Clinical, physiological, imaging and molecular responses to cannabis smoking: the Canadian Users of Cannabis Smoke (CANUCK) study

Graphical abstract

Cannabis smoking is associated with worse respiratory symptoms, lower pre-bronchodilator forced expiratory volume in 1 s (FEV₁) to forced vital capacity (FVC) ratio and forced expiratory flow at 25–75% of FVC (FEF_25–75%), more radiographic abnormalities, and altered airway epithelial immune response and mucin (MUC5AC) expression. CAT: COPD Assessment Test; CT: computed tomography; SGRQ: St George's Respiratory Questionnaire; ¹²⁹XeMRI: ¹²⁹Xenon magnetic resonance imaging.

Abstract

Background

The growing popularity of cannabis smoking in an era of legalisation has prompted concerns about respiratory health.

Objective

To investigate clinical and airway epithelial transcriptomic features associated with cannabis smoking.

Methods

This cross-sectional study analysed data from 139 cannabis-smoking participants categorised by joint-year exposure (low: ≤5; moderate: >5–20; high: >20 joint-years) and 57 never-smokers. We evaluated respiratory symptom questionnaire scores, lung function measurements, chest computed tomography and hyperpolarised ¹²⁹Xenon pulmonary magnetic resonance imaging measurements across groups. We compared the expression of immune response signatures and mucin genes in airway epithelial brushings collected from bronchoscopy. Using air–liquid interface cell cultures, we quantified epithelial mucin 5AC (MUC5AC) protein and correlated its expression with clinical outcomes.

Results

Among cannabis-smoking participants (48% male, median age of 27 years), 84% reported current or former cigarette smoking or vaping. Cannabis-smoking groups reported worse respiratory symptoms than never-smokers. High joint-year cannabis-smoking participants showed lower pre-bronchodilator forced expiratory volume in 1 s to forced vital capacity ratio, lower forced expiratory flow at 25–75% of the forced vital capacity, more radiographic emphysema and more ventilation abnormalities than never-smokers. Airway epithelial brushings from cannabis-smoking participants demonstrated an increased type 2 immune response, decreased type 17 immune response and higher MUC5AC gene expression than non-cannabis-smoking participants. Epithelial MUC5AC protein expression in cell cultures correlated with worse clinical outcomes and imaging abnormalities.

Conclusions

Cannabis smoking, particularly at high exposures, is associated with worse respiratory symptoms, lower lung function, functional imaging abnormalities and dysregulated immune responses in the airway epithelium. These observations suggest respiratory harm associated with cannabis smoking and underscore the concerns for future respiratory morbidities related to persistent cannabis use.

Shareable abstract

This study shows objective evidence of the adverse physiological, imaging and molecular changes in cannabis smoke-exposed small airways that may contribute to long-term respiratory morbidity https://bit.ly/4gvZqGf

Introduction

There are approximately 228 million users of cannabis globally, a prevalence that has been steadily rising over the last decade [1]. Growing interest in the therapeutic applications of cannabis has prompted some countries to ease restrictions around its use. On 17 October 2018, Canada became the second country in the world to fully legalise cannabis for recreational use [2]. Since then, cannabis use has increased amongst Canadians, with over one third of young adults now reporting cannabis use each year [3]. Smoking remains the most popular method to consume cannabis, which has raised concerns over long-term respiratory health implications. Although cannabinoids can reduce inflammation in many models of chronic disease [4], smoking cannabis releases combustion toxicants similar to cigarette smoke [5], potentially resulting in deleterious structural remodelling and immune modulation of the airways [6]. Cannabis smoking has been linked to increased respiratory symptoms, though its impact on lung function and the development of chronic lung diseases remain inconclusive [7]. As a result, public perceptions of the harms of cannabis smoking have generally been more favourable when compared with cigarette smoking.

