Microglia Imaging in Methamphetamine Use Disorder: a Positron Emission Tomography Study with the 18-kDa Translocator Protein Radioligand [F-18]FEPPA

Abstract

Activation of brain microglial cells, microgliosis, has been linked to methamphetamine (MA)-seeking behaviour, suggesting that microglia could be a new therapeutic target for MA use disorder. Animal data show marked brain microglial activation following acute high dose MA, but microglial status in human MA users is uncertain, with one positron emission tomography (PET) investigation reporting massively and globally increased translocator protein 18 kDa (TSPO; [11C](R)-PK11195) binding, a biomarker for microgliosis, in MA users. Our aim was to measure binding of a second generation TSPO radioligand, [F-18]FEPPA, in brain of human chronic MA users. Regional total volume of distribution (V_T) of [F-18]FEPPA was estimated with a two-tissue compartment model with arterial plasma input function for 10 regions of interest in 11 actively-using MA users and 26 controls. A RM-ANOVA corrected for TSPO rs6971 polymorphism was employed to test significance.

There was no main effect of group on [F-18]FEPPA V_T (p=0.81). No significant correlations between [F-18]FEPPA V_T and MA use duration, weekly dosage, blood MA concentrations, regional brain volumes, and self-reported craving were observed.

Our preliminary findings, consistent with our earlier postmortem data, do not suggest substantial brain microgliosis in MA use disorder, but do not rule out microglia as a therapeutic target in MA addiction. Absence of increased [F-18]FEPPA TSPO binding might be related to insufficient MA dose or blunting of microglial response following repeated MA exposure, as suggested by some animal data.

INTRODUCTION

Methamphetamine (MA) is an addictive psychostimulant abused worldwide that has a high treatment relapse rate¹. As current pharmacotherapeutic approaches lack substantial efficacy², there is a need to discover useful drug targets. Emerging preclinical^3–6 and clinical^7,8 findings employing drugs that inhibit microglial activity (e.g., minocycline and ibudilast), now suggest that modulating microglia might attenuate rewarding effects of MA; thereby helping treatment of MA use disorder. Microglia are an innate part of the neuroimmune system, changing from a “resting” to an “active” state in response to brain injury, immune threat (pathogens, endotoxins): the process of microgliosis^9,10.

The question of whether MA actually “damages” the human brain, in addition to leading to a state of compulsive addiction, is debated¹¹; however, our previous postmortem brain finding of markedly increased concentration of reactive lipoperoxidation products (4-hydroxynonenal, malondialdehyde) in MA users suggests that MA might cause some oxidative damage in human brain¹². The strongest evidence linking MA and a brain microgliosis response is provided by animal studies demonstrating microglial activation and inflammasome activation following very high dose-short term MA exposure^13,14 – the explanation given is that microgliosis is an expected response to a neurotoxic action of MA. However, information on brain microglial status in human MA users who used the drug for years is uncertain. In our own postmortem brain studies of chronic MA users, little signs of obvious above-normal brain microgliosis were observed on routine (qualitative) histological analyses and levels of microglial protein markers were normal¹⁵. In another postmortem brain study of human MA users, Kitamura and colleagues¹⁶ demonstrated above normal number of microglial cells stained with one marker (hGLUT5) but not with a second (CR3.43), and with no morphometric evidence of activated microglia. In sharp contradistinction to the negative postmortem findings, the Sekine group¹⁷, which employed an in vivo brain imaging positron emission tomography (PET) approach with the radiotracer [¹¹C] (R)-PK11195 that targets the 18 kDa translocator protein (TSPO) -- a microglial protein marker --- reported a massive (3–15x) global brain increase in TSPO binding in 12 MA users who had been abstinent on average for 1.8 years (range 0.5–4 years). However, conclusions derived from this imaging investigation might be compromised because of use of a first generation TSPO tracer, [¹¹C] (R)-PK11195, reported to have high non-specific binding¹⁸.

