flow. These measures are sensitive to disorganization and damage to axons and their myelin sheaths^82-83. Decreased white-matter integrity was inferred from abnormally low FA in bilateral frontal white matter voxels at the anterior commissure-posterior commissure (AC-PC) plane and in right hemisphere prefrontal white matter 5 mm above the AC-PC plane in 32 abstinent (>1 month) methamphetamine abusers, as compared to 30 healthy comparison subjects. Frontal executive functions, as assessed by WCST total errors, were also impaired in methamphetamine abusers. WCST errors correlated with FA values in right prefrontal white matter 5 mm above the AC-PC plane. Separate gender analyses revealed that both the FA and WCST abnormalities associated with methamphetamine abuse were significant only in the male, but not in the female subjects, suggesting the possibility for a protective factor in female methamphetamine abusers. Whether such differences represent a predisposition to addiction or are secondary to stimulant toxicity needs to be clarified through prospective study designs.

Structural brain abnormalities and MDMA

Three imaging studies assessed brain structure in MDMA abusers. The first one used VBM to compare 31 polydrug users who took MDMA (26% 5-10 times, 48% 11-39x, 26% > 40x), with 29 polydrug users who had never taken MDMA⁸⁴. No subject had taken MDMA for at least three weeks prior to testing. In cerebellum, occipital cortex (BA 18), left hemisphere temporal (BA21) and frontal cortices (BA45), and medial brainstem, MDMA users had lower gray matter concentration than polydrug users who had not used MDMA. There were no group differences in white matter concentration, and no differences between men and women. The specificity of the gray matter findings for MDMA is compromised because MDMA abusers who self-reported polydrug abuse also reported more severe substance abuse than polydrug users who reported never using MDMA. They thus reported significantly more episodes of use of seven additional classes of abused drugs (hallucinogens, cocaine, cannabis, opiates, PCP, sedatives, non-MDMA amphetamines). Use of non-MDMA amphetamines was reported by 55% of MDMA users but only 10% of comparison subjects. The authors reported similar VBM results when they used a less conservative statistical threshold after removal of covariation associated with individual drug categories, but they make no attempt to remove effects of overall severity of drug use. It would not be surprising, for example, if abuse of all sympathomimetic stimulants produced some effects similar to those produced by MDMA. Separate removal of covariance associated with lifetime incidence of cocaine use and non-MDMA amphetamine abuse could not rule out the possibility that use of the other sympathomimetic drugs, rather that MDMA use per se, contributed to group differences in the morphometric measures obtained in this study.

A DTI study compared healthy control subjects to 12 chronic MDMA abusers. MDMA abusers had consumed MDMA an average of 181 times. All had consumed it at least 15 times. These subjects had been abstinent from MDMA between 0 (two subjects were urine +) and 101 days (mean = 40 days). The MDMA abusers did not significantly differ from 20 comparison subjects in FA, mean diffusivity or transverse diffusivity in any of six corpus callosum sub-regions. However, longitudinal diffusivity, the diffusion in the direction of the white-matter fibers, was lower in MDMA abusers within the rostral body of the corpus callosum, suggesting the presence of damage to or differences in anterior callosal white matter ⁸⁵. Moreover, rostral body longitudinal diffusivity was correlated with the proportion of advantageous choices on a measure of effective decision making, the Iowa Gambling Task, in the subset of 11 MDMA users and 15 control subjects who completed this task. Although this relationship suggests that MDMA-mediated damage to or differences in rostral body axons may impair some forms of judgment, as with the VBM study⁸⁴, greater use of other drugs by the MDMA abusers provided a potential confounding influence. Because MDMA use is associated with a lifestyle that often includes polysubstance abuse, it is difficult to control for use of all other abused substances.

Beyond controlling for use of other drugs, it is of importance to determine if structural brain differences in those who have used MDMA, or have abused other substances, are truly consequences of their drug use, or antecedents that might predispose to such use. As pointed out above, it is possible that some abnormalities represent preselection or biological vulnerability markers that predate actual drug taking. Because MDMA is occasionally administered to human beings in research studies, and has been used therapeutically as a psychotherapy adjunct in disorders characterized by anxiety^86-88, including post traumatic stress disorder and terminal cancer⁸⁹, it is also important for ethical decision making to determine the possible negative consequences, if any, of low doses of MDMA.

