The study was initially approved by the Colorado Multiple Institutional Review Board (COMIRB Panel D IRB number: IRB00002760) in August 2018 and registered as an observational study on www.clinicaltrials.gov (NCT03617692). Participants were recruited with advertisements placed in local oncology clinics, mailed advertisements, and posts on social media. Participants were compensated with grocery store gift cards for completion of the baseline ($60) and acute administration ($100) appointments.

Trained research staff screened participants via telephone for the following inclusion criteria: (1) stated willingness to comply with all study procedures and be available for the duration of the study; (2) aged 21 years or older; (3) have a diagnosis of any solid tumor type and has undergone or is undergoing either curative or palliative treatment; (4) have the intent to use cannabis to treat their symptoms; (5) no use of other non-prescription drugs of abuse (e.g., cocaine, heroin, or methamphetamine) in the past 60 days; (6) not actively seeking or in treatment for any substance use disorder; (7) no acute illness other than cancer that could affect cognition or compliance per the decision of the study physician; (8) not pregnant or trying to become pregnant; and (9) score better than 20 (above the moderately to the severely impaired range) on the Telephone Interview for Cognitive Status [50]. Participants did not need to be naive to cannabis to participate in the study. Participants were subject to polypharmacy while participating in the study and were not given instructions about prescription medication use from research staff.

Changes to the eligibility criteria were made after the study’s commencement to enhance recruitment. Specifically, the allowed cancer diagnoses were expanded from including only lung cancer patients to including any solid tumor type. As a result, only edible cannabis products were allowed in the present study because the original design did not allow the use of inhaled cannabis products in the lung cancer patient population.

Due to the spread of coronavirus disease 2019 (COVID-19), the study transitioned from an in-person to a remote modality to reduce the risk of exposure. Remote participants (n = 17) interacted with research staff via Zoom or over the phone during study appointments and completed all the same procedures and measures as in-person participants (n = 11) with the exception that remote participants did not have blood collected. All study data were collected and managed using Research Electronic Data Capture (REDCap) electronic capture tools hosted at the Colorado Clinical and Translational Sciences Institute [51].

In-person participants provided written informed consent at the University of Colorado Boulder. Remote participants provided electronic informed consent via REDCap. Female participants of child-bearing age completed a urine pregnancy test to verify eligibility (research staff delivered a urine receptacle to remote participants). All participants completed self-report questionnaires and a measure of objective cognitive function. Participants were then provided with basic information about edible cannabis products, including examples of commonly found products, their various cannabinoid profiles, safety guidelines for use, approximate prices, and locations where these products can be purchased. Research staff also conducted a medical record review for demographic information and cancer diagnosis.

Following the baseline appointment, participants purchased the edible cannabis product of their choice from the local medical or recreational dispensary of their choice and used the product ad libitum for two weeks without any instruction from research staff about dosing or frequency. Participants were asked to not use any other cannabis products during this two-week period. Participants submitted photos of their cannabis product and its state-required label that details the cannabinoid content of the product via an online REDCap survey. A member of the research staff not responsible for data collection managed the product photo surveys to keep the research staff responsible for data collection blind to the products selected by participants.

Participants were instructed to not use any cannabis on the day of the acute administration appointment before it was scheduled to begin. The acute administration appointment included three rounds of assessments: (1) pre-use; (2) one-hour post-cannabis use; and (3) two-hour post-cannabis use, as previous research supports circulating cannabinoids peaking approximately two hours following oral cannabis ingestion [52–54]. In-person participants ingested their edible cannabis products in their homes but completed all assessments in a federally compliant, university-approved mobile pharmacology laboratory, as described in a previous publication [55]. Remote participants completed the entirety of the acute administration appointment in their homes, communicating with the research staff via Zoom or over the phone during assessments.

The pre-use assessment included self-report questionnaires and a measure of objective cognitive function. In-person participants also completed a blood draw for circulating cannabinoids (i.e., THC and CBD). Participants then ingested their desired amount of cannabis, recorded the amount of product ingested in a blinded fashion (e.g., “one gummy”, “one mL”), and waited one hour before resuming assessments. A member of the research staff not responsible for data collection translated the amount of product ingested into milligrams of THC and CBD consumed using the product photo surveys.

The one-hour and two-hour post-cannabis use assessments involved self-report questionnaires, a measure of objective cognitive function, and blood draws (for in-person participants only).

The progress of participants through the study is presented in Figure 1. Twenty-eight participants were recruited for the study. Twenty-five participants completed both the baseline and the acute use appointments and were included in the analyses. Descriptive information about study participants is provided in Table 1, and Table 2 provides information about participant cancer status and current treatment. Seventeen (68%) of participants reported no cannabis use in the seven days prior to the baseline appointment, and eight (32%) of participants reported no cannabis use in the 30 days prior to the baseline appointment. Participants reported no adverse events related to cannabis use during the two-week ad libitum use period, and no adverse events were observed following acute cannabis administration.