Given the above interest in microgliosis and MA addiction, and the literature deficiencies on this topic, our aim was to explore microglia status in actively using individuals with MA use disorder. We used brain PET imaging with [F-18]FEPPA, a second generation PET TSPO radioligand that has superior properties¹⁸ over the [¹¹C] (R)-PK11195 TSPO tracer used in the earlier imaging study¹⁷.

MATERIALS AND METHOD

STUDY PARTICIPANTS

11 MA users and 26 healthy controls provided written informed consent and completed a [F-18]-FEPPA PET scan and one magnetic resonance imaging (MRI) scan as part of a study approved by the Centre for Addiction and Mental Health Research Ethics Board. PET/[F-18]FEPPA data have been previously reported for some of the healthy controls^{9, 19, 20}. All subjects underwent a comprehensive medical and psychiatric screening (Semi-structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (DSM-IV))²¹. Drug history of MA use and other drugs of abuse was verified using scalp hair toxicology (USDTL, Illinois, USA) in MA users and urinalysis in all subjects at screening (LC-MS and EMIT immunoassay; Rapid Response BTNX Inc., Markham, Canada). Saliva samples (Oragene DNA, DNA Genotek Inc., Ottawa, Canada) were obtained to genotype for the TSPO rs6971 polymorphism which affects [F-18]FEPPA binding^18,22. Healthy controls were genotyped for TSPO using blood DNA (as described in Mizrahi et al., 2012)¹⁸. Subjects were eligible for participation if they were 17 years old or older, physically healthy, and had no other current or previous DSM-IV Axis I diagnosis (excluding stimulant use disorder in the MA group, Axis I disorders concurrent with MA use disorder, and nicotine dependence in both groups). MA users were included if they: (1) self-reported use of MA as the primary drug of abuse; (2) met DSM-IV criteria for MA use disorder; (3) tested positive for MA in scalp hair and urine; (4) were genotyped either as high-affinity (HAB) or mixed-affinity (MAB) binders based on the TSPO rs6971 polymorphism which affects tracer binding.

EXPERIMENTAL DESIGN

Urine samples for detecting pregnancy (in female subjects) and for testing drugs of abuse were taken just prior the scan. In MA users, recent alcohol and tobacco use, determined by breath alcohol and expired CO (> 10 ppm) measurements, respectively, were also assessed prior to the PET scan. MA users self-assessed their MA withdrawal and craving symptoms (Amphetamine Withdrawal Questionnaire²³, Amphetamine Selective Severity Assessment Scale²⁴, and Desire for Speed questionnaire²⁵) on PET scan day.

[F-18]FEPPA synthesis produced by reaction of cyclotron produced [F-18]-fluoride with a tosylate precursor has been described in detail in Wilson et al., 2008²⁶. The final formulation of the radiosynthesis of [F-18]FEPPA was sterile, pyrogen-free with a pH of 5–8. The radiochemical purity was greater than 96%.

The PET scanning was done with participants in supine position in a 3D high-resolution research tomograph (HRRT) (CPS/Siemens, Knoxville, TN, USA) which measures radioactivity in 207 1.2mm-thick brain slices. The participant’s head was held in place with a thermoplastic mask to minimize movement. A brief transmission scan was first acquired to correct PET images for attenuation using a single photon point source 137Cs. This was followed by a 120 minute emission scan after an intravenous saline solution of 5.1 ± 0.26 mCi of [F-18]-FEPPA administered as a ~1 min bolus into the antecubital vein followed by ~10 ml of saline. PET images were corrected for attenuation, reconstructed by filtered back projection algorithm using a HANN filter at Nyquist cutoff frequency and reconstructed into a series of 34 time frames (1 frame of variable length, followed by frames comprising 5×30 s, 1X45 s, 2×60 s, 1×90 s, 1×120 s, 1×210 s and 22×300 s).

