Extended-release naltrexone improves viral suppression levels in prisoners with HIV and alcohol use disorders who are transitioning to the community: Results from a double-blind, placebo-controlled trial

Abstract

Objective

To determine if extended-release naltrexone (XR-NTX) would improve or maintain viral suppression (VS) among incarcerated individuals with HIV and alcohol use disorders (AUDs) transitioning to the community.

Design

A randomized double-blind, placebo-controlled trial was conducted among incarcerated individuals with HIV and AUDs transitioning to the community from 2010 through 2016.

Methods

Eligible participants (N=100) were randomized 2:1 to receive 6 monthly injections of XR-NTX (n=67) or Placebo (n=33) starting at release and continued for 6 months. The primary and secondary outcomes were the proportion that maintained or improved VS at <200 copies/mL and <50 copies/ml from baseline to 6 months respectively using an intention to treat analysis.

Results

Participants allocated to XR-NTX improved VS from baseline to six months for <200 copies/mL (48.0% to 64.2%, p=0.024) and for <50 copies/mL (31.0% to 56.7%, p=0.001); while the placebo group did not (<200 copies/mL: 64% to 42.4%, p=0.070; <50 copies/mL: 42.0% to 30.3%, p=0.292). XR-NTX participants were more likely to achieve VS than placebo at six-months (<200 copies/mL: 64.2% vs. 42.4%; p=0.041; <50 copies/mL: 56.7% vs. 30.3%; p=0.015). XR-NTX independently predicted VS (<200 copies/mL: aOR=2.68, 95%CI=1.01–7.09, p=0.047; <50 copies/mL: aOR=4.54; 95% CI=1.43–14.43, p=0.009) as did receipt of ≥3 injections (<200 c/mL: aOR=3.26; 95% CI=1.26–8.47, p=0.010; <50 c/mL: aOR=6.34; 95%CI=2.08–19.29, p=0.001). Reductions in alcohol consumption (aOR=1.43; 95% CI=1.03–1.98, p=0.033) and white race (aOR=5.37; 95% CI=1.08–27.72, p=0.040) also predicted VS at <50 copies/mL.

Conclusion

XR-NTX improves or maintains VS after release to the community for incarcerated people living with HIV and AUDs.

Introduction

Globally, people living with HIV (PLH), and especially those with substance use disorders, are disproportionately concentrated in prisons and jails.¹ In the United States (U.S.), one-sixth of the 1.2 million PLH cycle through these criminal justice settings annually, resulting in an HIV prevalence that is 3-fold higher than the community.² Despite high levels of viral suppression (VS) achieved during incarceration,³ this transition is made tumultuous by low linkage to HIV care,⁴ high rates of relapse to substance use,⁵ homelessness,⁶ and overdose.⁷ Together, these factors undermine adherence to antiretroviral therapy (ART) after release and result in loss of VS.^4,8–10

Alcohol use disorders (AUDs) are 8-times more prevalent in criminal justice populations than in the community, with relapse especially common and problematic post-release.^11,12 In PLH, AUDs exacerbate ART non-adherence, resulting in suboptimal VS.¹³ The majority of prisoners in the U.S. do not receive medication to treat AUDs or prevent relapse to alcohol use at time of release.^11,14 Extended-release naltrexone (XR-NTX) is an evidence-based and FDA-approved treatment for AUDs that delays and reduces alcohol consumption,¹⁵ including for incarcerated PLH with AUDs and released to the community.¹⁶ We therefore hypothesized that XR-NTX would reduce alcohol consumption in PLH with AUDs who are transitioning to the community and either maintain or improve VS post-release, compared to those with untreated AUDs. We tested this hypothesis within a placebo-controlled trial using XR-NTX by specifically examining the influence of XR-NTX on achieving VS six months after release from prison or jail.

Methods

A double-blind, placebo-controlled prospective trial of XR-NTX among PLH with AUDs who were transitioning to the community was conducted from September 2010 through February 2016 in two sites in Connecticut. Eligible participants were randomized 2:1 to XR-NTX or placebo, receiving their first injection within one week before release from prison or jail; treatment was continued for 6 months post-release. Recruitment procedures and eligibility criteria,¹⁷ early and final retention levels,^17,18 hepatic safety,^16,19 alcohol consumption outcomes,¹⁶ and adverse events¹⁶ have previously been reported.