On average, participants used an edible cannabis product for 8.1 days (SD = 5.1, range = 0–14) and ingested a total of 116.4 mg (SD = 178.0, range = 0–625) of THC and 162.7 mg (SD = 420.3, range = 0–2,100) of CBD, or 8.3 mg of THC and 11.6 mg of CBD per day, during the two-week ad libitum use period. Participants reported using eighteen different brands of cannabis products. Fourteen participants used an edible cannabis product in the form of candies (e.g., gummies, chocolate), seven used tinctures, two used pill forms, one used beverages, and one used baked goods (stroopwafels). Milligrams of THC and CBD ingested at the acute administration appointment are presented in Figure 2. All participants confirmed using their purchased edible cannabis product at the acute administration appointment. Mean circulating THC at the acute administration appointment was 5.58 ng/mL (SD = 15.48, range = 0.39–46.87) at pre-use (n = 9), 10.43 ng/mL (SD = 26.72, range = 0.39–81.50) at one-hour post-use (n = 9), and 7.59 ng/mL (SD = 18.96, range = 0.39–54.50) at two-hour post-use (n = 8) for in-person participants with available blood data (participant agreed to blood collection and blood was successfully collected). Mean circulating CBD at the acute administration appointment was 3.57 ng/mL (SD = 3.42, range = 0.39–10.78) at pre-use (n = 9), 10.81 ng/mL (SD = 15.02, range = 0.47–43.70) at one-hour post-use (n = 9), and 9.67 ng/mL (SD = 8.40, range = 2.28–24.96) at two-hour post-use (n = 8) for in-person participants with available blood data. Circulating cannabinoid levels are also presented in Figure S1. Significant positive correlations were observed between THC ingested and circulating THC (r = 0.98, P < 0.001) and CBD ingested and circulating CBD (r = 0.85, P = 0.003) at one-hour post-use during the acute administration appointment for in-person participants with blood data (n = 9). The positive correlation between THC ingested and circulating THC was reduced (r = 0.65, P = 0.079) when one outlier (60.0 mg of THC ingested) was removed.

There was no significant effect of time on the Stroop effect on reaction time (ms, baseline mean = 84.93; pre-use at acute administration mean = 68.61; B = −23.46, SE = 14.54, P = 0.12), and no significant interactions were observed based on ingested cannabinoids for the Stroop effect on reaction time (Ps > 0.17). A significant effect of time was observed for congruent reaction time (ms, baseline mean = 1,091.99; pre-use at acute administration mean = 1,013.94; B = −71.72, SE = 29.57, P = 0.02) and incongruent reaction time (ms, baseline mean = 1,176.93; pre-use at acute administration mean =1,082.55; B = −93.98, SE = 25.43, P = 0.001), such that both reaction times decreased from baseline to the acute administration appointment. No significant interactions with ingested levels of cannabinoids were observed for Stroop congruent reaction time (Ps > 0.48) or Stroop incongruent reaction time (Ps > 0.27).

There was no significant effect of time on the Stroop effect on error rate (baseline mean = 0.04; pre-use at acute administration mean = 0.03; B = −0.01, SE = 0.01, P = 0.53), the error rate of congruent trials (baseline mean = 0.06; pre-use at acute administration mean =0.04; B = −0.02, SE = 0.02, P = 0.43), or the error rate of incongruent trials (baseline mean = 0.10; pre-use at acute administration mean = 0.07; B = −0.02, SE = 0.02, P = 0.22). No significant interactions with amount of cannabinoids ingested were observed for Stroop effect on error rate (Ps > 0.09), congruent error rate (Ps > 0.66), or incongruent error rate (Ps > 0.34).

Descriptive statistics for changes in acute administration outcomes are presented in Table 3.

The Stroop effect on reaction time exhibited a significant linear effect of time (B = 12.93, SE = 6.09, P = 0.04), such that there was a significant increase in the Stroop effect over the course of the acute administration appointment (i.e., performance worsened). None of the time X cannabinoid interaction terms were significant (Ps > 0.26). Specifically, while reaction time on congruent trials decreased over time (B = −25.75, SE = 8.44, P = 0.004), reaction time on incongruent trials did not significantly change (B = −12.82, SE = 8.04, P = 0.12).

The Stroop effect on error did not change over the course of the acute administration appointment (B = 0.004, SE = 0.007, P = 0.59), and none of the time X cannabinoid interaction terms were significant (Ps > 0.36). Specifically, while the error rate on incongruent trials increased over time (B = 0.01, SE = 0.007, P = 0.05), the error rate on congruent trials did not significantly change (B = 0.01, SE = 0.007, P = 0.14).