To generate the input function for the kinetic analysis arterial blood sampling were taken continuously for the first 22.5 min at a rate of 2.5 ml/min using an automatic blood sampling system (ABSS, Model #PBS-101 from Veenstra Instruments, Netherland). In addition, 7 ml manual samples were drawn at 2.5, 7, 12, 15, 20, 30, 45, 60, 90, 120 minutes to determine the metabolization of the radioligand. The fraction of parent radioligand in plasma was determined using Sep-Pak C18 Classic Cartridge and was fitted with a Hill function. The blood curve was divided by the bi exponential fitting the ratio blood to plasma, multiplied by the hill function and corrected for delay and dispersion to generate a parent compound in plasma curve to use as input function for the kinetic analysis (for further details, please see Rusjan et al., 2011²⁷)

Standard spin echo proton-density weight magnetic resonance images (MRI) were obtained in all subjects (Discovery MR750 3T scanner) for region of interest (ROI) delineation. The ROIs were delineated using the in-house image analysis software, ROMI (ROI delineation described in Rusjan et al., 2006²⁸) and fitted to the summed PET image (as described in Rusjan et al., 2011²⁷). ROI were chosen to cover the whole brain and included the cerebellum, hippocampus, caudate, putamen, thalamus as well as (anterior) cingulate, insula, prefrontal, temporal, and occipital cortices.

Time radioactivity curves (TAC) extracted in each ROI and in arterial plasma were fitted using a two-tissue compartment model (2-TCM) (as described in Rusjan et al., 2011²⁷). [F-18]FEPPA distribution volume (V_T), the ratio at equilibrium of the radioligand concentration in tissue to that in plasma, was estimated in each ROI²⁷. Partial volume correction was applied using the Geometric Transfer Matrix which calculates the fraction of true activity from each ROI by estimating transfer coefficients associated with spill-in and spill-out of the point spread function from adjacent ROIs (as described by Rousset et al., 1998²⁹).

TSPO GENOTYPING

Genomic DNA was obtained from peripheral leukocytes, using high salt extraction method from saliva samples. The polymorphism rs6971 was genotyped (as described in Suridjan et al., 2014³⁰). DNA specimens were analyzed for the TSPO genotype rs6971 variant using the manufacturer’s recommended TaqMan SNP genotyping protocol. In a subgroup of healthy controls, TSPO was genotyped using blood samples as described in Mizrahi et al., 2014¹⁸.

METHAMPHETAMINE LEVELS

Blood was collected in EDTA tubes at the beginning of the PET scan and frozen in −80°C freezer. Concentrations of MA and amphetamine (AM) in blood were assessed with High Performance Liquid Chromatography/ Tandem Mass Spectrometry (LC-MS/MS) (NMS lab, Horsham, PA). The detection limit of the test is 5 ng/ml.

STATISTICAL ANALYSES

Statistical analysis was conducted using IBM SPSS Statistics 24 (Armonk, New York, USA). Regional [F-18]FEPPA V_T between control and MA were compared using repeated-measures ANCOVA, with TSPO polymorphism as a covariate [ROI(10) x Group (2) with TSPO genotype]. Sphericity was corrected using the Greenhouse-Geisser method when required. Post-hoc Least Significant Difference pairwise comparisons were used to dissect significant interactions. Two-tailed partial correlations (controlling for TSPO genotype) were used to examine the relationship between [F-18]FEPPA V_T in the ROIs with clinical measures.

RESULTS

The demographics and drug history for participants, MA (n=11) and controls (n=26), are presented in Tables 1 and 2. There were no statistically significant differences in age, race, gender, BMI, and TSPO polymorphism. There were no differences in the amount (mCi) of [F-18]FEPPA injected between groups. Specific activity at time of injection and the mass injected were significantly lower in controls (Table 1).

Table 1.