Recruitment

After screening, 100 participants were ultimately recruited from prisons and jails within the Connecticut Department of Correction. Incarcerated PLH were first screened for self-reported hazardous drinking (≥4 drinks daily for women or ≥5 drinks daily for men),²⁰ followed by more detailed screening assessments.¹⁷

Screening, Eligibility and Consent

Inclusion criteria

1) documented HIV-infection; 2) adults ≥18 years of age; 3) transitioning to either New Haven or Hartford, Connecticut; 4) met DSM-IV criteria for alcohol abuse or dependence using the Mini International Neuropsychiatric Interview (MINI)²¹ or hazardous drinking using the Alcohol Use Disorders Identification Test (AUDIT) (score ≥4 for women and ≥8 for men);²² 5) provided informed consent; and 6) spoke English or Spanish.

Exclusion criteria

1) prescribed opioid pain medications or reported a medical indication for them post-release; 2) grade 3 or higher aspartate aminotransferase (AST) or alanine aminotransferase (ALT) elevations (>5× upper limit of normal); 3) evidence of Childs-Pugh Class C cirrhosis; 4) enrolled in another alcohol pharmacological or ART adherence study; or 5) breastfeeding, pregnant or unwilling to use contraception (women only).

Eligible and interested prisoners then completed verbal and written informed consent procedures, which were repeated immediately after release to prevent real or perceived coercion.

Ethical Oversight

All study procedures were reviewed and approved by the Institutional Review Board at Yale University, the Connecticut Department of Correction Research Advisory Committee, and the Office of Human Research Protections. A Certificate of Confidentiality was obtained from the National Institutes of Health. The study is registered at www.clinicaltrials.gov (NCT01077310).

Randomization

After consent, participants were randomly assigned 2:1 to receive 380 mg of XR-NTX (Vivitrol®) or placebo (provided in-kind by Alkermes, Inc.), administered intramuscularly every 4 weeks for six months. A covariate-adapted stratified block randomization was performed controlling for the presence or absence of concurrent opioid use disorder and whether prescribed ART or not.¹⁷

Study Measures

HIV-1 RNA levels, along with CD4 T-lymphocyte cell counts, were assessed at baseline, time of release and at 3 and 6 months post-release. Other study measures included: demographic information (age, gender, race, housing status), type of ART regimen (protease inhibitor, non-nucleoside reverse transcriptase inhibitor, or integrase strand transfer inhibitor regimens), health insurance status, duration of incarceration, symptoms of depression using the Brief Symptom Inventory (BSI; defined as a general T-score of ≥63 or any two primary dimension scores of ≥63)²³, alcohol and drug use, hepatitis C virus co-infection, and presence of co-morbid mental illness and other substance use disorders using the MINI. Our alcohol consumption variables¹⁶ were derived from a timeline follow back (TLFB)^24–26 that assessed self-reported daily totals of standard drinking units before incarceration and monthly for six months. Because drinking patterns are varied, we used a previously described Alcohol Improvement Score to more comprehensively measure heterogeneous drinking patterns.¹⁶ This score included the: 1) time to first heavy drinking day; 2) total number of drinks per drinking day; 3) percent of heavy drinking days (≥5 drinks per day for men, or ≥4 drinks per day for women); 4) pre- to post-incarceration change in average drinks per day; and 5) total number of drinking days. Each of these five variables were combined and weighted equally with a score of “1” for favorable outcomes and combined into a unit-weighted composite score using previously described criteria.¹⁶ This score generated a single summary measure that was more easily amenable to interpretation of heterogeneous drinking patterns, and more sensitive to data variance and subtle changes by reducing floor and ceiling effects.¹⁶

Adverse events were also monitored monthly and included hepatic transaminase levels and injection site reactions. Previous reports from this trial found no differences in adverse events between the two groups nor were there any serious adverse events in either treatment group.¹⁶

