Combination refers to patients consuming MC inflorescence concomitantly with MC oil extract; \* administration methods do not add up to 100% due to concomitant routes. Note: CI, confidence interval; IQR, interquartile range; BMI, body mass index; MC, medical cannabis.


#### **Table 2.** Migraine features.

Note: CI, confidence interval; IQR, interquartile range; +, positive for this manifestation.


**Table 3.** Sleep characteristics.

\* Components of the PSQI questionnaire global score. Note: CI, confidence interval; IQR, interquartile range; PSQI, Pittsburgh sleep quality index.

#### *3.3. MC Treatment Safety*

MC-related adverse effects (AEs) were reported by 37% (*n* = 53) of the sample. Notably, non-responders reported higher incidences of any AEs (*n* = 26, 46%) than responders (*n* = 27, 30%) (0.46 95% CI (0.21 to 0.99), *p* < 0.05). Most of the specific AEs did not vary significantly between responders and non-responders. However, itchy and red eyes (*n* = 8, 9%, for both) were reported only in the responder subgroup (χ<sup>2</sup> (1) = 6.9, *p* < 0.01 for both). Additionally, dry mouth was reported at higher rates among the responders (*n* = 9, 10%) than by non-responders (*n* = 2, 4%) (χ<sup>2</sup> (1) = 3.9, *p* < 0.05).

In descending order of frequency, reported AEs included central nervous system AEs (*n* = 33, 23%), psychological AEs (*n* = 21, 14%), ophthalmic AEs (*n* = 16, 11%), gastrointestinal AEs (*n* = 15, 10%), musculoskeletal AEs (*n* = 11, 8%), cardiovascular AEs (*n* = 10, 7%), and auditory AEs (*n* = 9, 6%).

We further evaluated the associations between MC administration routes and AEs. There were no significant differences between patients reporting MC-related AEs and MC administration routes (i.e., inflorescence, oil extract, or a combination of these administration routes) (0.08 95% CI (0 to 0.25); *p* = 0.59). Additionally, no differences were observed between the different consumption methods (e.g., smoking, vaping, sublingual etc., *p* > 0.05).

#### *3.4. MC Treatment Complexity*

The complexity of MC treatment in Israel is due to the variety of available cultivars in Israel (about 100 different cultivars or "strains") and the options for patients to consume more than one cultivar in the same month, with varying doses of each cultivar. Consequently, the 68 patients in the current study reported consumption of 50 different MC cultivars combinations were reported in the current study by the 68 patients we had full cultivar lab information on. Notably, 46 (92%), 1 (2%) and 3 (6%) of the 50 possible combinations were compiled of cultivars that were THC-, CBD-dominant or contained equally high contents of THC:CBD, respectively. These 50 combinations comprised 38 unique cultivars. Figure 2 shows a *z*-score clustered heatmap of the main phytocannabinoids (presented as total concentrations in % w/w) in the 38 cultivars consumed by the sample subgroup. Based on the phytocannabinoid concentration variability, these cultivars were clustered to nine different groups. Figure 2 also shows that in the combinations of cultivars consumed, ten cultivars were consumed only by responders, eight cultivars were consumed only by non-responders, and the rest of the cultivars (*n* = 20) were consumed by both groups.

**Figure 2.** Relative phytocannabinoid concentrations in the most frequently consumed cultivars. Colors on the graph represent the scaled phytocannabinoid concentration variations between cultivars; the numbers in each box represent the concentration (% w/w) of the specific phytocannabinoid within each cultivar. Note: \* for each phytocannabinoid, the concentrations of the acid and its neutral counterpart were summed and reported as the total content; Method used: package "pheatmap", function pheatmap, with the "Euclidean" (default) distance measure used in clustering rows, "complete" clustering method used on *z*-scored data scaled by row. Note: THC, (-)-Δ9-*trans*-tetrahydrocannabinol; CBD, cannabidiol; CBC, cannabichromene; CBG, cannabigerol; CBN, cannabinol; THC-C4, (-)-Δ9-*trans*-tetrahydrocannabinol-C4; THCV, (-)-Δ9-*trans*-tetrahydrocannabivarin.