A recent paper by investigators in the Netherlands demonstrates the advantages of a prospective study design in addressing these issues⁹⁰. MDMA-naïve young adults who reported a high probability they would soon start using the drug, or reported that their friends already used the drug, were assessed with proton magnetic resonance spectroscopy, perfusion-weighted imaging (PWI), DTI, and psychological questionnaires. The first thirty subjects to report MDMA use were reassessed between 0.9 and 29.5 months after the initial assessment (mean 8.1 +/- 6.5 months). Most had used MDMA a single time (mean 1.8 +/- 1.3 tablets), with last use at least 3 weeks (mean 7.7 +/- 4.4 weeks) prior to retesting. The only other drug to increase in frequency of use between assessments was cocaine. Separate analyses excluded the four subjects who increased their cocaine use. PWI revealed a significant reduction in the regional relative cerebral blood volume in dorsolateral frontal grey matter after MDMA use. No differences between sessions in the spectroscopy or DTI measures survived correction for multiple comparisons. However, a marginal increase in FA within the white matter centrum semiovale was noted after MDMA use. This increase was correlated with total amount of ecstasy tablets ingested. Since reduced rather than increased functional anisotropy is thought to be associated with axon damage, the authors concluded that while the frontal grey matter finding suggested the possibility of sustained local vasoconstriction, there were no indications of structural damage to axons or neurons after initial low doses of MDMA.

Discussion: gray matter and white matter

Gray matter abnormalities

The most consistently reported specific structural abnormality in amphetamine abusers is lower cortical gray matter density or volume^58,61,80,84. Reduced gray matter has been reported within all of the cortical lobes from at least one study: temporal^58,61,84, frontal^61,80,84, occipital^61,84 and parietal⁶¹. Only one study reported an increase in cortical gray matter volume, and this was localized to the parietal lobe⁶⁸. Increased parietal cortex gray matter in this study was associated with cognitive impairment, suggesting that might represent a measure of toxicity.

In contrast, both studies that assessed striatal grey matter in adult amphetamine abusers reported larger striatal volumes when compared to comparison subjects^63,68. A study of children exposed to methamphetamine in utero reported striatal volumes that were 13-30% smaller than healthy children, data that was interpreted as evidence of prenatal dopaminergic neurotoxicity⁶². In adults, association of larger volumes of the striatum with better cognitive performance, lower lifetime methamphetamine use⁶³, and with greater neuroplasticity⁶⁸ suggested that increasing the volume of the striatum may be a compensatory response to initial neurotoxicity which might fail above a cumulative neurotoxic load.

However, similar enlargement of the striatum induced by neuroleptic treatment of schizophrenic patients is generally considered a sign of toxicity. The classic neuroleptics that are most strongly linked to enlargement of the striatum are also associated with the choreoathetoid movements of tardive dyskinesia, which are similar to choreoathetoid movements reported in amphetamine abusers⁹¹. Choreoathetoid movement disorders are thought to result from a high ratio of dopamine to acetylcholine activity in the basal ganglia, which by inhibiting GABAergic influences on thalamic neurons, leads to increased glutamate-mediated excitation of neocortex. The second generation “atypical” neuroleptics, which produce less impact on the dopaminergic system, have a much lower incidence, and may even reverse both tardive dyskinesia and enlargement of the striatum⁹². Therefore, the larger basal ganglia volume in methamphetamine abusers might be construed as converging evidence that direct cathecholamine effectors have a high potential for toxicity.

One possible mechanism for such toxicity is the increase in free radical production during the metabolism of catecholamines. This process may be exacerbated by the high iron levels that are present in the basal ganglia, since iron is a well known catalyst of free radical reactions. A recent primate study showed that methamphetamine use increases iron in brain, which could further exacerbate such toxicity¹⁸. Antipsychotic medications have also been shown to influence iron levels⁹³ and to influence iron-related toxicity⁹⁴ in rodent models. Thus, pharmacologic disturbances of catecholamine metabolism, especially dopamine metabolism, may be a generalized way of disturbing iron metabolism, and thus producing toxicity in the basal ganglia⁹⁵. Recent demonstration of higher iron levels in males than females⁹⁶ suggests this mechanism may increase the vulnerability of males to stimulant neurotoxicity, consistent with two reports of greater white-matter abnormalities in male as compared with female methamphetamine users^69,81.

