MINOCYCLINE DOES NOT AFFECT EXPERIMENTAL PAIN OR ADDICTION-RELATED OUTCOMES IN OPIOID MAINTAINED PATIENTS

Abstract

Rationale:

Minocycline, a tetracycline antibiotic, inhibits activation of microglia. In preclinical studies, minocycline prevented development of opioid tolerance and opioid-induced hyperalgesia (OIH). The goal of this study was to determine if minocycline changes pain threshold and tolerance in individuals with opioid use disorder who are maintained on agonist treatment.

Methods:

In this double-blind, randomized human laboratory study, 20 participants were randomized to either minocycline (200mg/day) or placebo treatment for 15 days. The study had 3 test sessions (Days 1, 8 and 15 of treatment) and one follow-up visit one week after the end of treatment. In each test session, participants were assessed on several subjective and cognitive measures, followed by assessment of pain sensitivity using the Cold Pressor Test (CPT). Daily surveys and cognitive measures using Ecological Momentary Assessment (EMA) were also collected four times a day on Days 8 though 14 of treatment, and proinflammatory serum cytokines were assessed before and on the last day of treatment.

Results:

Minocycline treatment did not change pain threshold or tolerance on the CPT. Similarly, minocycline did not change severity of pain, opioid craving, withdrawal, or serum cytokines. Minocycline treatment increased accuracy on a Go/No-Go task.

Conclusions:

While these findings do not support minocycline’s effects on OIH, minocycline may have a potential use as a cognitive enhancer for individuals with opioid use disorder, a finding that warrants further systematic studies.

Introduction

Opioids have been used to treat acute and chronic pain for over one thousand years (Chapman et al. 2010). In spite of their strong analgesic effects, the efficacy of opioids is limited by several adverse effects including tolerance, dependence, and a paradoxical increase in pain sensitivity, also known as opioid-induced hyperalgesia (OIH) (Ricardo Buenaventura et al. 2008). Tolerance requires the use of higher opioid doses to achieve analgesia, increasing the risk of overdose and opioid dependence. Although OIH is a well-studied phenomenon in rodents, it remains controversial in humans as it is difficult to clinically separate from tolerance or withdrawal-induced hyperalgesia (Arout et al. 2015). Regardless of the exact cause, these adverse effects resulting from long-term opioid use not only limit the clinical utility of opioids, but also increase the risk for relapse to use and the development of an opioid use disorder (OUD) (Ricardo Buenaventura et al. 2008).

Most of the pharmacological effects of opioids (e.g., analgesia, euphoria, tolerance, dependence and withdrawal) are attributed to activity at neuronal opioid receptors, including the mu-, delta- and kappa-opioid receptors (Darcq and Kieffer 2018). However, more recent data suggest that immune signaling in the CNS, primarily mediated by glial cells, may also contribute to some of the negative effects of opioids. Such effects most notably include OIH, tolerance and dependence (Bachtell et al. 2017). Among glial cells, microglia are the main immune cells in the brain, detecting and responding to a range of infections and toxins. Notably, microglia are also activated by drugs of abuse, including opioids. One of the key mechanisms for this function is through activation of pattern recognition receptors (PRR), found on the cell membranes of microglia. One particular PRR, the Toll-like receptor 4 (TLR-4), binds opioids and activates microglia and results in the release of proinflammatory cytokines (TNF-α, IL-1β, and IL-6 and others), chemokines, lipid mediators of inflammation, matrix metalloproteases, and nitric oxide (Pocock and Kettenmann 2007). Commonly used opioids, including morphine, methadone, and buprenorphine activate TLR-4 (Hutchinson et al. 2010). Microglial cells also express mu and kappa opioid receptors; thus, stimulation of these receptors activates microglia as well (Trang et al. 2015).