#### *3.5. MC Treatment Characteristics*

In this subgroup analysis we included data only from patients who smoked or vaped MC inflorescences and not those who consumed oil extracts sublingually, in order to avoid comparing between different routes of administration (different pharmacokinetics). Since the inflorescences in this study were analyzed in their natural form, monthly consumption of phytocannabinoid doses were calculated according to total phytocannabinoid concentrations rather than analyzing separate acid or neutral concentrations, in order to simulate the neutral maximum content of phytocannabinoids consumed following smoking or vaporization. This calculation corrects for any differences that may arise in phytocannabinoid profiles as a result of decarboxylation due to mishandling or storage of the MC inflorescences. Thus, the minority of patients that reported sublingual consumption of oil extract (*n* = 12) or combined these with inflorescences (*n* = 19) were not included in this subgroup analysis. Consequently, 68 (47%) patients reported exclusive MC inflorescence consumption via inhalation. Of these, 45 (66%) of them were responders and 23 (34%) were non-responders.

For the abovementioned 68 patients, we first evaluated the differences in total MC monthly dose between responders and non-responders. No significant differences were found (30 (20–40) g and 30 (20–45) g, respectively) (0.25 95% CI (−0.26 to 0.76); *p* = 0.97) (Figure 3A). Therefore, we evaluated the impact of the monthly doses of specific phytocannabinoids. As the distribution of monthly doses of specific phytocannabinoids were non-normal, we separated specific phytocannabinoids into low and high monthly dose groups, based on the distribution of consumption in our patient sample.

**Figure 3.** (**A**–**C**) Phytocannabinoid dose differences between responders and non-responders. Note: MC, medical cannabis. Response refers to reduction in the monthly migraine attacks frequency following the initiation of MC treatment (i.e., ≥50%) compared to non-responders (i.e., <50%).

We found that responders were more likely to consume high doses (7.9–109.5 mg per month) of the phytocannabinoid ms\_373\_15c (*n* = 27, 60%) and low doses (0–9.9 mg per month) of the phytocannabinoid ms\_331\_18d (*n* = 28, 62%) compared to non-responders, who were more likely to consume low doses (0–7.8 mg per month) of ms\_373\_15c (*n* = 16, 70%) and high doses (10.0–46.8 mg per month) of ms\_331\_18d (*n* = 17, 74%) (3.40 95% CI (1.10 to 12.00); *p* < 0.05 and 0.22 95% CI (0.05 to 0.72); *p* < 0.01, respectively) (Figure 3B,C). The other phytocannabinoids monthly doses did not vary significantly between the subgroups. Importantly, no differences were found between responders and non-responders in the daily frequency of MC consumption (5 (2.5–7) times per day and 4.5 (3–6) times per day, respectively) (0.18 95% CI (−0.34 to 0.71), *p* = 0.99). Additionally, no

differences were found in the number of monthly cannabis cultivars combinations (2 (1–2) cultivars, respectively) (0.04 95% CI (−0.47 to 0.56), *p* = 0.99). Interestingly, among the 38 unique cultivars that patients consumed in their combinations, 12 contained considerable amounts of ms\_373\_15c and none or very low amounts of ms\_331\_18d. These cultivars appeared more frequently among the responders (42 appearances in cultivar combinations) than the non-responders (14 appearances in cultivar combinations).

#### *3.6. Migraine Treatment Characteristics*

A total of 65 (45%) of the patients reported any current consumption of pharmaceutical analgesic medications. Although not significant (0.51 95% CI (0.23 to 1.10), *p* = 0.09), more of the non-responders (*n* = 30, 54%) reported consumption of analgesics compared to the responders (*n* = 35, 39%). Nonetheless, there was a significant difference in the type of analgesic intake between the two groups. Non-responders consumed significantly higher rates of weak opioids (*n* = 13, 23%; e.g., tramadol hydrochloride, buprenorphine hydrochloride, etc.), strong opioids (*n* = 14, 25%; e.g., oxycodone hydrochloride, fentanyl, etc.), and triptans (*n* = 9, 16%; e.g., sumatriptan, rizatriptan, etc.) compared to responders (*n* = 4, 5%; *n* = 7, 8% and *n* = 4, 5%, respectively) (0.15 95% CI (0.03 to 0.53); *p* < 0.005, 0.25 95% CI (0.07 to 0.72); *p* < 0.005 and 0.24 95% CI (0.05 to 0.93), *p* < 0.05). No statistically significant variations were found between responders and non-responders in the consumption rates of over-the-counter analgesics, NSAIDs, anticonvulsants, antidepressants, and antiemetics.

#### **4. Discussion**

In this cross-sectional study, we evaluated patient reports on the frequency of their monthly migraine attacks, both pre- and post-MC treatment. Patients were classified as responders if they reported greater than 50% reduction in monthly migraine attacks post-MC treatment. As expected, responders reported lower current migraine disability and lower negative impact compared to non-responders.