Can the idea that amphetamines damage the dopamine-rich basal ganglia through iron-related toxicity or a similar process be reconciled with the evidence suggesting that striatal enlargement in methamphetamine abusers can be a compensatory adaptation? Although there is evidence that several mechanisms contribute to the induction of tardive dyskinesia by typical neuroleptics, the most prominent theory implicates postsynaptic dopamine receptor hypersensitivity⁹². We propose that dopamine hypersensitivity may be an adaptive response to reduction of dopamine activity in the striatum mediated either through chronic dopaminergic blockade or through amphetamine-mediated loss of dopaminergic terminals. Although typical neuroleptics have been largely replaced in order to reduce the risk for development of tardive dyskinesia early in the disease process, their efficacy is comparable to any of the newer antipsychotic medications. A recent study of psychotic symptomatology and treatment-associated changes in the basal ganglia revealed significant volume increases in caudate and putamen after only four weeks of antipsychotic treatment. The increases in left striatum were not associated with drug treatment per se, but were associated with the reduction of positive symptoms⁹⁷. Prenatal exposure to methamphetamine was associated with striatal deficits in children which were correlated with cognitive deficits⁶², suggesting that the mechanism that increases local volume to compensate for striatal damage is not available in utero.

White matter abnormalities

Overall, white matter abnormalities in amphetamine abusers have now been reported more often than gray matter abnormalities. Of the 13 studies in Table 1, 11 assessed white matter and 7 reported at least one significant white matter abnormality. In addition, among the four papers that did not describe white matter structural abnormalities, one reported that findings of low gray matter in methamphetamine abusers were invariably accompanied by trends for congruent local increases in white matter⁵⁸. Another reported that initial use of MDMA produced trends for increased FA in frontoparietal white matter that were significantly correlated with the number of MDMA tablets consumed⁹⁰. The specific form of white matter abnormality associated with amphetamine use is less clear, since increased volume^61,63, decreased volume^66,74, increased T2-hyperintensities⁶⁹, and reduced DTI measures of normal white matter diffusion^81,85 have all been reported. Gray matter reductions and white matter expansion are seen in healthy individuals through middle age⁵⁹. Deviations from these normal developmental trajectories may result from use of psychostimulants as has been suggested in cocaine dependence⁷⁷. In the dynamic background of healthy adult brain developmental trajectories, the timing of MRI studies in relation to abuse onset and age of subjects may account for some of the discrepancies.

Limitations and possible reasons for inconsistency

Multiple interpretations of structural abnormality: vulnerability factors, signs of damage, or compensatory responses?

Extant studies of the effects of amphetamines on brain structure suffer from several important limitations. Most studies employ a cross-sectional one-assessment design, which cannot determine whether volume differences in amphetamine abusers predated drug abuse (e.g., are risk or protective factors for developing addiction), directly represent brain damage, or are compensatory responses to such damage. Is having less frontal lobe grey matter a risk factor for amphetamine abuse? Do large brains indicate greater tissue resources against toxicity, thereby conferring reduced vulnerability to damage from amphetamine exposure? Investigations employing prospective longitudinal designs such as the Netherlands XTC study⁹⁰ are sorely needed to determine if abnormalities in the brain structure of amphetamine abusers were actually caused by amphetamines at all.

Determination of the relationship between structural abnormalities and multiple indices of both the severity of abuse and of cognitive performance are very helpful in characterizing the functional significance of structural abnormalities, but they have been unevenly employed. For example, amphetamine abusers have exhibited deficits in tests of executive cognitive functions similar to those exhibited by patients with frontal damage⁹⁸ (see also review by Robbins and colleagues, this volume). Table 1 shows that worse performance on the WCST was associated with evidence of frontal structural abnormality in methamphetamine abusers in both white⁸¹ and gray matter⁸⁰, and that a frontal white matter abnormality in MDMA abusers was associated with decreased performance on the Iowa Gambling Task⁸⁵. However, four other studies in which amphetamine abusers exhibited structural abnormalities in the frontal lobes did not measure executive functions^66,69,74,84.

