Quantifying a potential protective effect of buprenorphine on fatality risk during acute fentanyl exposures

Abstract

Introduction.

Buprenorphine is an important therapy for opioid use disorder and may also reduce the risk of fatal overdoses in fentanyl exposures. However, the role of buprenorphine in reducing this risk has not been quantified. This cross-sectional study examined the association between buprenorphine presence, decedent characteristics, and other factors with the predicted fentanyl concentrations in overdose deaths.

Methods.

The study identified unintentional fentanyl overdose decedents (n = 3,036) from the West Virginia Forensic Drug Database, 2011 through mid-2020. The main outcome was fentanyl concentrations in overdose deaths in the presence and absence of buprenorphine. A multiple linear regression model examined the association of fentanyl concentrations with buprenorphine presence based on the concentrations of the parent drug buprenorphine (B) and its metabolite norbuprenorphine (N), adjusting for demographics, toxicological characteristics (presence of multiple opioids, benzodiazepines, stimulants, marijuana, and alcohol), and comorbidities. We used a B/N concentration ratio <1 as an indirect indicator of longer-term buprenorphine exposure prior to drug overdose death.

Results.

The median fentanyl concentration was 65% higher when buprenorphine was present (N=168) vs. absent (N=2,868) (0.028 vs. 0.017 μg/mL, p < .001). In the multivariable model, statistically significant associations occurred between buprenorphine presence and increased fentanyl concentrations (+28.7%) with a B/N ratio <1. Obesity, male sex, alcohol presence, and comorbid cardiovascular diseases were statistically significantly associated with lower (−11.3% to −20.7%) fentanyl concentrations, whereas marijuana presence and a history of substance use disorder were associated with statistically significant higher fentanyl concentrations (+8.8% to +31.3%).

Conclusions.

These findings suggest that sustained or longer-term buprenorphine intake might exert some protective effect on fatalities resulting from fentanyl exposure as documented by the association of higher fentanyl blood concentrations with buprenorphine presence among fatal drug overdoses. As fentanyl availability and overdose rates increase nationally, buprenorphine is a vital tool for effective opioid use disorder treatment that might also reduce the risk of fatality in an acute fentanyl exposure.

1. Introduction

Buprenorphine has been increasingly used to treat opioid use disorder (OUD) due to its safety profile, effectiveness, relatively high degree of patient compliance, and proven effectiveness in reducing deaths (Larochelle et al., 2018; McLean & Kavanaugh, 2019; Wakeman et al., 2020; Woodruff et al., 2019). This use is especially important with the substantial increase in opioid overdose deaths experienced with the COVID-19 pandemic, most of which involved fentanyl (Kuehn, 2021). Buprenorphine is a partial opioid agonist with high binding affinity to the μ-opioid receptor, has slow dissociation kinetics facilitating displacement of full agonists like fentanyl, and causes respiratory depression with ceiling effects (Infantino et al., 2021; Seldén et al., 2012). These properties suggest that buprenorphine may have the potential to reverse opioid overdose (Zamani et al., 2020). Buprenorphine may also reduce the risk of fatal overdose following an acute fentanyl exposure (Wightman et al., 2021), although this risk reduction has not been quantified. To indirectly assess a possible risk reduction, studies measured the concentrations of fentanyl, widely identified in recent overdose deaths (DEA, 2019; O’Donnell et al., 2021), and compared them in fentanyl overdose decedents both with and without buprenorphine presence. Studies have also determined the association of other variables with predicted fentanyl concentrations (Dai et al., 2020).