Reliance on spirometry in previous cohorts to detect lung injury associated with cannabis smoking could overlook damage in the small airways, from where diseases such as chronic obstructive pulmonary disease (COPD) are thought to originate [8]. This may be particularly true in young adults (who report the highest use of cannabis amongst all age groups) without apparent airflow obstruction. Detection of early small airway damage from cannabis smoking may require the use of novel imaging and molecular techniques not yet available in a clinical setting. Furthermore, adequate assessment of pulmonary risks needs to reflect real-world habits of cannabis smoking, which are often accompanied by concurrent cigarette smoking and vaping. Finally, previous attempts at clarifying the potential harms associated with cannabis smoking have been hampered by the shadow of the justice system, where honest disclosure of cannabis habits may have been impeded by the risk of criminal retribution. As one of three nations in the world with full legalisation, Canada is in an excellent position to assess the respiratory impact of cannabis smoking without these legal barriers. The Canadian Users of Cannabis Smoke (CANUCK) study was designed to understand small airway dysfunction at a clinical and molecular level through detailed phenotyping and respiratory health assessments paired with leading molecular technologies, advanced imaging techniques and bronchoscopic sampling of airway epithelial cells.

Methods

Participants

Adults ≥19 years old were recruited through community-based strategies between 2021 and 2024 in Vancouver, Canada. We included a healthy never-smoker (NS) and a current cannabis-smoking group based on self-reported smoking habits. The NS group reported no history of cannabis smoking, cigarette smoking or vaping. We excluded participants with a pre-existing history of physician-diagnosed chronic lung disease, such as COPD, lung cancer, bronchiectasis or asthma (supplementary figure E1 and table E1). Cumulative cannabis exposure was quantified by joint-years, which was the number of joints smoked per day multiplied by the number of years of smoking. For participants who reported their use in grams, a conversion was made based on the average size of the top 50 most popular joints sold at provincial cannabis stores across British Columbia, Canada (0.5 g per joint). We classified cannabis-smoking participants by joint-years into low (≤5 joint-years), moderate (>5–20 joint-years) and high (>20 joint-years) exposure groups [9, 10]. To reflect real-world habits, participants with prior and/or concurrent cigarette smoking and/or vaping of nicotine, tetrahydrocannabinol or cannabidiol products were included in the cannabis-smoking group (supplementary table E2). This study received ethics approval from The University of British Columbia and Providence Health Care Research Ethics Board (H19-02222, H19-01163 and H21-01237). All participants provided written informed consent.

Clinical characterisation

In this cross-sectional analysis of the CANUCK study (supplementary figure E1), participants underwent baseline characterisation using clinical questionnaires, blood sampling and pulmonary function testing. A subset of participants underwent thoracic computed tomography (CT) imaging and ¹²⁹Xenon magnetic resonance imaging (¹²⁹XeMRI). A subset of participants in the imaging cohort consented to a research bronchoscopy for the collection of airway epithelial brushings. Participants undergoing bronchoscopy were stratified into a current cannabis-smoking group and non-cannabis-smoking group. Additional participants who underwent bronchoscopy from a previously described cohort were included to increase sample sizes in both groups (supplementary appendix) [11].

Respiratory symptoms and quality of life were reported using the St George's Respiratory Questionnaire (SGRQ) and COPD Assessment Test (CAT). Pulmonary function tests were completed according to European Respiratory Society/American Thoracic Society standards, including spirometry before and after 400 μg of salbutamol, body plethysmography for lung volumes and diffusing capacity of the lung for carbon monoxide. On thoracic CT imaging, qualitative measures of emphysema, bronchiectasis, mosaic attenuation, ground glass abnormalities and reticular changes were assessed by a clinical radiologist. Gas trapping and emphysema were quantified by the percentage of low attenuation areas (LAA), i.e. the percentage of voxels with density <−856 Hounsfield units (HU) (LAA856) at residual volume and percentage of voxels with density <−950 HU (LAA950) at total lung capacity [12]. Disease probability measures (DPMs) quantified emphysema and functional small airway disease (fSAD) according to established protocols [13]. Pi10 represents the square root of the wall area of a hypothetical airway with a luminal perimeter of 10 mm [12]. ¹²⁹XeMRI measured the areas with ventilation defects and low ventilation, expressed as percentages [14]. Further details are included in the supplementary appendix.