Along with exploration of risk factors for unwanted consequences of stimulant abuse, it is equally important to identify potential protective factors. For example, it has been reported that both lithium and valproate treatment protect against amphetamine-induced alterations of brain choline concentrations noted in bipolar disorder patients⁹⁹. Recent animal studies have produced evidence for neuroprotection against amphetamine-mediated toxicity by several substances, including nomifensine¹⁰⁰, methyllycaconitine¹⁰¹, coenzyme Q10¹⁰², baicalein¹⁰³ and melatonin¹⁰⁴. Impairment of learned place preference consolidation by amphetamine-induced neurotoxicity was ameliorated by administration of a glutathione precursor¹⁰⁵. Evidence suggests that lower brain volume is a predisposing factor for developing alcohol dependence¹⁰⁶. Larger premorbid brain size may protect against cocaine-related brain damage¹⁰⁷. The mechanisms proposed for cocaine neurotoxicity are similar to those proposed for amphetamine toxicity. This suggests that having a larger brain may increase a general brain reserve for coping with neurotoxic influences. It is highly likely that both risk and neuroprotective effects are genetically modulated. Charting the relationship of amphetamine use to both a priori genetic polymorphisms and to a posteriori brain structural, brain functional, and behavioral consequences of chronic amphetamine use will lead to better understanding and treatment of amphetamine-mediated brain damage.

Developmental timecourses

During the years of greatest risk for addiction (teens to middle age) there are normal changes in gray and white matter volumes (increases in white and decreases in gray)⁵⁹ which may interact with effects of amphetamine. At the far end of the lifespan, the hypothesis of amphetamine-mediated acceleration of aging in older adults could be explored by assessing chronic amphetamine abusers using multimodal functional and structural neuroimaging indices of cerebral health during normal aging. Such indices have undergone extensive development in recent years¹⁰⁸. Cross-sectional designs cannot unambiguously characterize the influence of amphetamine abuse on normal developmental trajectories. Longitudinal assessment of amphetamine-mediated changes could provide more informative assessment of drug influences on brain structure⁵⁸.

Sex differences

The importance of subject gender on amphetamine-associated structural brain abnormalities has not been well studied. Seven studies in Table 1 tested more than 4 amphetamine abusers of both sexes but only 5 reported explicit assessment of sex effects and only 2 tested more than 12 subjects of both sexes. Two studies of MDMA abusers found no sex differences^84,90. One study reported increased volume of the posterior midbody of the corpus callosum in female, but not male methamphetamine abusers⁶³. Two other studies of methamphetamine abusers reported white matter abnormalities in men but not in women^69,81. It has been hypothesized that since aromatase, the enzyme that produces estrogen in astrocytes, is involved in glial repair after brain injury, estrogen might have neuroprotective effects⁶⁵. In addition, female sex has been described to be “promyelinating”, and may accelerate brain myelination. Such acceleration could potentially produce different developmental timecourses of vulnerability to toxicity in men and women. These effects could explain some of the male preponderance in the epidemiology of addiction and other developmental disorders that are often comorbid with addiction^109,110.

Possible nonspecificity for amphetamines

Many amphetamine abusers, particularly MDMA abusers, are polydrug abusers, and are particularly prone to additional abuse of nonamphetamine stimulants such as cocaine. In many studies reporting structural brain abnormalities in amphetamine abusers, effects of exposure to a specific amphetamine are confounded with greater use of other abused substances, classes of abused substances (i.e. CNS stimulants), or the overall severity of substance abuse, even in studies where attempts were made to compare polydrug abusers who abuse amphetamines with polydrug abusers who do not. Cognitive impairment may also be associated with structural abnormalities. Such impairment in MDMA abusers has been more closely related to use of cannabis than to use of MDMA¹¹¹. Amphetamine abusers may be predisposed to abuse drugs by differences in personality or early experiences, which could also have brain structural correlates. Other potential confounds derive from lifestyle related effects, such as the circadian disruption, extended aerobic exercise, and high-volume music exposure¹¹² that have been associated with MDMA use within the “rave” dance culture. Differences in nutritional practice could predate or be caused by amphetamine abuse.