Subject’s demographic information

		Controls (n=26) Mean ± SD (Range)	MA (n=11) Mean ± SD (Range)	Group Difference p-value
Age (years)		37.9 ± 12.2 (24–58)	40.0 ± 14.7 (21–71)	0.85
Gender (M/F)		15/11 (42% Female)	8/3 (27% Female)	0.48
NIH Race		7 Caucasian 1 Black 12 Asian 2 Hispanic 4 Mixed	6 Caucasian 1 Asian 4 Mixed	0.11
BMI		24.4 ± 3.17 (19–35)	24.3 ± 3.77 (20–33)	0.93
Alcoholic Drinks/ week1		2.70 ± 3.94 (0–12.5)	2.25 ± 3.14 (0–10)	0.76
Daily Smoker		0	8	p<0.01
Cigarettes/day		--	18.3 ± 13.2 (5–25)	--
TSPO Genotype (MABs/HABs)2		8/18 (31% MAB)	5/6 (45% MAB)	0.47
Urine toxicology (Positive Drugs on scan day)		Acetaminophen: 1 Quinine: 1 THC: 1	MA/AM: all Cocaine: 2 Opiates: 2 THC: 4 OTHERS: 1 anti-depressant; 1 anti-psychotic; 1 anti-convulsant; 1 ibuprofen; 2 acetaminophen; 1 opioid antagonist	--
[F-18]FEPPA Tracer	Amount Injected (mCi)	4.90 ± 0.34 (4.2–5.56)	5.10 ± 0.27 (4.68–5.52)	0.09
	Specific Activity at Time of Injection (mCi/μmol)	3681 ± 3446 (950–12740)	1555 ± 960 (448–3667)	p<0.01
	Mass injected (μg)	0.95 ± 0.62 (0.12–2.16)	1.70 ± 1.06 (0.53–4.26)	0.01

¹HC n=14;

²MAB = mixed affinity binders, HAB = high affinity binders

Table 2.

Drug characteristics and history for MA users (n = 11)

Characteristics		Mean ± SD (Range)
Age of onset (years)		28.9 ± 16.5 (17–68)
Duration of use (years)		11.1 ± 9.70 (2–34)
Frequency (days/week)		4.84 ± 2.08 (1–7)
Amount used/week (grams)		1.81 ± 1.83 (0.1–5.3)
Route of MA administration (smoked, IV, oral, nasal)*		9, 1, 2, 2
MA Severity of Dependence Scale		7.27 ± 2.34
Polysubstance (hair)		All MA/AM; 8 cocaine; 5 THC-COOH; 2 opiates; 1 MDMA
# Marijuana users (hair)		5
MA abstinence before PET scan (hours)		39.2 ± 28.8 (14.3–110)
Blood levels of MA/AM on PET day (nmol/ml)		2.00 ± 2.56 (0–5.57)
MA withdrawal:	ASSA (n = 8)	55.8 ± 20.1
	AWQ (n = 8)	22.9 ± 8.00
DSQ	Total	117 ± 33.6
	Reinforcement	22.0 ± 8.78
	Strong desire	26.8 ± 12.4
	Mild desire	16.0 ± 8.07
	Control	7.38 ± 3.34

^*some MA users had mulitple routes of administration

MA = methamphetamine, PET = positron emission tomography, AM = amphetamine, ASSA = Amphetamine Selective Severity Assessment, AWQ = Amphetamine Withdrawal Questionnaire, DSQ = Desire for Speed Questionnaire, AM = amphetamine, THC-COOH = 11-Nor-9-carboxy-Δ⁹-tetrahydrocannabinol, MDMA = 3,4-Methylenedioxymethamphetamine

On average, MA users used MA for ~11 years with the typical amount being ~2 grams a week (shown in Table 2). Forensic hair analysis confirmed current use in MA users (who were using additional drugs, particularly cocaine) (Table 2). Four MA users and 1 control tested positive for THC (metabolites) on the day of the PET scan. Eight MA users were currently daily cigarette smokers whereas there were no current smokers in the control population. Reported average alcoholic drinks a week were not different between groups. At the time of the scan 10 of the 11 MA users had MA present in blood with metabolite AM, as expected, present at a lower concentration in 9 subjects (MA but not AM was detected in one of the 9 subjects). One of the 11 MA users had no detectable MA or AM in blood but did test positive for both in urine at PET scan. The 11 MA users had on average 2 nmol/ml of MA plus AM in their blood at the time of the scan. The ten blood MA-positive subjects had more than two-fold MA vs. AM in blood with the exception of one user who had similar MA and AM concentrations. This suggests that AM was largely metabolism-derived, rather than ingested, in the majority of the subjects^{31, 32}. MA users self-reported moderate withdrawal based on the Amphetamine Selective Severity Assessment²⁴ and the Amphetamine Withdrawal Questionnaire²³.