2. Methods

Our study examined West Virginia (WV) unintentional deaths from 2011 to mid-2020 in which fentanyl was identified as a direct cause of, or contributor to, death. The study obtained decedents’ demographics, medical history, key autopsy findings, and toxicological testing data from the Forensic Drug Database (FDD), which compiles data from all WV drug-related deaths (Dai et al., 2020). Toxicological testing is routinely performed on all deaths investigated by the WV Office of the Chief Medical Examiner, a statewide medical examiner system, with confirmatory tests conducted for most positive screens, including therapeutic and nonprescription drugs. Blood and/or tissue samples are screened for volatile compounds using gas chromatography with flame ionization detection and for drugs of abuse using automated enzyme immunoassays. Screening for fentanyl, buprenorphine, and norbuprenorphine has remained consistent over the study period. The most frequently involved opioids in deaths, including fentanyl and fentanyl analogs (carfentanil, furanyl fentanyl, para-fluoro(iso)butyryl fentanyl, acetyl fentanyl, acryl fentanyl, cyclopropyl fentanyl, butyryl fentanyl, 3-methyl fentanyl, methoxyacetyl fentanyl, tetrahydrofuran fentanyl), buprenorphine, methadone, morphine, codeine, hydrocodone, hydromorphone, oxycodone, and oxymorphone are screened at a routine basis. We determined buprenorphine and norbuprenorphine concentrations using liquid chromatography with tandem mass spectrometry following solid supported liquid extraction. The quantitation limit for fentanyl is 0.0005 μg/mL, and the limits for buprenorphine and norbuprenorphine are 0.0005 μg/ml and 0.001 μg/ml, respectively. Femoral or subclavian blood samples were used for all concentration analyses. The study used an immunoassay specific for carboxy-THC to screen for the presence of marijuana (delta-9-THC), with confirmatory testing performed as appropriate. Additional details of testing methods and thresholds for other substances are reported elsewhere (Dai et al., 2020).

We compared fentanyl concentrations in deaths by buprenorphine presence or absence, and buprenorphine (B) and its metabolite norbuprenorphine (N) blood concentrations when available. We created a flow chart (Supplemental Figure 1) to illustrate the number of fentanyl deaths and numbers of cases with available buprenorphine and norbuprenorphine concentrations. Since buprenorphine has a long elimination half-life (20–32 hours) (Häkkinen et al., 2013), higher N concentrations relative to parent drug B concentrations in a decedent indicate that buprenorphine was unlikely to be involved in an acute overdose case, but had been taken more long-term, since metabolism to N had sufficient time to occur (Infantino et al., 2021; Seldén et al., 2012). Studies have used B/N concentration ratios to indicate whether buprenorphine exposure likely represented acute intake or longer-term exposure. A cutoff value of 1 for the B/N concentration ratio has been proposed to distinguish longer-term buprenorphine use from acute exposure. A B/N ratio ≥1 would indicate likely acute buprenorphine exposure (i.e., high B concentrations with relatively low N concentrations) since little time might have been available prior to death for metabolism to N. A ratio <1 (higher N relative to B concentrations) would more likely represent non-acute, longer-term exposure with considerable metabolism to N present (Häkkinen et al., 2013; Seldén et al., 2012). The study also compared fentanyl concentrations at B/N ratios ≥1 or <1 to help differentiate possible effects of acute vs. longer-term buprenorphine exposure. We plotted distributions of deaths by cumulative fentanyl concentrations (0.005 μg/mL intervals) with and without buprenorphine by B/N ratio. Wilcoxon rank sum tests compared median fentanyl concentrations. A multivariable model examined associations between buprenorphine presence and postmortem fentanyl concentrations, accounting for demographics, toxicological characteristics, comorbid cardiovascular disease, and history of substance use disorder. Fentanyl concentration outcomes were log transformed due to skewed distributions. Average percent changes in predicted fentanyl concentrations were calculated based on the corresponding coefficient $\mathrm{\beta }$ using the formula $({\mathrm{e}}^{\mathrm{\beta }}-1)*100\%$ to present the findings in a clinically relevant manner. The study defined statistical significance as p <0.05. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies. The West Virginia University IRB deemed the study exempt. Study staff performed all analyses using SAS 9.4 (SAS Institute, Cary, NC).