A recent retrospective study conducted by Rhyne et al. (2016) showed that migraine patients who inhaled MC had a significant reduction in migraine frequency [30], which is in line with the results demonstrated here, and supports our finding of high rates of patient reporting of migraine frequency reduction. Migraine is classified as a pain condition. Mechanistically, endocannabinoids have been shown to have an inhibitory effect on serotonin receptors in vivo [31], which is shown to modulate pain and emetic responses. Additional in vivo data showed that THC induced an antinociception effect on the periaqueductal gray matter [32], which is believed to be involved in migraine pathophysiology [33]. Moreover, relatively low levels of the endocannabinoid anandamide (AEA) in the cerebral spinal fluid (CSF) were found to be associated with the mechanism of migraine initiation [11]. A reduction in pain in in vivo models following endocannabinoid [31] and cannabinoid [32] treatments supports our finding regarding a reduction of migraine disability in the responders group. Nonetheless, these studies still do not incorporate all the complexities of whole-plant cannabis treatment.

In this study, responders reported better migraine disability status, less negative headache impact, and better sleep quality. Whilst this result is logical, conflicting results were reported in another cross-sectional study, which demonstrated an association between improved headache disability and migraine intensity, but found no such association with headache frequency [34]. Taken together, our findings suggest that improved migraine disability status and negative impact among MC treatment responders might be attributed directly to MC treatment effects, rather than being secondary to the reduction of the frequency of migraine attacks. Here, we also reported an association between patients with poor sleep quality and less responsiveness to MC treatment in reducing the frequency of migraine attacks. A previous cross-sectional study demonstrated similar results, showing that even without MC treatment, an association was found between poor sleep quality and higher migraine attack frequency [35]. Thus, it might be suggested that migraineurs that responded to MC treatment and demonstrated a decrease in their monthly migraine frequency also had a concurrent sleep quality improvement. However, due to the current study design, we are unable to conclude whether the reported improved sleep quality can be attributed to the decrease in monthly migraine attack frequency or directly due to MC treatment effects.

There is increasing evidence that MC treatment has opioid-sparing effects [36–40]. Here, we found that responders to MC treatment also reported lower rates of consumption of opioids and triptans compared to non-responders. Both opioids and triptans are usually prescribed for migraine pain relief and not for prevention [6,41]. Thus, although we do not have information regarding the use of these medications prior to MC treatment initiation, this might be an indication that patients that responded clinically to MC treatment substituted this conventional treatment for MC.

In this study, we evaluated the differences in relative monthly dose of phytocannabinoids in each cultivar consumed, in both the responders and non-responders groups. To the best of our knowledge, this is the first study to assess the dose consumption of a wide variety of specific phytocannabinoids administered in combinations of cultivars. By doing so, we were able to elucidate associations between specific cannabinoids consumed over a monthly dose and the clinical response of migraine frequency reduction following MC treatment initiation. The most novel finding of this study was the identification that higher rates of patients that reported significant migraine frequency reduction following MC treatment also consumed higher monthly doses of ms\_373\_15c and lower monthly doses of ms\_331\_18d. Our group has previously identified these compounds in both THC- and CBD-dominant chemovars according to LC/MS/MS [18,19], however their absolute structure still needs to be elucidated. According to their MS/MS fragmentation spectra, ms\_373\_15c and ms\_331\_18d are acidic and neutral phytocannabinoids, respectively. Additionally, we identified specific cultivars that contain this favorable ratio between those compounds. However, it is important to note that we cannot attribute the anti-migraine effect of MC specifically to these phytocannabinoids, as we are yet to understand whether they are biological active. Nevertheless, we suggest using the presence of these phytocannabinoids to help in choosing specific MC chemovars for further research. Unfortunately, due to the relatively small sample size of patients in this study and a large number of cultivars with variable chemical constituents, translating these findings to the clinical setting will require a larger sample size and a more comprehensive approach. However, the work presented here could be the foundation of such a study to include these "lesser known" phytocannabinoid compounds. Currently, there are no clinical trials on migraine and MC [42]. Previous studies on migraine did not assess the phytocannabinoids mentioned in our study [43], and usually regarded "cannabis" as a single adherent medication [30], therefore disregarding the inherent complexity in MC treatment, with differences in over 90 phytocannabinoids [18] between cannabis cultivars [44].

We also found that the incidence of MC-related AEs was higher among non-responders. This may be explained by responders tolerating MC-related AEs better than non-responders. It could also be explained by the responders' success during trial-and-error to identify a specific MC chemovar that provided them relief with lower rates of AEs. Nevertheless, due to our study design, we could not corroborate these findings. Future studies should, therefore, investigate the association between MC-related AEs and treatment response a priori. Importantly, none of the patients reported aggravation of migraine AEs as a result of MC treatment.