BRAIN TSPO BINDING: Healthy Control vs. Methamphetamine Users

As expected, the repeated-measures, analysis of co-variance [Group (2) * ROI (10) with TSPO (2)] indicated that there was a main effect of TSPO genotype (F(1,34)=13.6, p=0.001, ANCOVA) suggesting that subjects with the HAB genotype had significantly higher [F-18]FEPPA V_T values in comparison to MAB subjects. There was no significant main effect of Group (F(1, 34)=0.02, p=0.88, ANCOVA); however there was a significant ROI * Group interaction (F(4.25, 144)=3.79, p=0.01). A Least Significant Difference (LSD) pairwise-comparison t-test revealed a non-significant trend for lower [F-18]FEPPA V_T (−24%; p_uncorrected=0.08) in hippocampus of MA users (reported in Table 2 and Figure 1). An ANCOVA with age, smoking status and sex as covariates (in addition to controlling for TSPO genotype) did not have a significant effect on [F-18]FEPPA V_T (age: F(1, 33)=0.05, p=0.83; smoking status: F(1, 33)=0.26, p=0.61; sex: F(1, 33)=0.01, p=0.94). None of the healthy controls reported cigarette use. An ANCOVA controlling for TSPO found no significant difference in the small group of daily smokers (n=8) vs non-smokers (n=3) (F(1,8)=0.20, p=0.67) or in MA users who used cannabis (based on Nor-9-carboxy-Δ9-tetrahydrocannabinol (THC-COOH) in hair n = 5) vs those who did not (n = 6) (F(1, 8)= 0.01; p = 0.94).

Figure 1.

Scatterplot of [F-18]FEPPA V_T for controls (n=26) vs methamphetamine users (n=11) in regions of interest. The non-significant main effect of the Group was (F(1, 34)=0.02, p=0.88) with a significant interaction (ROI * Group) (F(4.25, 144)=3.79, p<0.01)). Post-Hoc pairwise comparisons revealed a marginal decrease in the hippocampal region (−24%; p_uncorrected=0.08). MA: methamphetmine; HC: Healthy Controls; MAB: mixed affinity binders; HAB: high-affinity binders.

We investigated potential groups differences in regional brain volume and found no main effect of group (F(1,35)=1.15, p=0.29) and no ROI * Group interaction (F(2.18,76.4)=1.31, p=0.27). Partial volume correction of regional spread function did not alter the finding of no main effect of group (F(1,34)=0.032, p=0.86); the interaction was however less statistically significant (F(4,136)=2.29, p=0.06).

CLINICAL AND BEHAVIOURAL EXPLORATORY CORRELATIONS

Partial correlations accounting for TSPO polymorphism were conducted to assess the relationship between [F-18]FEPPA V_T and MA use severity (average weekly dosage, frequency of use, years of use, age of first MA use), time since last use, MA and AM on board at scan time, clinical scales measuring craving (Desire for Speed Questionnaire), withdrawal symptoms (Amphetamine Withdrawal Scale, Amphetamine Selective Severity Assessment) and mood (Beck’s Depression Inventory, Snaith Hamilton Pleasure Scale, Marin Apathy Evaluation Scale). No significant correlations were found (10 ROIs, p>0.05) except for a relationship between [F-18]FEPPA V_T and frequency of MA use (days used a week), such that those who had used MA on more days in the week had higher [F-18]FEPPA V_T. This correlation was significant in 9 of the 10 brain regions examined however did not survive correction for multiple comparisons (hippocampus r=0.64, p=0.05, insula r=0.65, p=0.04, occipital r=0.73, p=0.02, cerebellum r=0.70, p=0.02, temporal r=0.66, p=0.04, frontal r=0.73, p=0.02, caudate r=0.69, p=0.03, thalamus r=0.76, p=0.01 and putamen r=0.63, p=0.05). To further analyze this correlation (which was driven by MA users reporting 7-day a week use), the MA subjects were grouped into daily (7 days a week; n=4) vs intermittent users (5 or less days a week; n=7). There was a significant main effect of frequency of MA use on [F-18]FEPPA V_T (F(1,8)=7.80, p=0.02) suggesting that continuous vs intermittent pattern of use was associated with greater [F-18]FEPPA V_T across brain regions.