Outcome Variables

The original pre-planned primary outcomes were defined as the proportion who achieved or maintained HIV VS at the <400 copy/mL and the <50 copy/mL levels from baseline to six months. The Department of Health and Human Services HIV treatment guidelines changed their definition of VS to <200 copies/mL as the goal of therapy soon after study initiation,²⁷ resulting in a change in one of the primary outcomes from <400 copies/mL to <200 copies/mL.¹⁷ The other pre-planned co-primary outcome, change in maximal viral suppression defined as <50 copies/mL, remained a clinically important outcome in real-world practice, and is listed as a secondary outcome in this manuscript. Using an intention-to-treat (ITT) strategy, the primary and secondary outcomes involved a comparison of the changes in VS levels of <200 copies/mL and <50 copies/mL from baseline to 6 months after release in both treatment groups respectively. This outcome best reflected how participants did over time (baseline to 6 months), since our hypothesis was that participants who received placebo would be more likely to lose VS after release, while those who benefited from XR-NTX would either maintain or improve VS levels, or if not on ART, initiate it due to increased post-release stability.^28,29 Additionally, the difference between groups in proportions with VS at the 6 month time-point was assessed.

Sample Size and Power Calculations

We calculated an original sample size of 125 (XR-NTX=83 and placebo=42) needed to detect a statistically significant difference in the primary outcome at 6 months between the two groups. This incorporated a two-sided alpha=0.05, beta=0,20, and a compound symmetry true correlation structure of 0.5 (the most conservative, based on our results from earlier studies where our prison-release data suggested that 59% of HIV+ inmates leave prison with VS,¹⁰ and where 28% of HIV prisoners with an AUD leaving prison have VS after six months).³⁰ Power calculations as previously published¹⁷ also included oversampling (2:1 randomization) those receiving XR-NTX due to concern for potential increase in adverse events³¹ in the XR-NTX group and were also based upon prior aforementioned studies of released prisoners that had 80% retention after 6 months.^30,32

Participants’ disposition

Of the 195 PLH referred to the study, 118 consented and completed baseline interviews, and 100 were fully reenrolled and randomized and included in the final analytical sample; 67 were randomized to receive XR-NTX and 33 to receive placebo. The CONSORT diagram is depicted in Figure 1. Study retention was not statistically different between study arms as previously reported in a prior manuscript.¹⁶

Figure 1.

CONSORT Flow Diagram

Analytical Approach

Missingness Analysis

Overall, 13% of participants who were randomized had missing HIV-1 RNA data 6 months post-release. Therefore, we explored the structure of the missing data to determine if the data were “Missing Completely at Random” (MCAR) and not related to the dependent or independent variables. The structure of the missing data was assessed using Little’s MCAR³³ test implemented with code within the BaylorEdPsych package in R.³⁴ The highly non-significant results for Little’s MCAR test (p=0.560) suggested that the missing data were not statistically related to the main outcome (VS), viral load at baseline, nor any of the variables used in the analysis, most importantly, treatment assignment or number of XR-NTX injections. High p-values for Little’s MCAR Test also suggested that further missingness inquiries using sensitivity analysis were not merited because the data were clearly neither Missing at Random (MAR) nor Not Missing at Random (MNAR).^35,36 Although the MCAR result allowed the application of multiple imputation, when this was done it gave inconsistent results due to co-linearity of some variables. Consequently, we were able to maintain the most conservative standard ITT assumption that missingness from participant attrition equals viral non-suppression (missing = failure). This is the standard analytic method for regulatory submission of HIV-1 RNA data to the U.S. Food and Drug Administration,³⁷ which provides the most sensitive and conservative detection limits available and used previously in prospective trials of PLH where HIV-1 RNA is the outcome.

Intention-to-Treat Analysis of viral suppression from baseline to 6 months post-release

An intention-to-treat (ITT) analysis was conducted first by dichotomizing VS as suppressed or non-suppressed and comparing viral suppression levels for primary and secondary outcomes at six months were performed using Chi-Square testing. Further, change in VS over time from baseline to six months were deployed to more accurately reflect the differences in viral suppression levels. These analyses were conducted using T tests in the R statistical software.