The resultant inflammatory state induced by prolonged opioid treatment is thought to contribute to neuronal hyperexcitability and pain hypersensitivity by enhancing the conductance and number of the AMPA- and NMDA-type glutamate receptors, and down-regulation of GABA receptors (Li 2012; Watkins et al. 2009). In preclinical studies, a broad range of pharmacological treatments that reduce activation of microglia have prevented development of OIH and tolerance. One such medication is minocycline, which has been examined in animal models of chronic pain, as well as OUD, with promising findings (for a review, see Zhou et al., 2018). For instance, minocycline has been shown to potentiate opioid analgesia in a model of experimental pain (Hutchinson et al., 2008) as well as neuropathic pain (Popiolek-Barczyk et al., 2014), and also reduces abuse-related behavioral outcomes (Arezoomandan & Haghparast, 2015). Further, the same dose of minocycline used in the current study was recently shown to reduce levels of pro-inflammatory cytokines in individuals with schizophrenia, a finding which was predictive of improved treatment outcomes (Zhang et al. 2018).

In preclinical studies, minocycline has been shown to enhance opioid analgesia (Akgűn et al. 2018; Hutchinson et al. 2008), reduce opioid reinforcement (Arezoomandan and Haghparast 2015; Hutchinson et al. 2008), delay development of opioid tolerance (Cui et al. 2008), and prevent hyperalgesia (Mika et al. 2007). In human studies, minocycline has yielded mixed findings regarding pain. While some studies found improved pain or attenuated hyperalgesia (Samour et al. 2017a), others failed to illustrate any significant effects on pain (Curtin et al. 2017; Vanelderen et al. 2015). These inconsistencies between rodent and human studies regarding the effects of medications targeting microglia on pain and related outcomes have been previously recognized (Smith and Dragunow 2014).

Minocycline has also been evaluated for its potential effects on cognitive performance in animal models as well as in human studies. For example, minocycline improved cognitive function deficits induced by neuroinflammation (Hou et al. 2016), sleep deprivation (Wadhwa et al. 2017), and surgery (Jin et al. 2014) in mice. In healthy humans, minocycline improved the speed of performance in a response inhibition task (Sofuoglu et al. 2011) and improved performance in social decision making tasks (Kato et al. 2012; Watabe et al. 2012). Given the poor decision making and impulsivity associated with opioid addiction (Baldacchino et al. 2017), cognitive enhancement has suggested to be a potential treatment strategy (Sofuoglu et al. 2013).

Minocycline has yet to be examined in humans for its potential therapeutic effects addressing pain associated with OUD, and for its potential to improve cognitive function in this patient population. Given the promising preclinical findings with minocycline, we conducted a study examining the effects of short-term minocycline treatment on experimental pain and other OUD-related outcomes. Our sample was comprised of individuals with OUD maintained on opioid agonist treatment (i.e., methadone or buprenorphine/naloxone). Our outcomes included experimental pain threshold and tolerance (assessed with the Cold Pressor Test [CPT]), severity of pain, self-reported mood, opioid withdrawal, craving for opioids, and serum cytokine levels. Additionally, based on the cognitive enhancing effects of minocycline observed in both preclinical and clinical studies, we also included selective tests of cognitive function. We hypothesized that minocycline treatment, compared to placebo, would increase CPT pain threshold and tolerance, and reduce daily routine pain experience (i.e. muscle cramps, knee pain, etc.) in individuals with OUD maintained on opioid agonist treatment. As an exploratory aim, we examined the relationship between opioid craving and pain utilizing Ecological Momentary Assessment (EMA) surveys administered during the treatment protocol. The results of previous studies have been conflicting regarding the impact of craving on pain and vice versa (Garland et al. 2016; Martel et al. 2016); thus, it was of interest to explore this relationship in our sample of long-term opioid treated individuals. Finally, given the potential neuroinflammatory mechanism underlying OIH along with the recent study suggesting that minocycline reduces serum proinflammatory cytokines, we assessed serum levels of pro-inflammatory cytokines.