Severity of methamphetamine abuse

Both within and across studies, the amphetamine abusers who have been assessed for structural brain abnormality vary widely in lifetime exposure to amphetamines. It can be difficult to validate exposure history variables, since almost all studies use self-report, and accept retrospective ratings over many years from subjects who may vary in the intactness of memory systems (see review article by Robbins and colleagues, this volume). Exposure is usually reported as the number of exposures or the cumulative grams of drug ingested. Both measures are imprecise in most patients due to the variable purity of nonpharmaceutical amphetamines. The duration of abuse is another important variable. Excluding studies of initial use, studies in Table 1 include minimal durations of abuse that range from 1 to 7 years, and average cumulative total use of methamphetamine that ranges from 276 to 4930g. All of these problems compromise attempts to relate severity of abuse or drug exposure to structural abnormalities. Future studies should report at least the range, a measure of central tendency, and a measure of dispersion for the cumulative dose and duration of amphetamine abuse. Studies should obtain external validation of histories from medical records, family reports, or other sources whenever possible.

Duration of abstinence

Postmortem studies of animals and humans suggest that the primary dopaminergic damage produced by amphetamines involves terminals and processes rather than cell bodies. Previous studies of functional or metabolic measures assessed during prolonged abstinence from prior chronic amphetamine abuse have reported evidence for partial abstinence-related normalization of low cingulate cortex perfusion⁵⁰ and striatal dopamine transporters in methamphetamine abusers⁵¹, and low 5-HT transporter binding in MDMA abusers (summarized in⁸⁸). There are substantial differences in the duration of time from participants’ last amphetamine exposure to their structural brain imaging between studies and among individuals in each study. Mean abstinence in Table 1 ranges from about 7 to 730 days. Only three studies explore these effects of abstinence on structural abnormalities in amphetamine abusers. Although assessment of abstinence was not a focus of a study of the effects of methamphetamine and subject gender on subcortical volume measures, the authors note that the primary effect of methamphetamine in their study, larger striatal volumes, was uncorrelated with duration of abstinence⁶³. White matter abnormalities in T2-signal hyperintensities of methamphetamine abusers were also unrelated to duration of abstinence⁶⁹. In contrast, a VBM study reported that low frontal lobe gray matter density and associated cognitive deficits in methamphetamine abusers were both ameliorated after at least 6 months of abstinence⁸⁰.

White matter density was not related to duration of abstinence. Frontal gray matter changes produced by methamphetamine may thus be more reversible during abstinence than white matter changes. This idea is consistent with results of a recent proton magnetic resonance spectroscopy study in which cumulative lifetime exposure to methamphetamine correlated inversely with frontal concentrations of N-acetyl-aspartate in both white and gray matter. These data, suggest dose-dependent damage, with only the frontal gray matter concentration providing correlations with the duration of abstinence¹¹³. The other studies that did not find effects of abstinence did not assess frontal gray matter. The study that examined white matter abnormalities alone found no effects of abstinence using a distribution (18 +/- 29 months)⁶⁹, which was similar to the VBM study (20 +/- 34 months), although the study of subcortical gray matter used more recently abstinent methamphetamine abusers (4 +/- 6 months)⁶³. This would be expected to decrease the power to detect normalization during longer durations of abstinence. More research, preferably with longitudinal study designs, is needed to characterize both “dose-response” relationships between chronic exposure to amphetamine and the development of structural brain abnormalities, and time dependent effects on structural normalization during protracted periods of abstinence.

Different techniques

Imaging techniques for quantifying brain macro- or microstructure of amphetamine abusers include segmentation, manual tracing/drawing methods, pattern matching, MRI hyperintensity rating, DTI, and VBM. These techniques have different strengths and limitations. Gray matter reductions can be accompanied by white matter volume increases. Such changes could cancel each other out in unsegmented volume measurements. Pattern matching methods can be made sensitive to shape changes that would also be undetected by drawing methods that assess overall volume. Region of interest and automated voxel based DTI assessment of white matter microstructure¹¹⁴ can speak to changes in connectivity between brain regions, but yield little signal in gray matter regions.