Multiple Logistic Regression Analysis of Predictors of Viral Suppression at 6 months

After confirming that a statistically significant difference was found for each level of VS both over time and at six months, we explored predictive variables guided by the literature that included receipt of ART,^18,38 as well as specific to this study, treatment assignment and the number of injections received, to further explain independent predictors for the primary (VS <200 copies/mL) and secondary (VS <50 copies/mL) outcomes. Of note the majority of participants were receiving ART and there was no difference in the number on ART with almost 90% receiving ART at time of release. The number of injections was dichotomized as two or fewer injections versus three or more to reflect better retention in the study due to prior studies showing that the first three months is enough time to lose viral suppression after release.^8,10 A backward stepwise model selection “step” algorithm in R software then sequentially eliminated variables until we achieved models with the best goodness-of-fit using the Akaike information criterion (AIC), as they yielded the most parsimonious results. Of note due to number of potential variables, those with the best AIC are demonstrated.

Results

Baseline Characteristics

The baseline characteristics are described in Table 1. Participants were, on average in their mid-40s, mostly men (77%), racial/ethnic minorities (84%), homeless or unstably housed (63%), prescribed ART (87%), and had major depression (75%). Central to the analysis, baseline viral suppression levels at <200 copies/mL and <50 copies/mL was 53% and 35%, respectively, and not significantly different (66% had VS<400 copies/mL, not significantly different). The two treatment arms did not differ significantly for any baseline variable aside from median CD4 count (XR-NTX: 490 vs. Placebo: 418 cells/mL; p=0.033); however, 90% of participants in both groups had CD4 counts >200 cells/mL.

Table 1.

Baseline Characteristics (N=100)