#### *Limitations*

There are four limitations in the current study. Firstly, our results may have been biased by the small sample size; non-parametric models were used to balance this limitation. Secondly, there may be self-reporting bias. Participants were able to respond to the questionnaire under strict anonymity, ensuring there were no risks that their current treatment plan may be altered by their physician. The questionnaire has also been validated. Thirdly, since we cannot compare the initial indications for which responders and non-responders obtained their MC license, it is possible that the presented results have been biased. Nonetheless, since we identified that chronic neuropathic non-cancer pain was the predominant indication for obtaining MC license, we assumed that it is unlikely that differences

between the subgroups exist. Lastly, since the frequency of migraine attacks prior to MC treatment was reported in retrospect, recall bias might have occurred.

#### **5. Conclusions**

Migraine is currently not indicated for a MC treatment license in Israel. Nevertheless, in some cases it falls under the approved chronic neuropathic non-cancer pain indication, making it possible to study migraine more extensively. In this study, we demonstrated that patients responding to MC treatment also reported less disability and lower conventional anti-migraine medications intake. Additionally, we highlighted the importance of recognizing and analyzing the doses of the pronounced MC constituents consumed by patients, which in turn allowed us to better understand MC treatment associations with reduction in migraine attacks frequency. We also identified specific cultivars that contain the favorable ratio of compounds that were associated with migraine frequency reduction. These results might shed light on the beneficial effects of MC on migraine and motivate future studies to utilize a cannabis cultivar with the specific phytocannabinoids mentioned here. This additional work could validate our results and possibly support making migraine an approved indication for MC license in Israel.

**Author Contributions:** Conceptualization, J.A., G.M.L., E.E., and D.M.; data curation, J.A.; formal analysis, Y.V.; funding acquisition, D.M.; investigation, J.A.; methodology, J.A., Y.V., P.B., G.M.L., E.E., and D.M.; project administration, D.M.; resources, D.M.; supervision, G.M.L. and D.M.; writing—original draft, J.A.; writing—review and editing, J.A., Y.V., P.B., G.M.L., E.E., and D.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** The study was funded by the Evelyn Gruss Lipper Charitable Foundation, Lauren Frank Rose Donation. This sponsor had no role or influence on the study or on this submission.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **E**ffi**cacy of Dronabinol for Acute Pain Management in Adults with Traumatic Injury: Study Protocol of A Randomized Controlled Trial**

#### **Claire Swartwood 1, Kristin Salottolo 2, Robert Madayag 2,3 and David Bar-Or 2,\***


Received: 3 February 2020; Accepted: 10 March 2020; Published: 12 March 2020

**Abstract:** Delta-9-tetrahydrocannabinol (Δ9-THC) and other cannabinoids present in cannabis (marijuana) have been shown to affect the normal inhibitory pathways that influence nociception in humans. The potential benefits of cannabinoids as an analgesic are likely greatest in hyperalgesic and inflammatory states, suggesting a role as a therapeutic agent for treating acute pain following injury. Dronabinol is a licensed form of Δ9-THC. The primary objective of this single center randomized controlled trial is to evaluate the efficacy of adjunctive dronabinol versus control (systemic analgesics only, no dronabinol) for reducing opioid consumption in adults with traumatic injury. Study inclusion is based on high baseline utilization of opioids ≥50 morphine equivalents (mg) within 24 h of admission for adults aged 18–65 years with traumatic injury. There is a 48-hour screening period followed by a 48-hour treatment period after randomization. A total of 122 patients will be randomized 1:1 across 2 study arms: adjunctive dronabinol versus control (standard of care using systemic analgesics, no adjunctive dronabinol). Patients randomized to the dronabinol arm should receive their first dose within 12 h of randomization, with a dose range of 5 mg up to 30 mg daily in divided doses, in addition to systemic analgesics as needed for pain. The primary efficacy endpoint is a change in opioid consumption (morphine equivalents), assessed post-randomization (48 h after randomization) minus pre-randomization (24 h prior to randomization). This is the first randomized trial to investigate whether adjunctive dronabinol is effective in reducing opioid consumption in acute pain management of traumatic injury. Trial Registration: ClinicalTrials.gov Identifier: NCT03928015.

**Keywords:** Delta-9-tetrahydrocannabinol; dronabinol; marijuana; randomized controlled trial; opioids; traumatic injury

#### **1. Introduction**

Delta-9-tetrahydrocannabinol (Δ9-THC) and other cannabinoids present in cannabis (marijuana) have been shown to affect the normal inhibitory pathways that influence nociception in humans. Cannabinoids act through the binding of two cannabinoid receptors coupled through G proteins; CB1 receptors are predominantly found at central and peripheral nerve terminals, where they mediate transmitter release, while CB2 receptors are highly expressed throughout the immune system [1].