Bulk RNA-sequencing from airway epithelial brushings

Research bronchoscopies were performed according to published protocols [15]. Airway epithelial cells were collected from airways ≤2 mm in diameter from the right or left upper lobe using a steel-tipped cytology brush. Bulk RNA-sequencing and differential gene expression analyses were performed on brushings using our standardised bioinformatics pipeline [15]. First, differentially expressed genes (DEGs) in cannabis smoking were identified, then the enriched and depleted biological pathways associated with cannabis smoking were explored using gene set enrichment analysis (GSEA). GSEA was performed on all genes ranked by effect size (−log₁₀(p-value)×log2 fold change (log₂FC)) using WebGestaltR (www.webgestalt.org) to identify Kyoto Encyclopedia of Genes and Genomes (KEGG) and Reactome pathways. Next, we identified cannabis smoking-associated DEGs that were significantly correlated with joint-years. Finally, we investigated the expression of immune response in airway epithelial cells by gene signatures that were determined a priori (supplementary table E3). Gene signature scores were calculated as the mean gene counts of log-transformed counts per million (logCPM). We investigated the expression of type 1, 2 and 17 immune responses and mucin genes (MUC5AC and MUC5B). Further details on bulk RNA-sequencing and analyses are included in the supplementary appendix.

Histology in air–liquid interface cell cultures

In a subset of participants with viable epithelial brushings collected from bronchoscopies, airway cells were expanded in Pneumacult Ex-Plus media (STEMCELL Technologies) and maintained as air–liquid interface (ALI) cultures. After 28 days, formalin-fixed ALI cultures were processed with immunohistochemical staining for MUC5AC. Protein expression of MUC5AC was calculated as the percentage of the total area. Further details are included in the supplementary appendix.

Statistical analyses

Clinical characteristics between NS and cannabis-smoking participants were compared using independent t-tests for normally distributed variables and Mann–Whitney U tests for non-normally distributed variables. Linear trends across the ordered subgroups (NS, low, moderate and high joint-year groups) were evaluated using one-tailed Jonckheere–Terpstra tests and linear regressions adjusting for age, sex, body mass index (BMI), vaping and cigarette smoking status. Owing to the high prevalence of cigarette smoking and vaping amongst cannabis-smoking participants, clinical outcomes were analysed in relation to inhaled exposure subgroups (cannabis only, cannabis+cigarette, cannabis+vaping, cannabis+vaping+cigarette) to understand subgroup differences.

The airway epithelial transcriptome was compared between cannabis-smoking and non-cannabis-smoking participants. Significant DEGs were identified with a model adjusting for age, sex, BMI and vaping status using a false discovery rate (FDR) <0.05 (Benjamini–Hochberg adjusted p-value). Model covariate selection was based on principal component analysis (supplementary appendix). Significantly enriched and depleted biological pathways were identified with FDR <0.05. We reported the association between the expression of various DEGs and joint-year exposure using Spearman's correlation ranked by FDR. The percentage of positively stained MUC5AC protein in ALI cultures was were compared between cannabis-smoking and non-cannabis-smoking participants. We reported the relationships between MUC5AC protein expression and clinical outcomes of the ALI donors using Spearman's correlation. Statistical analyses were performed using Stata (version 18.5; StataCorp LLC), GraphPad Prism (version 10.4.1; GraphPad Software) and R (version 4.3.1; www.r-project.org). Statistical significance was defined as a two-sided p-value <0.05.

Results

Concurrent cigarette smoking and vaping is common in cannabis-smoking participants

Among 196 participants, 57 (29%) were in the NS group and 139 (71%) were in the cannabis-smoking group (table 1). The cannabis-smoking group was younger and consisted of more male participants (median age 27 years, 48.2% male) than the NS group (median age 31 years, 29.8% male). Among cannabis-smoking participants, the median cannabis exposure was 7 joint-years (interquartile range (IQR) 1–25 joint-years) (figure 1a). Cannabis-smoking participants were grouped into joint-year subgroups, with 61 (44%) reporting low, 39 (28%) reporting moderate, and 39 (28%) reporting high cannabis exposure (figure 1b, supplementary table E4). Among the 139 cannabis-smoking participants, 84% reported concurrent cigarette smoking and/or vaping (cannabis+vaping: n=47; cannabis+cigarette: n=27; cannabis+vaping+cigarette: n=43) (figure 1c, supplementary table E5).