3. Results

Of the 3,036 unintentional fentanyl overdose decedents in WV, 168 (5.5%) were buprenorphine positive (Table 1). Of the cases in which buprenorphine was identified, a third were less than 35 years of age, with two-fifths between 35–44 years old; nearly 30% were obese; almost two-thirds were male; 99% were white, and 83 of the 168 (49.4%) had a B/N ratio < 1. The median fentanyl concentration was 65% higher when buprenorphine was present vs. absent (0.028 vs. 0.017 μg/mL, p <.0001). The B/N ratio could be determined in 61% of cases (102/168), with a B/N ratio ≥1 in only 19 (11.3%) of the buprenorphine positive cases. Compared to when buprenorphine was absent, higher fentanyl concentrations were found in the presence of buprenorphine and for both B/N ratios (p<0.0001). The median fentanyl concentration with a B/N ratio <1 (0.030 μg/mL) was not statistically significantly different than the median concentration with a B/N ratio ≥1 (0.026 μg/mL) (p=0.2943). Buprenorphine concentrations overall ranged from 0.0005 to 0.038 μg/mL (Table 2). When cumulative percentages of fentanyl deaths by B/N ratios were plotted against fentanyl concentrations (0.005 μg/mL intervals, Figure 1), a “right shift” occurred in fentanyl concentrations compared to no buprenorphine. This shift demonstrates that if buprenorphine was present, higher concentrations of fentanyl were needed to reach the same proportion of deaths if no buprenorphine was present, especially for B/N ratios < 1. For example, at the 90^th percentile for cumulative fentanyl concentrations in the absence of buprenorphine, the fentanyl concentration was about 0.05 μg/mL, but the fentanyl concentration was doubled (about 0.1 μg/mL) at the 90^th percentile for fentanyl deaths with a B/N ratio <1. The fentanyl concentration was about 30% higher at the 90^th percentile for cumulative fentanyl deaths when comparing fentanyl with no buprenorphine present vs. B/N ≥ 1.

Table 1.

Fentanyl concentrations by buprenorphine presence and ratio of buprenorphine concentration versus norbuprenorphine concentration (N = 3036).

Opioid concentrations by buprenorphine presence	N (%)	Mean (SD), μg/mL	Median (Min, Max), μg/mL	P-valuea
Fentanyl	3036	0.028 (0.067)	0.017 (<0.001, 0.620)
Buprenorphine absent	2868 (94.5)	0.027 (0.069)	0.017 (<0.001, 0.620)
Buprenorphine presentb	168 (5.5)	0.037 (0.036)	0.028 (0.001, 0.230)	<0.0001
B/N ratio < 1	83	0.038 (0.035)	0.030 (0.001, 0.157)	<0.0001
B/N ratio ≥ 1	19	0.029 (0.027)	0.026 (0.001, 0.084)	0.2943

^aComparisons of buprenorphine presence and B/N ratios to buprenorphine absence; Wilcoxon rank sum test; P-values smaller than 0.05 are highlighted in bold.

^bExcluded 66 cases in which the B/N ratio could not be determined.

Table 2.

Buprenorphine decedent concentrations (μg/mL).

Substance	Na	Minimum	Maximum	Median [IQR]
Buprenorphine (B)	119	0.0005	0.038	0.0011 [0.0008, 0.0025]
Norbuprenorphine (N)	136	0.0005	0.022	0.0023 [0.0014, 0.0038]

^aIn some deaths, the concentration of a parent drug (buprenorphine) or a metabolite (norbuprenorphine) cannot be quantified.

Figure 1.