The evidence demonstrating a therapeutic effect of THC and cannabis-based medications is still emerging but is well established for treating chronic pain based on three influential peer-reviewed publications [2–4]. These publications also provide conclusive evidence for a therapeutic effect of cannabis-based medications as anti-emetics and for multiple sclerosis symptoms. There is moderate evidence for improving sleep outcomes associated with sleep apnea, fibromyalgia, multiple sclerosis, and chronic pain. There is insufficient or low-quality evidence in all remaining conditions that have been studied. For instance, there is a dearth of research on cannabinoid use for acute pain management. A 2017 systematic review identified seven randomized controlled trials (RCT) assessing the analgesic efficacy of cannabinoid medications for acute pain [5]. Of these studies, five RCTs demonstrated that cannabinoids were equivalent to placebo, in one RCT cannabinoids were superior to placebo, and in one RCT cannabinoids were inferior to placebo. These limited and inconsistent data justify the necessity to perform additional studies on the analgesic effects of cannabinoids for acutely painful conditions.

Patients commonly experience severe, acute pain following traumatic injury that is treated with analgesics, particularly opiates. The antinociceptive properties of cannabinoids may be greatest in hyperalgesic and inflammatory states, suggesting a therapeutic role for treating pain following injury [6]. Moreover, pre-clinical studies support a potential role of Δ9-THC and cannabinoids as an adjunctive agent to opioids in painful conditions, via synergistic enhancement of mu opioid antinociception as well as the prevention of tolerance to and withdrawal from opiates [7–9].

Recently published preliminary clinical research from our group examined the effect of adjunctive dronabinol for acute pain management among 66 trauma patients [10]. Cases demonstrated a significant reduction in opioid consumption (morphine equivalents) from baseline with adjunctive dronabinol (−79 mg, *p* < 0.001), while the change in opioid consumption for matched controls was unchanged from baseline (−9 mg, *p* = 0.63), resulting in a nine-fold greater reduction in opioid consumption for cases versus controls that was significantly different between pairs (difference: −70 mg, *p* = 0.02). There were no differences in secondary outcomes. These results suggest that adjunctive dronabinol used as part of a multimodal analgesia regimen may result in a marked reduction in opioid consumption

Two subset analyses of this matched cohort study provide mixed evidence that the opioid sparing effect of dronabinol may be greater in patients who are marijuana users. Among the subset of 19 cases who were marijuana users, opioid consumption was significantly reduced with adjunctive dronabinol (−97 mg, *p* < 0.001) versus no change in opioid consumption in 19 matched controls (1 mg, *p* = 0.70), with a difference between pairs that was significant: −108 mg, *p* = 0.01) [10]. However, when examining the subset of patients who received dronabinol, there were no differences in the change in opioid consumption for patients who were marijuana users (*n* = 21, −97 mg reduction with dronabinol) compared to non-marijuana users (*n* = 15, −64 mg reduction with dronabinol), *p* = 0.41 (unpublished).

We are recruiting patients in a RCT to evaluate the efficacy of adjunctive dronabinol on opioid utilization for acute pain management. The primary trial objective is to evaluate the efficacy of adjunctive dronabinol versus control (systemic analgesics only, no dronabinol) for reduction in opioid consumption in adults with traumatic injury. Dronabinol is a licensed form of Δ9-THC. Dronabinol is not FDA approved for acute pain management; however, it has been in use at our level I trauma center system formulary without restriction since 2015.

#### **2. Materials and Methods**

#### *2.1. Study Design and Setting*

This is an open label RCT being performed at a single level I trauma center: St. Anthony Hospital in Lakewood, CO. This RCT was designed primarily to determine whether adjunctive dronabinol reduces opioid consumption compared to control. The study was designed with a stratified randomization by baseline marijuana use, which is intended to determine whether the treatment effect of dronabinol is greater in chronic marijuana users compared to recreational or non-marijuana users. This stratified randomization design was incorporated based on the gestalt that cannabis-based medication has a greater benefit for marijuana users.

There is a 48-hour screening/randomization window, a 48-hour treatment window, and a total participation period extending through the acute hospitalization. A description of the clinical trial is posted at ClinicalTrials.gov.

#### *2.2. Study Subjects*

Patients are being recruited from the participating trauma center to which they are acutely presenting. A total of 122 adult trauma patients will be randomized 1:1 across 2 study arms: adjunctive dronabinol or control (systemic analgesics only), as shown in Figure 1.