There were no significant differences between the intermittent and daily MA groups in any demographic aspects (i.e.: age, BMI, TSPO genotype, smoking status, years of MA use, grams a week of MA use, cannabis use) excluding frequency of use in a week.

In MA users who smoked cigarettes (n = 8) we found no correlation between the number of cigarettes smoked and regional [F-18]FEPPA V_T (10 ROIs, p>0.05).

An ANCOVA controlling for [F-18]FEPPA mass injected did not affect group differences in regional [F-18]FEPPA binding (F(1,34)=2.41; p = 0.62). There were no significant correlations between [F-18]FEPPA mass injected and regional [F-18]FEPPA V_T (10 ROIs, p>0.05).

DISCUSSION

We found that brain [F-18]FEPPA V_T in the MA group was “normal” (i.e., not elevated) in all examined brain areas and that binding did not correlate with most demographic and behavioural measures.

Present findings vs. previous human brain microglial marker studies.

Our observation of generally “normal” binding of [F-18]FEPPA to TSPO strikingly differs from the imaging findings of Sekine¹⁷ who reported a 3–15 fold global increase in brain binding of a different TSPO tracer, [11-C] (R)-PK11195, in MA users. Given that we were able to distinguish between TSPO genotypes that influence medium vs. high binding it is likely that we would have been able to detect a TSPO difference given the very large effect size reported by Sekine. The different outcomes could be explained by uncertainty in validity of measurement of TSPO binding by Sekine. Specifically, use of a first generation tracer having low brain penetration, low specific to nonspecific binding ratio, high plasma protein binding and a reference region not devoid of TSPO (see Narendran et al., 2014³³) could have confounded the finding. In contrast, we employed a second generation TSPO tracer having superior binding characteristics and an arterial input function to obtain the PET outcome measures.

Some characteristics of the MA users in the two studies were generally similar (mean age, 40 years present study vs. 32 years, Sekine; mean duration of MA use, 10 vs. 11 years, severity of craving (likely “moderate” in both), but information on estimated typical dose, pattern of use or blood/hair toxicology was not provided in by Sekine. MA users in the Sekine study might have used higher drug doses than those in our study, but Sekine notes that their MA users “had no history of using toxic doses of the drug”. Blood MA levels (the molar sum of MA plus AM) in current study (mean: 2 nmol/ml) are similar to those reported by Melega³¹ in 105 MA users (mean: 2.4 nmol/ml). Our autopsied brain study of MA-related deaths found no biochemical evidence of microglial activation¹⁵, and mean postmortem blood molar concentrations of MA plus AM were much higher (mean: 10.8 nmol/ml)³⁴. A potentially important difference between the two imaging studies however, was in duration of abstinence, which needs to be explored further in follow-up studies.

The abstinence time range in the Sekine investigation was a lengthy six months to four years vs. 14 to 110 hours in our study. As expected from the long MA half-life (approximately 11 hours³⁵), 10 of 11 subjects tested positive for MA in blood (one testing MA and AM -positive only in urine) suggesting that all had some MA present in brain at time of scan. As such the study cannot separate chronic from acute effects of MA. Although microgliosis is generally considered to be an “early” (e.g., days) response to brain insult, conceivably, the Sekine finding of increased [11-C] (R)-PK11195 binding years after last use of MA could be explained by a triggering of a microglial process not present in early abstinence but develops much later and far outlasts presence of the toxin. Our finding, however, of generally normal [F-18]FEPPA binding data in human MA users is consistent with postmortem brain observations of no¹⁵ or no/little³⁶ evidence of increased brain microglial number or activation in MA users who, for the most part, were suspected of dying as a result of MA intoxication and had five or more times higher MA concentrations in blood (see above).