Variable	XR-NTX N=67 (%)	Placebo N=33 (%)	Total N=100 (%)	p Value
Gender
Male	50 (74·6)	27 (81·8)	77 (77·0)	0·422
Female	16 (23·9)	5 (15·2)	21 (21·0)
Transgender	1 (1·5)	1 (3·0)	2 (2·0)
Ethnicity
Black	46 (68·7)	19 (57·6)	65 (65·0)	0·528
Hispanic	11 (16·4)	8 (24·2)	19 (19·0)
White	10 (14·9)	6 (18·2)	16 (16·0)
Mean Age, years	44·9 (SD ± 8·12)	45·2 (SD ± 8·92)	45·0 (SD ± 8·35)	0·866
Completed High School (N=99)	35 (53·0)	15 (45·5)	50 (50·5)	0·568
Referred from
Prison	17 (25·4)	6 (18·2)	23 (23·0)	0·085
Jail	43 (64·2)	27 (81·2)	70 (70·0)
Community (within 30 days)	7 (10·4)	0 (0)	7 (7·0)
Duration of current incarceration (months)	13·99 (SD ± 30·32)	11·64 (SD ± 18·05)	13·21 (SD ± 26·82)	0·683
Study Site
New Haven	40 (59·7)	19 (57·6)	59 (59·0)	0·839
Hartford	27 (40·3)	14 (42·4)	41 (41·0)
Housing status (N=99)
Stable	22 (33·3)	14 (42·4)	36 (36·4)	0·440
Unstable	17 (25·8)	10 (30·3)	27 (27·3)
Homeless	27 (40·9)	9 (27·3)	36 (36·4)
Chronic Hepatitis C virus infection	29 (47·5)	16 (50·0)	45 (48·4)	0·822
Currently prescribed ART	57 (85·1)	30 (90·9)	87 (87·0)	0·470
Prescribed ART Regimen
Protease Inhibitor (PIs)	29 (51·7)	16 (55·2)	45 (51·7)	0·441
Non-Nucleoside Reverse
Transcriptase Inhibitor (NNRTIs)	20 (35·7)	10 (34·4)	30 (34·5)
Integrase Inhibitor	5 (8·9)	3 (10·3)	8 (9·4)
Combination	3 (5·4)	1 (3·4)	4 (4·7)
HIV-RNA (copies/mL)
< 400	22 (63.6)	42 (67.7)	66 (66.0)	0.292
< 200	32 (48.0)	21 (64.0)	53 (53·0)	0·135
< 50	21 (31·0)	14 (42·0)	35 (35·0)	0·293
HIV-RNA (copies/mL)
Mean (SD)	4427 (± 14386)	8119 (±37959)	5683 (± 24884)	0·492
Log10 Mean (SD)	2·43 (±1·03)	2·22 (±1·04)	2·36 (±1·03)	0·368
CD4 count ≥ 200 cells/mL (cells/mL)	60 (89·6)	30 (90·9)	90 (90·0)	0·832
Median CD4 count (cells/mL)	490	418	410	0·033
Psychiatric Conditions***
Bipolar Disorder	12 (18·8)	4 (12·1)	16 (16·5)	0·457
Major Depressive Disorder	10 (15·6)	4 (12·1)	14 (14·4)	0·481
PTSD	5 (7·8)	3 (9·1)	8 (8·2)	0·536
Panic Disorder	6 (9·2)	1 (3·1)	7 (7·2)	0·260
Psychotic Disorder	9 (14·1)	2 (6·1)	11 (11·3)	0·325
Brief Symptom Index, Depression	46 (78·0)	22 (68·8)	68 (74·7)	0·334
Alcohol Use Severity+
Abstinent or Low-Risk Drinking	2 (3·0)	0 (0·0)	2 (2·0)	0·174
Hazardous Drinking	5 (7·5)	2 (6·0)	7 (7·0)
Harmful Drinking	0 (0·0)	4 (12·1)	4 (4·0)
Dependent Drinking	60 (89·6)	27 (81·8)	87 (87·0)
Opioid Dependence*	15 (22·4)	7 (21·2)	22 (22·0)	0·894
Mean Substance use duration (years)^
Cannabis (SD)	6·63 (± 9·550)	4·14 (± 7·395)	5·81 (± 8·937)	0·424
Cocaine (SD)	8·68 (± 9·939)	8·34 (± 10·495)	8·67 (± 10·074)	0·747
Heroin (SD) (N=99)	3·06 (± 7·512)	2·68 (± 6·763)	2·94 (± 7·239)	0·890
Substance Use Disorder from M·I·N·I#
Cannabis Use Disorder (N=77)	9 (18·8)	3 (10·3)	12 (15·6)	0·259
Cocaine Use Disorder (N=89)	38 (63·3)	15 (51·7)	53 (59·6)	0·324
Opioid Use Disorder (N=87)	11 (19·6)	3 (9·7)	14 (16·1)	0·361
XR-NTX Injections received
0–3	40 (59·7)	19 (57·6)	59 (59·0)	0·839
4–6	27 (40·3)	14 (42·4)	41 (41·0)
Cumulative Injections received
6	10 (14·9)	6 (18·2)	16 (16·0)	0·061
5	19 (28·4)	8 (24·2)	27 (27·0)
4	27 (40·3)	14 (42·4)	41 (41·0)
3	38 (56·7)	15 (45·5)	53 (53·0)
2	49 (73·1)	17 (51·2)	66 (66·0)
1	61 (91·0)	24 (72·7)	85 (85·0)
Re-incarcerations
0	43 (64)	23 (70)	66 (66)	0·857
1	14 (21)	6 (18)	20 (20)
2 or more	10 (15)	4 (12)	14 (14)

Abbreviations: ART= antiretroviral therapy; AUDIT= Alcohol Use Disorders Identification Test; PTSD = post-traumatic stress disorder; SD = standard deviation

^*Using Rapid Opioid Dependence Screen; ^Using Addiction Severity Index;

^#Mini International Neuropsychiatric Interview (M·I·N·I·)

⁺AUDIT (Alcohol Use Disorders Identification Test)

Intention-to-Treat (ITT) Analysis: Comparison of viral suppression at baseline and 6 months post-release