**Figure 1.** Subject Disposition.

Patients should fulfill all of the following inclusion criteria:


**Table 1.** Oral morphine milligram equivalents (MME) conversion factors.


<sup>1</sup> Formula: Strength per Unit X (Number of Units/Days Supply) X MME conversion factor = MME/Day.

Patients fulfilling one or more of the following criteria may not be enrolled in the study:


#### *2.3. Study Visits*

The following procedures will be performed at screening, within 48 h of hospital admission: ensure patient meets inclusion and exclusion criteria; record 24-hour total morphine equivalents; record habitual marijuana usage; obtain informed consent via patient or proxy.

Once patients are confirmed to meet all criteria and have signed an informed consent, they will be randomized 1:1 across the two study arms (dronabinol or control, Figure 1). The following procedures will be performed during randomization: randomize the patient using the Microsoft Excel blinded randomization schema; record pain using the patient self-reported pain numeric rating scale (NRS, 0–10 scale).

The following procedures will be performed during the acute hospitalization, post-randomization: if randomized to the dronabinol arm, administer the first dose within the first scheduled dose window and within 12 h of randomization. Record all doses received, including date, time and dose; record all opioid and non-opioid systemic analgesics received, including route/dose/frequency; record all non-analgesic concomitant medications; record pain NRS scores at the following time points: once admitted in hospital bed, preoperatively in the OR prior to anesthesia, one hour post operatively; record all analgesic complications; record all documented drug use from patient self-report and urine drug screening results. Detailed information on the regularity of marijuana use will also be recorded.

The following procedures will be performed at hospital discharge: record discharge pain NRS score; record discharge location; re-consent, if necessary.

#### *2.4. Randomization and Blinding*

Patients will be assigned to treatment by a randomization schedule developed and maintained by an independent statistician. The randomization allocation sequence was computer generated and is blinded, with allocation hidden until a patient has met all inclusion and exclusion criteria and provided informed consent. Randomization will occur in a 1:1 fashion in blocks of 2 and 4 and stratified by habitual marijuana user (yes/no).

The assessor, participants, treatment team, and statisticians are unblinded. All assessments are standard and routinely collected by the assessors (ICU and general ward nursing staff), including pain NRS scores and analgesia administration.

#### *2.5. Intervention*

The study drug is dronabinol (Marinol®, AbbVie, Inc; Chicago, IL, USA). Eligible patients will receive adjunctive dronabinol vs. control (no dronabinol, systemic analgesia only). Patients will be allocated to a treatment in accordance with the randomization schedule following confirmation of eligibility.

Patients who consent to participate in the study will have an order in the electronic medical record that will be used to assist with treatment compliance and for dispensing dronabinol, when applicable. Patients randomized to the dronabinol arm should receive their first dose within 12 h of randomization. The initial dosing and any changes in dosing will be determined by the prescribing/treating clinician. The target dose is 5 mg twice daily; the dose may be adjusted to within 5 mg to 30 mg daily in divided doses (e.g., 2.5 mg twice daily–10 mg three times daily). Patients who are randomized to the control arm will have an order set that specifies no administration of dronabinol for 48 h.

Patients in both arms will receive as needed (pro re nata, PRN) non-opioid/opioid analgesia as determined by the care team; patients who are not randomized to the dronabinol arm will receive these analgesics only, while patients randomized to the dronabinol arm will receive dronabinol in addition to PRN non-opioid/opioid analgesia. A target pain numeric rating score for trauma patients is 4 or less on a 0–10 scale. Higher pain scores ≥5 typically warrant analgesia, as determined by the attending physician and care team for the patient's specific needs. These established guidelines will ensure patients are receiving analgesia based on self-reported pain, independent of treatment arm.

After the 48-hour treatment window post-randomization, the use of adjunctive dronabinol for the remaining acute hospitalization will be at the patient's and physician's discretion. Except for the analgesia protocol, all other interventions will follow techniques used in the context of everyday clinical practice, and thus will be identical for participants in both arms. The following medications are discouraged: neuraxial and locoregional nerve blocks.

#### *2.6. Outcome Measures*

Patients will be followed to hospital discharge for outcomes of morphine equivalent use, length of stay, pain NRS scores, hospital complications, and analgesic complications.

The primary outcome is morphine equivalents. All opioids consumed will be converted to morphine equivalents, as shown in Figure 1 [11]. The clinical effects of treatment arm on morphine equivalents will be evaluated at 48 h after randomization.