Materials and Methods

Subjects

Adults (ages 23 to 56) who were diagnosed with opioid dependence as defined by the DSM-IV, and who were currently enrolled in an opioid agonist treatment program (methadone or buprenorphine/naloxone) in the New Haven, Connecticut area were recruited. Participants were required to be compliant with their program and on a stable agonist dose for at least 2 weeks. They were medically healthy, as determined by physical examination and laboratory tests, not taking medication that would change pain responses (e.g., NSAIDS, antidepressants and anti- convulsant), and women were not pregnant and on a reliable birth control method. Those with current major psychiatric disorders (e.g., bipolar disorder, schizophrenia, depression) and other drug abuse or dependence (except tobacco or marijuana) as determined by the Structured Clinical Interview for DSM-IV (SCID) (First et al., 1996) and urine drug screening, were excluded. Fifty-five participants were screened, with 27 deemed eligible for participation and thus randomized. Of these, seven did not complete the study: two dropped out before beginning study medication, three dropped out after the first test session, one presented with illicit opioids in their urine at the second test session, and one dropped out before beginning the third test session. A total of 20 participants (15 male; 5 female) completed all study procedures. Baseline characteristics of the study participants are shown in Table 1 and were not significantly different between groups. The protocol was approved by both the Yale University and VA Connecticut Healthcare System institutional review boards, and written informed consent was obtained prior to participation. Participants were compensated for their participation.

Table 1.

Demographics

	Minocycline	Placebo	p value
	n = 10	n = 10
Female (%)	3 (30%)	2 (20%)	0.61
Mean (SD) age (years)	46.5 (4.53)	47.9 (10.00)	0.69
Race (n)			0.62
Black	1	2
White	7	6
Hispanic	2	1
Other	0	1
Methadone
Number (%)	8 (80 %)	10 (100%)
Mean (SD) dose (mg)	71.88 (25.49)	63.00 (35.64)	0.81
Buprenorphine
Number (%)	2 (20%)	0 (0%)
Mean dose (mg)	8mg (0)	-	-

Experimental Design

In this double-blind, randomized human laboratory study, participants were randomized to either minocycline (200mg/day) or placebo for 15 days. The study had one adaptation/baseline session prior to medication initiation, 3 test sessions (Days 1, 8 and 15 of treatment) and one follow-up visit one week after the end of medication. In order to capture a finer-grain measurement of study variables in the participant’s real-world environment (Stone and Shiffman 1994), EMA were collected four times a day on Days 8 through 14 of treatment using an HP iPAQ Pocket PC 2003 Pro Personal Digital Assistant (PDA).

During the adaptation session, participants were familiarized with study procedures and asked to complete the subjective, cognitive, and experimental pain measures described below. Test sessions lasted two hours or less and were conducted in the morning to minimize interference with participant’s typical dosing schedule. For each visit, participants were asked to refrain from taking their methadone or buprenorphine dose until the end of the session; compliance was confirmed by communication with the dosing nurse at their respective treatment clinic, or by bottle/sublingual film count when possible. This strategy was implemented in accordance with previous research on medication effects on experimental pain in opioid-maintained populations. These prior studies have shown increased pain sensitivity both immediately prior to methadone or buprenorphine dosing (at trough plasma opioid levels), as well as three hours after (at peak plasma opioid levels) (Compton et al. 2012). Thus, we chose to conduct all assessments prior to methadone/buprenorphine dosing in order to eliminate the confound of peak analgesia from the opioid agonist, and to maintain consistency across participants.

Before the start of each session, participants underwent a breathalyzer and a urine toxicology dipstick test to confirm presence of methadone and absence of any illicit drugs other than THC. Buprenorphine could not be assessed using this method and was confirmed during screening bloodwork, and otherwise by the patient’s verbal confirmation and prescription sublingual film count at each session. Baseline vital signs were also taken, and women were given a urine pregnancy test. After baseline measures were collected, participants received their assigned study medication. Starting one hour later, coinciding with the peak plasma levels of minocycline, assessment of study outcomes was initiated.