TABLE 1.

Clinical characteristics of cannabis-smoking participants

	Never-smoking	Cannabis-smoking	p-value#
Total (n)	57	139
Demographics
Age (years)	31.0 (2.0–41.0)	27.0 (24.0–37.5)	0.09
Male sex	17 (29.8)	67 (48.2)	0.03
BMI (kg·m−2)	24.5 (22.6–27.2)	24.5 (21.7–28.8)	0.69
Race			0.11
White	27 (47.4)	71 (51.1)
Asian/Pacific Islander	16 (28.1)	15 (10.8)
South Asian	4 (7.0)	9 (6.5)
Latino/Hispanic	3 (5.3)	15 (10.8)
Black	1 (1.8)	2 (1.4)
First Nations	0 (0.0)	3 (2.2)
Other/Mixed Race	6 (10.5)	23 (16.6)
Prefer not to answer	0 (0.0)	1 (0.7)
Education			<0.001
College/Undergraduate	26 (45.6)	83 (59.7)
Graduate school	26 (45.6)	21 (15.1)
High school	2 (3.5)	19 (13.7)
Less than high school	0 (0.0)	2 (1.4)
Trade/technical/vocational training	3 (5.3)	14 (10.1)
Employment			0.90
Employed, full time	23(40.4)	56 (40.3)
Employed, part time	6 (10.5)	14 (10.1)
On disability	3 (5.3)	14 (10.1)
Student	14 (24.6)	34 (24.5)
Other/prefer not to answer/missing	4 (7.0)	10 (7.2)
Unemployed	7 (12.3)	11 (7.9)
Income ($)			0.05
<30 000	22 (38.6)	59 (42.5)
30 000–49 999	5 (8.8)	30 (21.6)
50 000–99 999	22 (38.6)	31 (22.3)
100 000–149 999	1 (1.8)	8 (5.8)
>150 000	2 (3.5)	2 (1.4)
Prefer not to answer	5 (8.8)	9 (6.5)
Marital status			0.09
Married/common law	16 (28.1)	38 (27.3)
Separated/divorced/widowed	1 (1.8)	7 (5.0)
Single	40 (70.2)	90 (64.8)
Prefer not to answer	0 (0.0)	4 (2.9)
Pulmonary function
Pre-BD FEV1% predicted	107.5±12.5	105.4±15.0	0.35
Pre-BD FVC % predicted	110.2±13.0	112.4±15.2	0.34
Pre-BD FEV1/FVC (%)	82.5 (78.0–86.9)	80.2 (74.6–85.0)	0.016
Pre-BD FEF25–75% (% predicted)	99.1±27.4	88.4±25.2	0.012
Post-BD FEV1 (% predicted)	111.9±12.0	110.6±14.9	0.58
Post-BD FVC (% predicted)	111.0±12.7	113.2±14.7	0.33
Post-BD FEV1/FVC (%)	85.2 (81.6–88.9)	84.0 (78.0–87.1)	0.058
Post-BD FEF25–75% (% predicted)	112.5±25.3	103.6±28.5	0.049
FEV1 BD change (% predicted)	3.9 (1.5–6.0)	5.4 (2.6–8.0)	0.029
FVC BD change (% predicted)	0.2 (−1.7–2.3)	0.9 (-1.3–2.6)	0.27
RV (% predicted)	109.9 (97.4–131.6)	113.9 (99.7–132.5)	0.68
TLC (% predicted)	103.9±11.4	106.1±12.4	0.27
RV/TLC	27.26 (22.94–30.54)	24.10 (21.38–29.99)	0.09
DLCO (% predicted)	104.2±15.3	106.1±19.5	0.52
Symptom scores
CAT¶	3.0 (1.0–7.0)	8.0 (4.0–12.0)	<0.001
SGRQ Total+	4.8 (1.1–9.7)	11.6 (5.3–19.2)	<0.001
SGRQ Symptom	8.5 (0.0–15.2)	24.2 (12.7–38.1)	<0.001
SGRQ Impact	0.0 (0.0–4.0)	4.3 (0.0–13.1)	<0.001
SGRQ Activity	6.2 (0.0–12.2)	12.2 (0.0–29.7)	<0.001
Imaging §
Qualitative CT
Emphysema	1 (2.3)	11 (10.5)	0.19
Bronchiectasis	0 (0.0)	2 (1.9)	0.89
Mosaic attenuation	3 (7.0)	8 (7.6)	1.00
Reticular abnormalities	1 (2.3)	6 (5.7)	0.64
Ground glass	2 (4.7)	9 (8.6)	0.63
Quantitative CT
LAA950%	1.0 (0.2–3.3)	1.1 (0.3–3.7)	0.62
LAA856%ƒ	1.5 (0.5–2.7)	3.3 (0.6–6.1)	0.08
DPMfSAD	8.0 (4.6–15.0)	9.4 (5.5–17.7)	0.42
DPMemphysema	0.2 (0.0–0.8)	0.7 (0.0–1.8)	0.11
Pi10	3.9 (3.9–4.0)	3.9 (3.9–4.0)	0.85
129XeMRI
Ventilation defect (%)	1.3 (1.0–2.2)	1.7 (0.9–2.9)	0.27
Low ventilation (% )	12.4 (8.3–16.3)	13.3 (9.5–17.0)	0.51
Ventilation deficit (%)	13.6 (9.4–18.3)	14.9 (10.6–20.2)	0.45