Secondary outcomes include the following:


Analgesic complications will be recorded irrespective of the presence or absence of a causal relationship, and include the including:


#### *2.7. Statistics*

Significance is set at an alpha value of 0.05. SAS (Cary, NC) software will be used for statistical analysis. All efficacy analyses will be performed in the intent-to-treat population, defined as all patients who are randomized. Subset analyses will be performed by habitual marijuana use.

The primary endpoint is the change in morphine equivalents and will be assessed as: post-randomization (48 h after randomization) minus pre-randomization (24 h prior to randomization). No imputation will occur for the primary endpoint. The change in morphine equivalents (mg) will be analyzed with an analysis of covariance (ANCOVA) model to examine the effect of treatment arm, adjusted for age, gender, injury severity score, and clinical characteristics that differ between groups with *p* < 0.15. Of note, our study inclusion criteria allow for patients to present with polytrauma. We anticipate the majority of patients will have injuries to the thorax and extremities, with few patients presenting with severe TBI because administration of opioids and other drugs that alter a neurological assessment tend to be used sparingly. Should there be differences in injury patterns, despite the 1:1 randomization procedure, these differences will be adjusted for in the primary ANCOVA analysis.

Secondary efficacy analyses include the difference between treatment groups in: hospital disposition, hospital complications, and analgesic complications, reported as proportion (%) and analyzed with chi-square tests; morphine equivalents over the hospitalization, hospital length of stay (days), time (h) to transition to non-opioid analgesia, pain NRS scores at all specified time points, reported as median (IQR) and analyzed with a Wilcoxon rank-sum test. Analgesic complications will be described by severity as mild, moderate and severe.

#### *2.8. Sample Size*

The planned enrollment is 122 patients total randomized 1:1 across two study arms: dronabinol or no dronabinol (systemic analgesia only). The sample size is based on a 38% reduction in morphine equivalents with adjunctive dronabinol vs. an 8% reduction in morphine equivalents for systemic analgesics only, with a pooled standard deviation of 58. The analysis was performed using two sample mean tests with normal approximation and equal weights. These estimates were derived via bootstrapping of the final matched study sample of 66 patients. The power to demonstrate the main effect of dronabinol over systemic analgesics is 80% using a 2-tailed alpha of 0.05.

#### *2.9. Ethical Considerations*

The study was approved from the Institutional Review Board for St. Anthony Hospital (Catholic Health Initiatives). There may be patients incurring cognitive impairment (due to head injury or acuity of illness). The study coordinator will discuss with the treating team and will directly assess the consenting capacity of the patient. The study nurse follows the current hospital protocol regarding the use of consent by a legally authorized representative. In these clinical situations where the patient's representative initially consents, the patient will be "re-consented" when able to assure that they want to continue in the study.

Safety outcomes will be reported to the head of the medical executive committee at an ongoing basis. If/when the rate and/or severity of the monitored safety events becomes unacceptable, the medical executive committee has procedures in place to protect research subjects.

An interim analysis will be performed when >50% (*n* = 62) of patients have been enrolled and discharged from the hospital to determine clinical equipoise. A stopping guideline of *p* < 0.001 will be used for the primary end point.

#### **3. Discussion**

This is the first randomized trial to investigate whether the addition of dronabinol is effective at reducing opioid consumption for acute pain management of traumatic injury. There are numerous strengths of this study. This clinical trial improves upon our previously published matched cohort study and removes many of the limitations of that study: patients will now be matched by self-reported marijuana use; the pre-treatment period for the controls will be identical to cases rather than being estimated based on the median time from admission to first administration of dronabinol among cases; we will know why controls were not prescribed dronabinol; there are complications and adverse effects that are associated with both systemic analgesics and dronabinol, which will be recorded and analyzed in this trial by treatment arm and by severity. Additional strengths of this study are that it is investigator-initiated and independent from pharmaceutical or other industry interests, and the findings (whether positive or negative) will be submitted to a peer reviewed scientific journal for publication.

Another benefit of this study is the stratified randomization by chronic marijuana usage. Earlier work by the study investigators suggests that pre-injury marijuana use results in increased consumption of opioid analgesics and greater self-reported pain following traumatic injury compared to trauma patients who are marijuana naïve [12]. If the randomized trial demonstrates a greater treatment effect in the subset of chronic marijuana users, this will have wide-ranging clinical implications for acute pain management, because trauma patients have a high prevalence of marijuana use and other substance abuse issues, reported in 40–50% of patients [13,14] that appears to be increasing over time [15]. Thus, if marijuana use significantly affects acute pain management then chronic marijuana users will merit special consideration during acute pain management.