The main measures assessed 1) experimental pain, 2) subjective mood, opioid abuse- related, and pain outcomes, 3) cognitive performance, and 4) serum cytokine levels. In addition, heart rate and blood pressure were assessed for safety reasons and will not be further discussed. Experimental pain was assessed with the CPT, and our methods were adapted from Eckhardt et al. (1998). The CPT was chosen because of its particular sensitivity for robustly detecting opioid-induced changes in pain, including analgesia and hyperalgesia (Chu et al. 2006; Compton 1994; Doverty et al. 2001; Koltzenburg et al. 2006), as compared to other experimental pain measures such as noxious electrical stimulation. For this test, two water coolers were utilized; one was filled with warm water (100.04°F/37.8°C), and the other with ice water (32.9–34.7°F/0.5–1.5°C). Participants first immersed their hand into the warm- water bath for two minutes to standardize hand temperature before moving their hand into the ice water bath. They were instructed to verbally report the first time they experienced a painful sensation (pain threshold), but to keep their hand submerged as long as possible and withdraw from the water if the pain became too uncomfortable (pain tolerance). For this outcome, latency (in seconds) to pain threshold and pain tolerance was recorded.

Subjective measures included to capture the effects of minocycline on mood (the Profile of Mood States; POMS (McNair et al. 1992)), and severity and impact of occasional pain experiences (the Brief Pain Inventory Short Form; BPI-SF (Keller et al. 2004)). In addition, the Short-Form McGill Pain Questionnaire (SF-MPQ) was used to capture the sensory and affective dimensions of the pain experience immediately following the CPT (Melzack 1987). A visual analogue scale (VAS; 0 to 10) was used to assess pain intensity during the CPT.

In addition to questions about pain and opioid craving, the EMA assessments included the Sustained Attention to Response Test (SART) and the Digit Symbol Substitution Test (DSST) cognitive assessments. The SART is a Go/No-Go task and assesses the ability to withhold responses to an infrequently occurring target (Robertson et al. 1997). The DSST is a test of psychomotor performance, which measures motor persistence, sustained attention, response speed and visuomotor coordination (Thorndike 1919). These tests were chosen because of their sensitivity to drug and medication effects, including opioids and minocycline (Sofuoglu et al. 2011). The EMA also assessed potential opioid withdrawal by asking the Subjective Opioid Withdrawal Scale (SOWS; (Handelsman et al. 1987) at each timepoint.

Serum cytokine analysis included three cytokines that have been linked to inflammation and are sensitive to the dose of minocycline utilized in the current study (Zhang et al. 2018): Interleukin-1 beta (IL-1β), Interleukin-6 (IL-6), and Tumor and Necrosis Factor alpha (TNFα). They were assayed using electrochemiluminescence multi-array technology (Meso Scale Discovery, Gaithersburg, MD) as described in the methods of DellaGioia et al. (2013). Serum samples were obtained at baseline prior to treatment initiation, and once more on the last day of treatment.

Minocycline

Participants were randomized to active (200mg minocycline) or placebo, given as a single daily dose, by the research pharmacist. We chose a 200mg dose of minocycline based on the typical daily regime used in clinical trials for the treatment of infections, as well as our previous studies with minocycline (Sofuoglu et al. 2011; Sofuoglu et al. 2009; Zhang et al. 2018). Medication was initiated during the first laboratory session and continued for the proceeding two weeks, constituting 15 total days of administration. This treatment duration was shown to reduce response to experimentally-induced pain in humans (Samour et al. 2017b). Following oral administration, peak plasma levels of minocycline are reached within 1– 4 hour, and the elimination half-life of minocycline ranges from 11 to 24 hours. The most common side effects associated with minocycline include dizziness, vertigo, nausea and vomiting. Photosensitivity manifested by an exaggerated sunburn reaction has also been reported with minocycline, albeit rarely. As minocycline should be taken on an empty stomach, subjects were advised to fast for at least one hour prior to coming to the clinic.