Data presented as mean±sd, median (interquartile range) or n (%), unless otherwise indicated. BMI: body mass index; BD: bronchodilator; FEV₁: forced expiratory volume in 1 s; FVC: forced vital capacity; FEF_25–75%: forced expiratory flow at 25–75% of FVC; RV: residual volume; TLC: total lung capacity; D_LCO: diffusing capacity of the lung for carbon monoxide; CAT: COPD Assessment Test; SGRQ: St George's Respiratory Questionnaire; CT: computed tomography; LAA950%: % of low attenuation area with density <−950 Hounsfield units; LAA856%: % of low attenuation area with density <−856 Hounsfield units; DPM: disease probability measure; fSAD: functional small airway disease; Pi10: square root of the wall area of a hypothetical airway with a luminal perimeter of 10 mm.

^#: calculated as Chi-square test/Fisher's exact test for categorical variables, two independent sample t-test for normally distributed continuous variables and Mann–Whitney U test for non-normally distributed continuous variables; ^¶: measures the global impact of symptoms (cough, sputum, dyspnoea, chest tightness) on health status, with higher scores (range 0–40) denoting a more severe impact on the participant's life; ⁺: SGRQ measures impact on overall health, daily life and perceived well-being, with higher scores (range 0–100) indicating more limitations; ^§: number of participants varied in each imaging subset (CT: never-smoking, n=44; cannabis-smoking, n=106; DPM: never-smoking, n=25, cannabis-smoking, n=63; ¹²⁹XeMRI: never-smoking, n=38, cannabis-smoking, n=100); ^ƒ: obtained at RV.

FIGURE 1.

Cannabis smoking, cigarette smoking and vaping in the CANUCK study. a) Quantification of cannabis smoking by joint-years (number of joints per day×number of years smoked) in the cohort (median 7 joint-years, interquartile range (IQR) 1–25 joint-years). b) Number of cannabis-smoking participants by joint-year groups: low (≤5 joint-years), n=61; moderate (>5–20 joint-years), n=39; and high (>20 joint-years), n=39. c) Prevalence of concurrent cigarette smoking and vaping amongst cannabis-smoking participants. Among the 139 cannabis-smoking participants, 84% (n=117) reported concurrent cigarette smoking and vaping (cannabis+vaping, n=47; cannabis+cigarette, n=27; cannabis+vaping+cigarette, n=43).