Absence of a brain TSPO binding increase in our investigation might also be explained by co-use of other drugs, (e.g., cocaine, opiates, cannabis in some users) which might have antagonized any MA-induced microgliosis. However, results of a PET study showed no association with cocaine use on brain TSPO binding (using a second generation tracer) in the human³³. Brody and colleagues^37,38 have reported cigarette smoking linked to a modest reduction in TSPO expression and therefore might be a confounding factor, as most MA users of our study used tobacco; however, this finding was not recently replicated by Hillmer et al; (2019)³⁹ who employed, unlike Brody, arterial sampling to determine total distribution volume. Furthermore, we found no significant difference using an ANCOVA between brain TSPO binding in the small MA smoking (n=8) and non-smoking groups (n=3), and no significant correlation between [F-18]FEPPA binding and number of cigarettes typically used. A recent study found elevated [F-18]FEPPA binding in chronic cannabis users⁴⁰; in our small sampled study we do not find difference between MA users who used cannabis vs those who did not.

Present findings vs. animal literature.

Although it is not possible to model perfectly in a preclinical animal study extent and pattern of drug use of the human MA users of our investigation (e.g., subjects having on average ten years of drug use), results of several animal investigations suggest that some microglial activation, as inferred from increased TSPO binding, still occurs after “long-term” daily administration MA exposure in the rodent (40 days⁴¹; four months⁴²) although the possibility of some blunting or impairment in mounting a microglial response cannot be excluded. In this regard, some animal data also demonstrate a blunted microglial response in rodents re-exposed to MA 7–30 days after initial exposure^43,44. In the present study, slightly increased TSPO binding was found in daily MA users compared to those who used intermittently; a nominal difference which needs further investigation.

Hippocampal Finding.

We report, in an exploratory post-hoc analysis, a trend for modestly reduced TSPO binding in hippocampus in our small sampled study. This finding, unlikely to be explained by differences in brain volume, is in line with preclinical literature suggesting that this region may be vulnerable to MA exposure^45,46. Our results in MA users can be compared with those of a recent study finding of selectively decreased hippocampal TSPO / [11-C]PBR28 V_T in recently abstinent patients with alcohol use disorder. The mechanism behind reduced hippocampal TSPO in alcohol and potentially in the MA users in this study is not known but could include actual loss or atrophy of microglia, astrocyte or other cells that express TSPO (stem cells, endothelial cells) or an independent downregulation of the mitochondrial protein TSPO. Further investigations need to be carried out to uncover the mechanism of TSPO loss, if any, in hippocampus.

Limitations.

In addition to the uncertainties mentioned above, there is the generic concern that our prospective power analysis based on the effect size estimated from the Sekine paper was inflated and therefore that our small sampled study was underpowered to detect an effect. Effect sizes detected in current study were small (Cohen’s d: 0.1) for whole brain and moderate (Cohen’s d: 0.4) for hippocampus findings; in this regard the computed required sample size >400 subjects and ~40 subjects respectively were greater than the size of the current study. Furthermore, TSPO is not absolutely specific for the microglial cell and can be found in astrocytes, and endothelial cells^47,48. Although some postmortem and imaging studies do find increased TSPO binding in human brain when microglial activation is expected (e.g., Alzheimer’s disease; multiple system atrophy)⁴⁹ the gold standard for microglial activation remains histopathological assessment using (difficult to obtain) postmortem brain. Our study was not powered to assess effects of biological sex, found to be modestly related to TSPO binding ⁵⁰. The imaging outcome could also have been influenced by small differences in regional brain volumes, although we did not find significant differences in regional grey matter density between healthy controls and MA users and partial volume correction yielded the same main finding. There were differences in [F-18]FEPPA mass injected, however those did not affect the finding. This, together with the generally normal cognitive testing performance (data not shown) of the MA group also provide no evidence of substantial brain injury in the MA users in our study that might have elicited a microglial response.

Conclusion.

Employing a second generation TSPO probe for estimating microglial status, we found no evidence of a massive global increase in brain TSPO binding in chronic MA users as reported in an earlier PET imaging study. The negative findings do not however exclude the possibility that therapies targeting the microglial cell might be useful in treatment of MA addiction. Our observations are preliminary; a larger sample is needed to explore more thoroughly the possible influence of drug use factors including dose, pattern, and abstinence time on microglial status.