While there are now 11 states that have legalized recreational marijuana, we believe Colorado is uniquely able to study this issue because of the high utilization in our state. Colorado was the first state to legalize and commercialize recreational marijuana, with retail shops opening on January 1st, 2014. A recent study identified that commercialization of recreational marijuana in Colorado was associated with an increased use of marijuana or an increased risk of traumatic injury while using marijuana [15].

Opioids are established and effective analgesics for managing pain in the traumatic and critical care setting due to their proven efficacy in treating moderate to severe acute pain [16]. The Center for Disease Control and Prevention (CDC) estimates that approximately 130 Americans are dying each day from opioid overdose, resulting in an opioid epidemic. We believe the use of dronabinol as a tool in the clinician's tool kit to decrease reliance on opioids is an appealing option. Some possible benefits of this study include better pain control and a lower need for opiates for participants. Use of dronabinol to reduce or maintain the opioid regimen, rather than increasing narcotic dosages to detrimentally high levels, may also reduce the negative effects of opioids on vascular neurologic response and respiratory depression.

One of the primary limitations of this trial is that the study is open label. Patients are still prospectively randomized to active treatment vs. control and all assessments are standard and routinely collected by the assessors (ICU and general ward nursing staff), including pain scores and analgesia administration. However, we are unable to blind patients because there are no orally administered placebo pills that are on hospital formulary to be used for this study (unavoidable blinding). We did not blind clinicians because the dosing of dronabinol may need to be modified and is allowed within the range of 5 mg to 30 mg daily in divided doses. Although a blinded study would be preferred to reduce knowledge bias, the study design is compatible with real-world situations and increases the external validity of the study.

Additional limitations are as follows. First, our preliminary study was conducted in 2017, around the peak of the opioid epidemic [17,18]. Since that time, there have been enterprise-wide initiatives to use alternatives to opioids [19,20], which could impact our enrollment criteria. However, our study has potentially greater implications in the current setting where opioid alternatives are sought. Second, and related, the data used to power the RCT were recorded in 2017, and it is possible that opioid consumption will be less in both groups (dronabinol and control), but whether this translates to a different treatment effect with dronabinol remains to be seen. Third, marijuana use is based on self-reporting because admission urine toxicology testing is only utilized in about 50% of patients, with a bias towards screening younger patients. Unlike blood alcohol tests, urine toxicology testing

seldom results in a change in care and thus are not routinely ordered following traumatic injury. We will not be requiring a change in practice for ordering urine toxicology testing as part of our study. However, our unpublished research demonstrates the percent agreement between urine toxicology findings and patient self-report is 81% for cannabis. The negative predictive value of 95% demonstrates that a negative self-report correctly identifies 95% of patients who test negative for cannabis, while the specificity provides an 85% chance that a patient will not test positive for cannabis if the patient denies use. Fourth, the results of this study are only be applicable to dronabinol and not to other cannabinoids, such as the recently trending cannabidiol (CBD). Finally, the study is currently approved as a single-center RCT, which limits its generalizability. The authors are amenable to adding additional sites which use dronabinol on formulary without restrictions.

There are two additional risks to the patient that need to be mentioned. First, this study involves an experimental (investigational) drug that has not been approved by the U.S. Food and Drug Administration (FDA) for the specific indication of acute pain management. Dronabinol is only FDA approved for loss of appetite due to HIV and chemotherapy-induced nausea and vomiting. This study is not intended to result in an FDA Investigational New Drug Application. Second, dronabinol is a synthetic version of THC. There is a risk that the study medication will result in a positive urine drug screen test for cannabis for two weeks or more in patients who are not a current user of marijuana products. In most cases, if an employee has a recent prescription for dronabinol, that is sufficient to report the result to the employer as a negative.

#### *Trial Status*

The trial has been recruiting patients since October 2019 and will continue until 122 patients have been randomized. Protocol version 1.2. Two amendments have occurred since trial commencement. First, the inclusion criteria of a minimum baseline pain score ≥5 was removed. The second amendment modified the sample size calculation to incorporate the full preliminary study findings, rather than a smaller pilot population.

**Author Contributions:** Conceptualization, C.S.; methodology, K.S. and C.S.; software, K.S.; formal analysis, K.S.; investigation, C.S. and R.M.; resources, D.B.-O.; writing—original draft preparation, K.S.; writing—review and editing, C.S., R.M., D.B.-O.; supervision, D.B.-O.; project administration, C.S. and D.B.-O. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Brain Sciences* Editorial Office E-mail: brainsci @mdpi.com www.mdpi.com/journal/brainsci

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18