Cumulative Fentanyl Concentration Distributions Overall and By Buprenorphine/Norbuprenorphine (B/N) Ratio

The multivariable model (Table 3) found a statistically significant association between buprenorphine presence and increased fentanyl concentrations for a B/N ratio <1 (+28.7%), but a non-statistically significant decreased predicted fentanyl concentration with a B/N ratio ≥1 (−17.4%) compared to deaths without buprenorphine. Obesity (−13.5%), male sex (−12.1%), alcohol presence (−20.7%), and co-morbid cardiovascular disease (−11.3%) were statistically significantly associated with lower fentanyl concentrations, while marijuana presence (+31.3%) and a diagnosis of substance use disorder (+8.8%) were associated with significantly higher predicted fentanyl concentrations. The presence of other opioids (e.g., fentanyl analogs, oxycodone, etc.), benzodiazepines (e.g., alprazolam, diazepam, etc.), and stimulants (e.g., methamphetamine, cocaine, etc.) were not significantly associated with changes in fentanyl concentrations.

Table 3.

Average percent change in predicted fentanyl concentrations (μg/mL).^*

Variables	N(%)	Coefficient	95 % Confidence Interval	Percent change
Obesity [BMI ≥ 30] (ref = BMI < 30)	1005 (34.2 %)	−0.145	−0.148, −0.142	−13.50 %
Male (ref = female)	2193 (74.7 %)	−0.129	−0.131, −0.127	−12.10 %
Ethanol presenta (ref = ethanol absent)	442 (15.1 %)	−0.232	−0.237, −0.227	−20.70 %
Buprenorphine present (ref = buprenorphine absent)	102 (3.5%)
B/N ratio < 1	83 (2.8 %)	0.252	0.226, 0.278	28.70%
B/N ratio ≥ 1	19 (0.6 %)	−0.192	−0.303, −0.081	−17.40 %
Other opioids presentb (ref = no other opioids)	1277 (43.5 %)	−0.013	−0.015, −0.011	−1.30 %
Benzodiazepine presentc (ref = benzodiazepines absent)	752 (25.6 %)	−0.022	−0.025, −0.019	−2.20 %
Stimulant present (ref = stimulant absentl	1405 (47.9 %)	−0.049	−0.051, −0.047	−4.80%
Marijuana present (ref = marijuana absent)	168 (5.7%)	0.272	0.259, 0.285	31.30%
Cardiovascular diseases (ref = No cardiovascular diseases)	654 (22.3 %)	−0.12	−0.122, −0.118	−11.30 %
History of subsunre use disorder (ref = No history of SUD)	2816 (95.9 %)	0.084	0.082, 0.086	8.80%

Bold indicates statistical significance p < 0.05 front the model.

^*N = 2936; some cases were removed from modeling due to missing variables.

^aEthanol presence was defined as a presence of ethanol concentrations in cases where ethanol was considered a cause or contributor to death.

^bExcluding buprenorphine and fentanyl, including fentanyl analogs, heroin, oxycodone, hydrocodone, and other opioids.

^cIncludes alprazolam, clonazepam, chlordiazepoxide, diazepam, lorazepam, oxazepam, temazepam.

4. Discussion

Buprenorphine presence was associated with a significantly higher predicted fentanyl concentration in fatalities compared to when it was absent, suggesting that when buprenorphine is present higher doses of fentanyl are needed to result in a fatality. Multivariable analyses controlling for other factors also found fentanyl concentrations significantly higher by almost 29% among those likely taking longer-term buprenorphine (B/N ratio <1) before a fatal fentanyl poisoning. With possible acute buprenorphine intake (B/N ratios ≥1) in the presence of fentanyl, the negative association might be indicative of additional respiratory depression, although this association was not statistically significant. No definitive conclusions can be made due to the small number of cases with higher B/N ratios (n=19). Our finding of a significantly higher fentanyl concentration with a B/N ratio <1 is consistent with a partial protective effect of nonacute buprenorphine exposure on fentanyl overdosage, i.e., relatively more fentanyl is needed to result in a fatal outcome in these cases, as the cumulative distribution of fentanyl concentrations illustrates (Figure 1). This finding provides the first quantitative evidence of buprenorphine’s risk reduction for fatalities suggested in earlier studies (Wightman et al., 2021; Zamani et al., 2020) and is consistent with a recent clinical trial that found taking sustained buprenorphine protects against fentanyl-induced respiratory depression (Moss et al., 2022).