Statistical Analysis

Primary Analysis

For lab data, linear mixed models (LMM; SAS Systems, Cary, NC) tested the effect of Group (Active vs. Control), Week (treated as a categorical variable) and the Group x Week interaction on subjective, behavioral, and cognitive measures. Pre-treatment measures of each dependent variable were included as covariates. For cognitive data, assessment occurred before and after minocycline administration at laboratory visits at Weeks 1, 2, and 3. LMMs therefore tested the main effect of Group, the Group x Week interaction and the Group x Pre-Post interaction. For analysis of cytokine data, ANCOVA was used; data from screening was included as a covariate. For EMA data, LMMs were used for analysis of subjective and cognitive data, testing the (Treatment) Group, Day, and Group x Day interaction. See Supplementary Materials for additional details.

Exploratory Analysis

In exploratory analyses, using Generalized Estimating Equations (GEE) we examined associations between Craving and Pain. Mean Craving scores (average of all craving ratings during EMA) and Deviation Craving scores (difference scores between craving at each assessment and the Mean craving rating) were created for each subject. A significant effect for Mean Craving would indicate a between-subject effect, where individuals who report generally more Craving report generally higher Pain ratings during EMA. A significant effect for the deviation score (Deviation Craving) would reveal a within-subject effect, where individuals report higher Pain ratings when they experience more Craving than usual. The GEE also included Day of EMA Study and Treatment condition as covariates. In GEE analyses, Craving was treated both as a continuous variable and an ordinal variable (in separate models).

Justification for the sample size

In a previous study with healthy controls, minocycline treatment attenuated d- amphetamine induced subjective drug effects and plasma cortisol increses and improved performance on a Go/NoGo task, with large effects sizes for all 3 outcomes (Sofuoglu et al. 2011). A sample of 20 participants were sufficient to detect similar large effect sizes from minocycline treatment for the main outcome measures.

Results

Summary statistics of laboratory, EMA, and cytokine data for each group are reported in Table 2. Neither pain threshold nor tolerance for pain in the CPT was affected by minocycline treatment (ps > 0.05) (See Table S1 in Supplementary Materials).

Table 2.

Summary Statistics on Behavioral, Subjective, and Cytokine Data by Training Group and Time

Group ↓		Setting →	Screening	Lab	Lab	Lab	EMAa	Lab	Lab
		Week →	−1	0	1	2		3	4
					Treatment
Active	CP-Threshold (sec)			18.72 (11.30)	16.16 (6.46)b	16.13 (5.67)		16.21 (5.86)	17.59 (9.34)
Mean (SD)	CP-Tolerance (sec)			40.86 (21.71)	37.42 (12.76)b	34.41 (11.84)		36.61 (20.45)	31.58 (12.56)
	BPI Severity			1.28 (1.83)	0.88 (1.56)	0.78 (1.27)		0.80 (1.30)	1.10 (1.45)
	BPI Interference			0.93 (1.74)	0.90 (1.69)	0.63 (1.23)		0.67 (1.30)	0.52 (0.89)
	POMS – Depression			20.80 (5.33)	18.90 (5.84)	21.00 (6.46)		19.20 (5.43)	18.80 (7.57)
	POMS - Total Mood Disturbance			43.10 (15.04)	44.70 (14.43)	50.30 (18.29)		43.40 (17.51)	40.70 (15.52)
	McGill Sensory			10.33 (7.88)	10.9 (8.32)	10.80 (7.54)		11.30 (8.26)	12.50 (7.80)
	McGill Affective			2.00 (3.32)	1.70 (2.11)	1.90 (3.45)		2.40 (3.24)	1.89 (3.48)
	IL-1β (pg/ml)		0.80 (0.58)					0.67 (0.44)
	IL-6 (pg/ml)		1.64 (1.30)					1.11 (0.94)
	TNF-α (pg/ml)		2.86 (1.79)					2.61 (1.74)
	Pain (1–7)						1.22 (0.63)
	SOWS (1–7)						1.05 (0.24)
	Craving (1–7)						1.04 (0.18)
Control	CP-Threshold (sec)			24.78 (13.77)	24.37 (17.57)	18.46 (6.86)		25.75 (15.45)	21.92 (7.93)
Mean (SD)	CP-Tolerance (sec)			59.72 (30.40)	52.56 (28.54)	48.42 (36.62)		51.52 (37.53)	43.14 (15.46)
	BPI Severity			1.10 (1.93)	0.95 (1.96)	1.00 (1.81)		1.18 (2.66)	1.28 (2.57)
	BPI Interference			0.97 (2.37)	1.33 (2.07)	1.04 (2.15)		1.49 (2.57)	1.20 (2.57)
	POMS – Depression			18.90 (5.22)	17.90 (3.35)	16.70 (3.71)		17.00 (5.01)	17.80 (7.87)
	POMS - Total Mood Disturbance			47.40 (22.63)	41.88 (13.62)	40.40 (15.38)		42.30 (17.51)	42.20 (23.61)
	McGill Sensory			12.90 (8.69)	13.10 (4.77)	14.20 (11.05)		13.30 (9.44)	10.70 (8.82)
	McGill Affective			1.50 (2.22)	1.70 (2.11)	2.20 (3.74)		2.00 (3.74)	2.30 (4.22)
	IL-1β (pg/ml)		0.72 (0.55)					0.62 (0.53)
	IL-6 (pg/ml)		2.14 (1.63)					1.99 (2.53)
	TNF-α (pg/ml)		2.66 (0.80)					2.53 (0.96)
	Pain (1–7)						1.14 (0.49)
	SOWS (1–7)						1.03 (0.22)
	Craving (1–7)						1.02 (0.20)