VBM is enjoying increasing popularity because of the ease of comprehensive automated voxelwise whole-brain analysis, relative to the more labor-intensive techniques that involve drawing regions or structures onto individual brains in order to make volume measurements. However, the reliability and generalizability of VBM results remain controversial^115-117. While some studies have validated the technique against manual tracing methods^79,118-121, other studies have reported somewhat discordant results^122-124. Measurement of total cerebrospinal fluid volume obtained with VBM are higher than those obtained with manual tracing methods¹²⁵. Since the operational definition for gray matter density has not been resolved, it may be that different types of VBM measure very different things (i.e. volume vs. the statistical probability of a voxel being gray matter). Although the underlying statistical assumptions are reasonably conservative^126,127, proper use of VBM requires proper preprocessing^128,129, and the selected smoothing kernel may need to differ for optimal VBM of gray and white matter^119,130. Understanding of the proper uses and limitations of VBM are still developing. A recent editorial summarizes core principles for conducting and reporting studies in ways that make them most useful and interpretable in the context of other techniques for structural image analysis¹³¹.

Finally, a myriad of technical issues are involved in the quantification of gray and white matter. These include differences in gray/white contrast produced by different MRI sequences and differences due to alignment of images¹³². As high-field strength instruments are becoming more prevalent, contrast differences caused by differences in the field strengths of MRI scanners may also contribute to inconsistent results. Such differences can result from a combination of changes in T1 relaxation-related contrast that are known to occur at different field strengths, and increased sensitivity to the effects of tissue iron on T2 relaxation-related contrast at higher field strengths. Thus, different field strengths may change the image contrast in ways that “shift” the apparent border between gray/white matter and thereby influence volumetric results. This possibility is further exacerbated for the study of structural effects of amphetamine abuse by the recent observation that methamphetamine use can directly influence brain iron levels¹⁸.

Panel A of Figure 3 shows that two groups of methamphetamine abusers and control subjects who were assessed with MRI instruments operating at 1.5 and 3 Tesla show no scanner-related differences in total brain volume. In contrast, Panel B shows a significant difference between the fraction of brain volume comprising gray matter between methamphetamine abusers and healthy control subjects scanned at the lower (1.5 Tesla), but not higher (3.0 Tesla) field strength (Welch two sample t-test p-value = 0.013). Although this observation suggests that group differences may be harder to detect at higher field strengths, there are probably not yet enough studies to assess this idea. Of 6 papers in Table 1 that probed both white and gray matter abnormalities in amphetamine abusers, both of the two studies that used a 3.0 Tesla field and 3 of 4 studies that used a 1.5 Tesla field reported gray matter abnormalities. Half of the studies at both field strengths reported white matter abnormalities.

Figure 3.

Box-and-whisker plots of structural MRI data in two scanners with different field strengths. A) Total brain volume from MRI images acquired in a 3.0 T General Electric (GE) scanner compared to images acquired from a 1.5 T Siemens Sonata scanner. There is no difference between the two samples. B) Differences in whole-brain gray matter fraction between control subjects and methamphetamine abusers scanned in the GE 3.0 T scanner and the Siemens Sonata 1.5 T scanner. There was no significant difference between the control group and methamphetamine group in the GE scanner. However, there was a significant difference between groups scanned in the Siemens scanner (Welch two sample t test p value = 0.013). Dark lines indicate the median for each group, the first and third quartile for each group are indicated by the lower and upper boundaries of the box respectively, and the dashed lines represent the range of values for each group. Outliers are shown as open circles (S. Fears, J. O’Neill, G. Bartzokis, and E. D. London, unpublished).

One approach to maximizing the multiple strengths and minimizing the weaknesses associated with different imaging techniques is to incorporate multiple types of information into one investigation. This was done in a recent study that correlated results of the same MRI dataset of patients with panic disorder analyzed with manual volumetry and with automated VBM¹¹⁹, and in another study that combined univariate and multivariate analysis of VBM and voxel-based relaxometry data to improve identification of the network of brain abnormalities associated with temporal lobe epilepsy¹³⁰. New ways of using high resolution MRI data, such as computation of voxel-based cortical thickness (VBCT) maps¹³³, are developing rapidly. Collecting multimodal imaging datasets and analyzing them using multiple complimentary techniques is becoming increasingly feasible. A single MRI session can incorporate several sequences that assess several tissue parameters (volumes of gray and white matter, T1, T2, DTI, iron measures, etc) and also collect functional information. Exploration of such multimodal datasets will increase our understanding of both the underlying pathophysiology and the techniques themselves. Understanding of the technical aspects of brain imaging tools that we use on clinical patients must progress towards increasing the specificity of measures or combinations of measures to further the twin goals of using brain imaging to inform both medication development and clinical decision making.