Figure 2 depicts the change in VS levels from baseline to 6-months post-release using the ITT analysis in each treatment group and the between group differences at the 6 month time point for both outcomes. At baseline as previously mentioned, there were no differences in VS levels for either the primary or secondary VS outcome as shown in Table 1. The primary VS outcome (<200 copies/mL) did not statistically significantly improve from baseline (64%) to six months (42.4%) for the placebo group (p=0.070) nor did it for the secondary VS outcome (<50 copies/mL) (42.0% to 30.3%, p=0.292); (VS <400 copies/mL also did not improve for the placebo group from baseline 66.7% to six months 42.4%, p=0.030). The XR-NTX group significantly improved from baseline to six months for the primary VS outcome (<200 copies/mL) (48.0% to 64.2%, p=0.024) and for the secondary VS outcome (<50 copies/mL) (31.0% to 56.7%, p=0.001); (VS <400 copies/mL in the XR-NTX group was maintained from baseline 63.6% to six months 53.7%, p=0.260). Furthermore at six months, compared to placebo, participants who received XR-NTX were also significantly more likely to have VS for both the primary (<200 copies/mL) (64.2% XR-NTX vs. 42.4% placebo; p=0.041) and secondary VS outcomes (<50 copies/mL) (56.7% XR-NTX vs. 30.3% placebo; p=0.015; (VS <400 copies/mL at six-months was not significantly different between groups (53.7% XR-NTX vs. 42.2 % placebo, p=0.289)). Of note additional analyses found that no participants in the placebo group who were not on ART at baseline achieved VS, while 14.3% of those not on ART at baseline in the XR-NTX group did achieve VS at that level at the 6 month time-point.

Figure 2.

Intention-to-Treat Analysis: Comparison of viral suppression levels at <200 copies/mL and <50 copies/mL for participants receiving extended-release naltrexone or placebo (N=100)

Legend: XR-NTX (extended-release naltrexone)

Independent Predictors of Achieving Viral Suppression

The independent correlates of achieving VS for the primary and secondary outcomes are provided in Table 2. For the primary outcome (VS <200 copies/ml), both allocation to receive XR-NTX (aOR: 2.68; 95% CI=1.01–7.09) and receipt of three or more injections (aOR: 3.26; 95% CI=1.26–8.47), irrespective of allocation, predicted VS at <200 copies/mL. Treatment allocation to XR-NTX, receiving three or more injections, decreasing levels of alcohol consumption using the alcohol improvement score and white race predicted maximal VS (<50 copies/mL). Additional multivariate analyses (not shown) compared (a) those in the XR-NTX group who received three or more injections compared to (b) the XR-NTX group who received 2 or fewer injections plus all Placebo participants. A second multivariate model was created placing (a) all of the placebo participants who had received 3 or more injections against (b) participants who received placebo with two injections and fewer and all XR-NTX participants. In the first model the XR-NTX with three or more injections was significantly related to VS (<50 copies/mL level; p≤ 0.05), while the second model was not significantly different for high injection placebo participants. Of note as shown in Table 2, concurrent opioid use disorder diagnosis was not found to be associated with explaining the difference in VS at either the <200 copies/mL (p=0.241) or <50 copies/mL outcomes (0.645).

Table 2.

Multivariate Models for Viral Suppression at <200 and <50 copies/mL

Viral Suppression at <200 copies/ml			Viral Suppression at <50 copies/ml
Variables	aOR (95% CI)	p-value	Variables	aOR (95% CI)	p-value
(Intercept)	0·21 (0·04, 1·2)	0·075	(Intercept)	0·12 (0·02, 0·84)	0·032
Placebo	referent		Placebo	referent
XR-NTX	2·68 (1·01, 7·09)	0·047	XR-NTX	4·54 (1·43, 14·43)	0·009
Gender			Gender
Female	referent		Female	referent
Male	1·17 (0·37, 3·71)	0·784	Male	0·47 (0·13, 1·70)	0·250
Transgender	2·09 (0·073, 52·71)	0·653	Transgender	0·00 (0·000, inf)	0·990
Homelessness	0·97 (0·37, 2·52)	0·953	Homelessness	0·45 (0·15, 1·32)	0·148
Number of injections			Number of injections
2 or less	referent		2 or less	referent
3 or more	3.26 (1.26, 8.47)	0.010	3 or more	6.34 (2.08, 19.29)	0.001
Major Depression	0·94 (0·24, 3·59)	0·924	Major Depression	0·77 (0·17, 3·52)	0·735
Cocaine Use Disorder	0·91 (0·74, 1·12)	0·378	Cocaine Use Disorder	0·83 (0·66, 1·04)	0·112
Opioid Use Disorder	0·90 (0·76, 1·07)	0·241	Opioid Use Disorder	0·96 (0·79, 1·15)	0·645
Race/Ethnicity			Race/Ethnicity
Black	referent		Black	referent
Hispanic	0·87 (0·25, 2·99)	0·821	Hispanic	3·65 (0·84, 15·94)	0·085
White	1·56 (0·41, 5·92)	0·514	White	5·37 (1·08, 26·72)	0·040
Alcohol Improvement Score	1·32 (0·98, 1·78)	0·068	Alcohol Improvement Score	1·43 (1·03, 1·98)	0·033
AIC: 137			AIC: 121

Legend: aOR: adjusted odds ratio; CI: 95% confidence interval; XR-NTX: extended-release naltrexone; AIC: Akaike Information Criterion; BOLD represents statistically significant.