^aTable Note EMA occurred between Week 2 and Week 3 Laboratory visits.

^bCP-Threshold and CP-Tolerance were assessed post-intervention at Week 1. All other assessments were assessed pre-intervention at that visit. CP = Cold Pressor; BPI = Brief Pain Inventory; POMS = Profile of Mood States; SOWS = Subjective Opiate Withdrawal Scale

Subjective Data

There was no evidence that minocycline had an effect on BPI, SOWS, POMS, or McGill measures assessed in the lab, or pain, craving, and opioid withdrawal assessed during EMA (ps > 0.05) (See Table S1 in Supplementary Materials).

Cytokine Data

Minocycline did not have a significant effect on serum cytokine levels (ps > 0.05) (See Table S1 in Supplementary Materials).

Cognitive Data

Summary statistics are reported in Tables 3. There was no evidence that minocycline had an effect on DSST assessed in the lab, or measures from the Go/No-Go task assessed during EMA (ps > 0.05). However, there was evidence that minocycline acutely improved performance in the lab (significant Group x Pre-Post interaction for errors of commission (No-go trials) and omission (Go trials)) (Table 4). On No-go trials, the effect of Pre-Post was significant for the Active group (p = 0.02); averaged over weeks the number of errors decreasing from 7.55 to 5.61 from pre- to post-, but not for the Control group, with the number of errors changing from 7.62 to 8.64 from pre- to post- (p = 0.24).

Table 3.