This study is among the first to explore associations of acute and sustained legal market cannabis edible use, including the use of both THC and CBD, with changes in pain, cognition, and quality of life in a group of cancer patients. Patients reported improvements in pain, sleep quality, and subjective cognitive function following a two-week ad libitum cannabis use period, as well as reductions in pain intensity, increases in feelings of subjective high, and some decreases in cognitive function following acute cannabis use. The observational nature of the study also allowed a window into the kinds of products cancer patients select, how frequently they choose to use their products, and what doses of THC and CBD they choose to take. Participants in the present study used a variety of edible forms of cannabis and a wide range of doses of both THC and CBD. This is perhaps somewhat unexpected given the rapidly growing interest in CBD over the past five years [71]. A significant effect of CBD use on baseline pain intensity was observed, such that greater CBD use over the two-week period was associated with greater pain intensity at baseline, suggesting that individuals with higher pain intensity at baseline may have intentionally selected products higher in CBD or consumed more CBD over the two weeks to cope with these heightened pain levels. A greater understanding of how cancer patients come to make decisions around cannabis use and how their expectancies may vary by cannabinoid content will be critical in order to develop guidance for both clinicians and patients who wish to use cannabis for palliative cancer care.

Improvements in pain intensity and pain interference were observed following sustained cannabis use, as well as improvements in pain intensity following acute use. Interestingly, a time X CBD ingested interaction emerged in the two-week models of pain intensity, with greater CBD use during the two-week ad libitum period being associated with steeper improvements in pain intensity. Notably, no moderating effect of CBD use was observed during the acute use appointment. Two early clinical trials in cancer patients that utilized THC oil capsules demonstrated a trend toward progressive pain relief with increasing doses of THC over a six-hour period [72], as well as significant improvements in pain following 20 mg of THC compared to placebo over a seven-hour period [73], although adverse reactions to this dose were observed in this group of predominantly inexperienced cannabis users [73]. In addition, a two-week trial of oromucosal sprays showed improvements in pain compared to placebo following the use of a combination of CBD and THC spray, but not with a THC-only spray [74]. More recent trials have yielded mixed results on the effectiveness of cannabinoids for pain [75, 76], but the picture that emerges from across the literature suggests that THC and CBD may influence pain relief over different time courses. Pharmacokinetic research will be key to addressing these potentially differential effects of THC and CBD depending on the time course of sustained versus acute use.

Improvements in sleep quality following sustained cannabis use were observed, with higher CBD use during the two-week ad libitum use period being associated with steeper improvements in sleep quality. A recent meta-analysis that included five randomized controlled trials of oromucosal sprays containing cannabinoids found a very small improvement in sleep disturbance compared to placebo in individuals living with chronic cancer pain [77]. Objective measures of sleep in the context of carefully controlled sleep studies would be ideal for better understanding the differential effects of THC and CBD across the full range of measures of sleep quality.

Two weeks of ad libitum use in the present study was associated with patient reports of improvements in subjective cognitive function. Some objective measures of cognitive function also improved, including reaction times for both congruent and incongruent trials on the Stroop task. Consistent with the literature on pain and cognitive function, an exploratory analysis supported a negative association, such that increases in pain were associated with subjectively worse cognitive function. Given widespread concern—by both clinicians and patients themselves—about decrements in cognitive function associated with cancer treatment, a potential indirect role of cannabis use in improving subjective cognitive function is an intriguing finding to pursue.

Also consistent with past research, some impairments following acute administration on objective cognitive performance were also observed, such that the Stroop effect on reaction time and error rate for incongruent trials both increased (i.e., got worse).

Limitations of the present study include a small sample size which limited statistical power to investigate demographic or other moderators in the analyses. From an internal validity perspective, the lack of random assignment to product condition and the lack of a placebo-control condition are clearly problematic for any causal conclusions regarding the influence of cannabis on any of the outcomes tested in this study. In addition, the study was interrupted by the COVID-19 pandemic, requiring the study to transition to a remote modality to continue data collection. The generalizability of the conclusions of the present study is limited by the largely white sample and by the restriction on the use of cannabis edibles. There is evidence that cancer patients also use inhaled cannabis products [7, 78]. The short two-week timeframe may have limited the capacity of the present study to observe changes following sustained cannabis use. The inclusion of participants with different cannabis use histories and undergoing different treatments may have influenced participant response to acute and sustained cannabis use.

Despite its limitations, this study benefits from increased external validity via its naturalistic design, which allowed participants to select and use legal market edible cannabis products in a manner and environment that reflected their own preferences, needs, and typical use behaviors. The study’s inclusion of multiple cancer types also allowed for a greater age range of participants and an equal representation of sex, which may not be feasible when restricting eligibility to certain cancer types (e.g., breast, prostate).

In this study of legal market edible cannabis products in cancer patients, two weeks of ad libitum cannabis use was associated with improvements in pain intensity and interference, sleep quality and subjective cognitive functioning. It is particularly of note that high CBD, not THC, use during this two-week period was associated with steeper improvements in pain intensity and sleep quality. Acute cannabis use decreased pain intensity, decreased some measures of objective cognitive function, and increased feelings of subjective high. The findings suggest that studies conducted over longer timeframes with larger samples that allow the capacity to test for moderating effects of product type (e.g., inhaled versus edible), cancer type/stage, and demographic characteristics are an important next step in understanding the potential benefits and harms of cannabis use for palliative care in cancer patients.