Discussion

To our knowledge, this is the first double-blind, placebo-controlled prospective randomized trial that examines whether an evidence-based pharmacotherapy to treat alcohol use disorders, XR-NTX, influences HIV viral suppression in PLH with alcohol use disorders who are transitioning to the community from prison or jail. The key findings from this trial, using an ITT analysis with the most conservative assessments for missing data, supports that both VS (<200 copies/mL) and maximal VS (<50 copies/mL) are more likely to be achieved in PLH receiving XR-NTX than in those receiving placebo. These findings support the use of XR-NTX in PLH with AUDs who are transitioning through prisons and jails, an intervention that may improve linkage to HIV care and promote VS. This is especially important given that as few as 21% and 34% of PLH in Connecticut prisons and jails are linked to HIV care within 14 and 30 days after release to the community respectively,⁴ the minimal time needed to get their ART prescriptions refilled.

To support the ITT findings, multiple logistic regression analyses provides further support for XR-NTX, but provides additional insights and interpretation of the data. Key among them is that receiving more XR-NTX injections results in improved HIV VS levels. This suggests that better retention on the intervention (i.e., monthly injections) was highly correlated with better engagement in HIV treatment and adherence to ART. This is consistent with findings that patients consistently engaged in care have higher levels of VS.^39,40 For maximal VS (<50 copies/mL), two additional factors contributed to the outcome – reductions in alcohol consumption and white race. While considerable health disparities by race/ethnicity persist in the U.S., especially among prisoners and PLH,^41–45 recent studies have not found that race/ethnicity predicts linkage to HIV care after release.⁴ Reductions in alcohol consumption were significantly predictive of maximal VS (<50 copies/mL) and approached significance (p=0.068) for the primary VS outcome (<200 copies/mL). One explanation for these findings may be that prior data suggest that the number of days of alcohol consumption is associated with missing ART doses, not only on drinking days but also for two days post-consumption.⁴⁶ Thus, maximal VS (<50 copies/mL) may have been better achieved due to better reductions in alcohol consumption, which in turn, reduced the number of days that patients missed taking their ART. Of note the majority of participants, almost 90%, were already receiving ART at the time of randomization (baseline or time of release) and there were no significant differences between the groups in those receiving ART. Additionally, at six months we found that no participants in the placebo group who were not on ART at baseline achieved VS at the <50 copies/mL level, while 14.3% of those in the XR-NTX who were not on ART at baseline group did achieve VS at that level at the 6 month time-point, suggesting that possibly XR-NTX through reductions in alcohol use also may have assisted in initiating ART. Thus the combination of ART with reductions in alcohol use via use of a medication (XR-NTX) to reduce alcohol relapse can improve the likelihood of maintaining or achieving HIV VS. While the effect of XR-NTX on VS is likely moderated by reductions in alcohol consumption thereby improving adherence to ART, further research is needed to determine whether these beneficial effects are either fully or partially mediated by antiviral or anti-inflammatory mechanisms.^47–49

Findings from this study have important implications for both individual and public health. Given the magnitude of PLH and AUDs who cycle through U.S. prisons and jails annually, and paucity of associated prescribed medications to reduce alcohol relapse prior to, during, or after incarceration¹¹ suggest that XR-NTX may help PLH with AUDs remain in HIV care and achieve VS after release. XR-NTX, if implemented properly, may markedly improve post-release HIV treatment outcomes. Whether or not these findings can be extended to others in confined settings, like hospitals or addiction treatment programs must be further examined. These findings do, however, have important implications for HIV treatment guidelines. Presently, the International Association of Physicians in AIDS Care (IAPAC) ART adherence guidelines⁵⁰ for criminal justice populations transitioning to the community recommend (1) directly administered antiretroviral treatment (DAART) for released prisoners at high risk of ART non-adherence and for people who inject drugs and (2) treatment of opioid use disorders with opioid agonist therapies like methadone or buprenorphine.⁵⁰ Findings here should extend these recommendations to include the use of XR-NTX for treatment of AUDs in eligible PLH transitioning from incarceration. Improved adherence with XR-NTX will likely reduce the need for DAART in this population.