Summary Statistics on Cognitive Data by Training Group and Time

Group↓	Setting →	Lab	Lab	Lab	Lab	Lab	EMAa	Lab	Lab	Lab
	Week →	0	1	1	2	2		3	3	4
	Pre-Post Medication→		Pre	Post	Pre	Post		Pre	Post
				Treatment
Active	DSST	37.11 (8.57)	42.33 (8.12)	41.78 (6.72)	45.67 (8.32)	43.25 (11.51)		48.44 (13.38)	46.56 (9.19)	50.75 (8.97)
	Go/No-go commission (0–25)	7.00 (6.41)	5.78 (4.60)	5.75 (5.09)	7.40 (5.76)	5.50 (5.80)		9.30 (5.58)	5.60 (4.78)	7.00 (4.54)
	Go/No-go omission (0–200)	10.90 (8.17)	9.11 (12.67)	6.88 (8.08)	15.80 (29.67)	7.70 (12.14)		18.40 (18.71)	13.20 (12.22)	11.10 (7.69)
	Go/No-go RT (ms)	451.75 (119.87)	429.78 (85.55)	438.25 (110.91)	462.05 (94.13)	466.15 (92.26)		436.22 (89.76)	453.91 (85.64)	422.17 (89.47)
	PDA Go/No-go commission (0–25)						9.59 (5.49)
	PDA Go/No-go omission (0–200)						15.34 (17.62)
	PDA Go/No-go RT (ms)						388.45 (162.29)
Control	DSST	37.11 (9.05)	40.89 (4.96)	40.89 (5.30)	43.33 (6.96)	41.11 (7.41)		43.00 (7.23)	40.11 (9.16)	46.67 (5.89)
	Go/No-go commission (0–25)	9.00 (5.40)	8.33 (6.16)	7.33 (5.27)	7.89 (6.47)	8.70 (5.10)		7.30 (7.67)	9.89 (7.27)	7.20 (6.88)
	Go/No-go omission (0–200)	17.90 (16.78)	13.78 (9.93)	13.44 (9.94)	21.00 (23.81)	27.00 (27.87)		25.40 (30.16)	22.33 (24.79)	21.10 (21.13)
	Go/No-go RT (ms)	395.56 (114.10)	378.35 (98.63)	410.08 (95.79)	433.10 (112.79)	426.83 (121.37)		437.87 (117.95)	405.11 (87.60)	438.90 (124.71)
	PDA Go/No-go commission (0–25)						9.36 (7.04)
	PDA Go/No-go omission (0–200)						17.73 (22.12)
	PDA Go/No-go RT (ms)						404.42 (190.76)

Table Note

^aEMA occurred between Week 2 and Week 3 Laboratory visits. DSST = Digit Symbol Substitution Test; RT = Reaction Time; PDA = Personal Digital Assistant

Table 4.

Results of Analyses on Cognitive Data

	Group					Group x Time					Group x Pre-Post
	Df	PE	SE	F	p	df	PE	SE	F	p	df	PE	SE	F	p
Lab Data
DSST	1, 85	3.01	2.26	1.77	.19	1, 83	4.81	2.62	1.99	.14	1, 84	−0.32	2.01	0.03	.87
Go/No-go commission (0–25)	1, 90	−0.62	1.60	0.81	.70	2, 88	2.01	1.79	0.67	.51	1, 89	−3.25	1.20	7.30	.008
Go/No-go omission (0–200)	1, 90	−3.51	6.07	0.34	.56	2, 88	1.91	6.98	0.41	.66	1, 89	−8.82	4.05	4.74	.03
Go/No-go RT (ms)	1, 90	−2.33	25.71	0.01	.93	2, 88	−30.68	26.99	1.34	.27	1, 89	13.89	11.85	1.37	.24
EMA Data
PDA Go/No-go commission (0–25)	1, 439	−0.68	2.35	0.08	.77	1, 438	0.18	0.29	0.40	.53	-	-	-	-	-
PDA Go/No-go omission (0–200)	1, 439	−3.81	6.67	0.33	.57	1, 438	−0.59	1.00	0.35	.55	-	-	-	-	-
PDA Go/No-go RT (ms)	1, 439	−28.44	71.56	0.16	.69	1, 438	−6.90	7.34	0.88	.35	-	-	-	-	-

Table Note: LMMs were used for analysis of cognitive data. For Lab Data, Time reflects Week (coded as a categorical variable with 3 levels, Week 1, Week 2, and Week 3), and Pre-Post reflects timing of drug administration at each lab visit (Pre- vs. Post). For EMA data, Week reflects Day of EMA (coded as a numerical variable). For Group x Time interaction (Lab Data), parameter estimate reflects comparison for Week 3 (end of treatment) vs. Week 1 (start of treatment). DSST = Digit Symbol Substitution Test; RT = Reaction Time; PDA = Personal Digital Assistant. Significant results are in bold.