The largest significant reduction in predicted fatal fentanyl concentrations was associated with ethanol presence (20% reduction), consistent with an increased fatality risk when combining central nervous system depressants (Dai et al., 2020). It is unclear why multiple opioid ingestion in the current study was not associated with a greater decrease in predicted fentanyl concentrations. Our previous study found significant reductions in predicted fentanyl concentrations in the presence of multiple opioids; however, we did not consider possible effects from buprenorphine involvement (Dai et al., 2020). High fentanyl potency relative to the potency of other prescription opioids might also help explain why a significant association was not seen between the presence of these opioids and predicted fentanyl concentrations. Our findings are also consistent with previous studies indicating that obesity and heart disease are independent risk factors for opioid overdose (Archibald et al., 2021; Boscarino et al., 2022).

Presence of marijuana (delta-9-THC) was however independently associated with a substantial increase of almost one-third in the predicted fentanyl concentration. Increased buprenorphine concentrations with cannabis use were recently reported, possibly because buprenorphine is a substrate of CYP3A4 which can be inhibited by cannabis constituents (Vierke et al., 2021). Fentanyl is also a substrate of CYP3A4 and a potential cannabidiol interaction has been documented (Duragesic Prescribing Information, 2018). However, the source or administration route of marijuana exposure was generally not available for decedents. The relationships among marijuana use, fentanyl concentrations, and fentanyl toxicity potential should be further explored.

The strength of the study includes the use of a comprehensive database that incorporates information from multiple sources and toxicological analyses to allow for estimation of the “risk modification” of buprenorphine on fentanyl fatalities that has been hypothesized in earlier studies (Wightman et al., 2021; Zamani et al., 2020). Limitations of this study include the possibility of postmortem redistribution affecting the interpreting of fentanyl drug concentrations. However, our analyses only included peripheral blood samples (femoral, subclavian) to lessen the impact of this redistribution (Olson et al., 2010). Some cases (39%) were missing norbuprenorphine or buprenorphine concentrations which prevented determination of a B/N ratio. Findings from forensic data collected in a single state might also not be easily generalized into other states or nationwide, due to variability in how medical examiners identify and report drugs as causes of death. Our study only examines decedents and lack of prevalence data regarding buprenorphine use among illicit opioid users prevents comparison of B/N ratios in these individuals with those ratios found in the deaths. Although West Virginia utilized a centralized medical examiner system to help ensure consistent drug screening across cases, incomplete, inconsistent, or missing data may still exist. Due to the inability to access complete medical records for all decedents, the impact of opioid tolerance and dependence, or other individual pharmacodynamic variability, on fentanyl concentrations observed could not be accounted for. Future research plans include repeating the modeling analysis of fentanyl deaths when a larger number of deaths involved buprenorphine, especially those with higher B/N ratios. Combining our forensic data with similar data from other states would also be useful to increase the overall sample size.

5. Conclusions

The association of higher fentanyl blood concentrations following fentanyl-associated fatalities when buprenorphine was present provides potential quantitative evidence for reduced fentanyl overdose risk with concurrent buprenorphine therapy. As fentanyl availability and overdose rates increase nationally, treatment of OUD with buprenorphine remains the most effective way to reduce overdose (Larochelle et al., 2018; Samples et al., 2023) and based on our findings it may also reduce fatalities from acute fentanyl exposures. The potential impact of marijuana taken with fentanyl and buprenorphine needs further exploration.

Highlights:

A statistically significant association was found between buprenorphine presence and increased postmortem fentanyl concentrations in fentanyl-associated deaths.

A predicted 28.7% increase in fentanyl concentrations was associated with sustained buprenorphine use, suggested by a ratio of buprenorphine to norbuprenorphine (metabolite) less than 1.

Buprenorphine may help reduce the risk of fatal overdose in fentanyl exposures.