The 90-90-90 Joint United Nations Programme on HIV/AIDS (UNAIDS) strategy⁵¹ in the U.S. has nearly achieved its first step – HIV diagnosis. Engagement in care and VS, however, fall considerably short. Because XR-NTX appears to exert its influence both on better treatment engagement/retention, but also on VS levels, it can potentially improve both individual health and ‘treatment as prevention’ ideals through higher VS levels. This is especially true in incarcerated PLH with AUDs who engage in high-risk HIV behaviors after release and in whom higher VS levels may translated into reduced transmission.

One of the major impediments of adoption and implementation of an evidence-based treatment for prisoners involves cost. The strongest evidence to improve VS levels for transitioning incarcerated individuals is directly administered antiretroviral treatment (DAART),⁵² yet it is rarely implemented due to limited community resources and elevated costs.⁵⁰ Though pharmacotherapy using methadone or buprenorphine for transitioning incarcerated individuals with HIV and opioid use disorder are recommended and are associated with low cost, they are based on non-experimental designs;²⁸ XR-NTX, however, remains costly. A recently published randomized placebo-controlled trial found that XR-NTX also led to improved VS for incarcerated individuals released to the community with HIV and opioid use disorders.⁵³ Findings from this trial provide the highest level of support for treating AUDs in transitioning incarcerated individuals with HIV and, pending cost-effectiveness analyses; it should be adopted as voluntary treatment of AUDs as part of the U.S. response to controlling the HIV pandemic.

Despite the many new and important findings presented here, some limitations remain, including the sample size, and lack of generalizability to other communities. Poor retention in clinical care, including missing follow-up data from research studies after release from criminal justice settings is common.^10,54 In this study, missing follow-up data were low (13%), but the MCAR analysis allowed us to use the most conservative assumption for comparing VS with missing observations for HIV-1 RNA levels being treated as not achieving VS. This limitation, however, is typically what is considered in real-world treatment settings of PLH where the association between poor retention, particularly missed visits, and poorer biological outcomes evidenced by virological failure and mortality is well established in the literature.^39,40 Using a missing value as “failure” to achieve VS is the most conservative strategy for NOT achieving or maintaining VS. Despite some missingness of viral load data and the relatively small sample size, the findings remain robust even controlling for other factors. Moreover, though including persons with mental illness and polysubstance use likely contributed to elevated attrition levels seen in this study,^16,18 these results are more reflective of the real-world effectiveness of XR-NTX as a community-based treatment.

Conclusions

Monthly-administered XR-NTX is an effective strategy to reduce alcohol consumption in PLH with AUDs transitioning to the community from prisons and jails. The present study extends the benefits of XR-NTX as a treatment for AUDs and supports the additional benefit it has on improving HIV treatment outcomes in combination with ART in this population, specifically the achievement and maintenance of VS and maximal VS, the cornerstone of individual and public health for PLH. This study is the first to demonstrate the multifaceted benefits of XR-NTX in this vulnerable population of incarcerated persons with HIV and AUDs, and adds to the important recent findings of XR-NTX improving VS among incarcerated persons with HIV and opioid use disorder.⁵⁵ Newer longer-acting naltrexone preparations, including implantable naltrexone that lasts for 3 to 6 months, might overcome challenges with returning for monthly injections. Also, though the cost of XR-NTX remains high, one must balance the costs with the societal benefits. Real-world implementation studies are needed to examine its effectiveness. Future work should evaluate cost-effectiveness of XR-NTX in incarcerated individuals transitioning to the community as well as in other community settings where HIV and AUDs are highly prevalent to assess the generalizability of this strategy to improve VS and further, to examine its impact on treatment as prevention efforts.