Exploratory Analysis

Using GEE, Mean Craving scores were not significantly associated with Pain (p = 0.17), but Deviation Craving scores were significantly associated with Pain (PE = 0.45, SE = 0.10, p < 0.001). The association between Deviation Craving scores and Pain was also significant when Pain was treated as an ordinal variable (PE = 1.06, SE = 0.33, p = 0.001). Thus, when subjects’ experience more craving than their (subject-specific) average, they reported more incidences of pain.

Discussion

In this study, 15 days of minocycline treatment did not change pain threshold and tolerance on the CPT in patients with OUD maintained on opioid agonist treatment (methadone or buprenorphine/naloxone). Similarly, minocycline did not affect pain severity, opioid craving, withdrawal severity, mood, or serum cytokine levels when compared to placebo. For cognitive outcomes, minocycline improved the accuracy of performance in the Go/No-Go task. Exploratory analysis suggested that participants reported more pain when they reported higher craving for heroin.

Before discussing the implications of these findings, several limitations of the study should be mentioned. First, this was a small sample with 10 opioid-maintained participants assigned to each of an active minocycline and a placebo condition. Further, given the pilot- study nature of this investigation, we did not include a third group of healthy control participants who were not opioid-maintained as a comparator. Secondly, participants were required to be clinically stable on their opioid dose (i.e., no current illicit drug use other than THC, minimal craving, withdrawal, or pain). While this relatively stable sample minimized the potential risks from study participation, it may have made it difficult to show changes in outcomes as a result of minocycline treatment. Third, treatment duration was relatively brief, inclusive of 15 days with only one dose of minocycline (200mg/day). It is unclear if longer or different doses would be more effective for the study outcomes (see, (Compton et al. 2010).

Minocycline treatment improved accuracy of performance in the Go/No-Go task. This finding is similar to our previous study in which 200mg/day minocycline for 4 days improved the speed of the same task in healthy controls (Sofuoglu et al. 2011). In previous studies, minocycline treatment was associated with more focused decision-making, balancing risk and trustworthiness in a trust game in healthy controls (Kato et al. 2012; Watabe et al. 2012). These findings led the authors to suggest that perhaps minocycline reduces the “noise” introduced by microglial activation during a decision process. If replicated, improvement of cognitive performance could be therapeutically beneficial in individuals with OUD, who display deficits in various cognitive functions (Baldacchino et al. 2017). Cognitive deficits are predictive of treatment response to multiple addictions, including OUD. Accordingly, cognitive enhancement (or cognitive remediation) has been suggested to be a potential treatment strategy for OUD (Rezapour et al. 2016; Sofuoglu et al. 2013). Further studies are needed to determine if such treatments enhance outcomes for OUD.

Given the dual focus of the study on pain and OUD-related outcomes, as an exploratory aim we also examined the relationship between craving for opioids and rating of pain assessed using EMA. Our finding of a strong association between craving and pain is consistent with previous cross-sectional studies, including a recent EMA study conducted in chronic pain patients receiving prescription opioids (Martel et al. 2016). A unique feature of our study was that our participants were clinically stable, having no illicit opioid use and low levels of craving and pain. Even in this sample, the association between craving and pain supports the importance of pain processes in chronic opioid use and potential relapse.

In summary, minocycline treatment did not affect the threshold or tolerance to pain assessed with the CPT in individuals with OUD who are maintained on an opioid agonist. Based on the positive findings from preclinical studies, the lack of minocycline effects on CPT response, pain severity, withdrawal and craving was somewhat unexpected. Perhaps more relevant for our study, a recent preclinical study found that while minocycline was effective in delaying development of opioid tolerance, it failed to reverse existing opioid tolerance (Zhang et al. 2015). Our findings do not support the potential use of minocycline to reduce OIH, or pain tolerance in individuals with OUD. Minocycline also did not affect opioid craving, withdrawal or serum cytokine levels. However, minocycline treatment was associated with more accurate performance on the Go/No-Go task. While these findings do not support minocycline’s use in attenuating OIH, minocycline may have a potential use as a cognitive enhancer for individuals with OUD.