Cannabis smoking is associated with worse respiratory symptoms and lung function

Cannabis-smoking participants in each joint-year subgroup reported higher SGRQ scores (figure 2a) and CAT scores (figure 2b) than NS participants. The unadjusted increasing linear trends in SGRQ and CAT scores were significant across the groups, and remained significant after adjusting for age, sex, BMI, vaping status and cigarette smoking status (supplementary table E6). There were no differences in SGRQ total scores amongst the cannabis exposure subgroups (supplementary figure E2a). The cannabis+vaping+cigarette subgroup reported higher CAT scores than the cannabis only group (supplementary figure E2b).

FIGURE 2.

Cannabis-smoking participants have more respiratory symptoms and worse airflow limitation. a) St George's Respiratory Questionnaire (SGRQ) total scores in never-smoker (NS) and cannabis-smoking participants. Cannabis-smoking participants had higher SGRQ total scores versus the NS group (median 4.8, IQR 1.08–9.74) in the low (median 8.4, interquartile range (IQR) 4.0–14.9, p=0.01), moderate (median 13.3, IQR 6.8–21.0, p<0.001) and high (median 12.1, IQR 5.9–29.6, p<0.001) joint-year groups. Linear trend test across ordered groups: p<0.001 (unadjusted), p=0.003 (adjusted). b) COPD Assessment Test (CAT) scores in NS and cannabis-smoking participants. Cannabis-smoking participants had higher CAT total scores versus the NS group (median 3, IQR 1–7) in the low (median 6, IQR 3–9, p=0.006), moderate (median 9, IQR 6–14, p<0.001) and high (median 10, IQR 6–16, p<0.001) joint-year groups. Linear trend test across ordered groups: p<0.001 (unadjusted), p=0.0003 (adjusted). c) Pre-bronchodilator forced expiratory volume in 1 s (FEV₁) to forced vital capacity (FVC) ratio in NS and cannabis-smoking participants. The high joint-year group had a lower FEV₁/FVC (mean±sd 74.6±7.2%, p<0.001) than the NS group (82.3±6.9%). Linear trend test across ordered groups: p<0.001 (unadjusted), p=0.006 (adjusted). d) Pre-bronchodilator % predicted forced expiratory flow at 25–75% of FVC (FEF_25–75%) in NS and cannabis-smoking participants. The high joint-year group had a lower FEF_25–75% (mean±sd 79.7±27.1% predicted, p<0.001) than the NS group (99.1±27.2% predicted). Linear trend test across ordered groups: p<0.001 (unadjusted), p=0.024 (adjusted). Adjusted models accounted for age, sex, body mass index, vaping status and cigarette smoking status. *: p<0.05 compared with NS; ^#: p<0.05 in adjusted linear trend test.

Cannabis-smoking participants in the high joint-year group showed lower pre-bronchodilator forced expiratory volume in 1 s (FEV₁) to forced vital capacity (FVC) ratios than NS (figure 2c). The proportion of participants with FEV₁/FVC <lower limit of normal (pre- and post-bronchodilator) did not differ between groups (supplementary figure E3a). Additionally, cannabis-smoking participants in the high joint-year group showed lower % predicted forced expiratory flow at 25–75% of FVC (FEF_25–75%) (figure 2d), a spirometric metric used to identify airflow obstruction in the smaller airways. The unadjusted decreasing linear trends in FEV₁/FVC and FEF_25–75% across groups were significant across the groups. The FEV₁/FVC relationship across groups remained significant after adjusting for age, sex, BMI, vaping and cigarette smoking status (supplementary table E6). The FEF_25–75% % predicted relationship remained significant after adjusting for vaping and cigarette smoking status (supplementary table E6). The cannabis+cigarette subgroup had significantly lower pre-bronchodilator FEV₁/FVC than the cannabis only group (supplementary figure E3b). There were no differences in pre-bronchodilator FEF_25–75% amongst the cannabis exposure subgroups (supplementary figure E3c).

High joint-year exposure is associated with increased radiographic emphysema, airway wall thickness and ventilation abnormalities

Cannabis-smoking participants with moderate and high joint-years had more gas trapping measured by %LAA856 at residual volume (figure 3a), and those with moderate joint-years also had higher DPMs of fSAD (DPM_fSAD) than NS (figure 3c). Cannabis-smoking participants with high joint-years had more radiographic emphysema measured by %LAA950 at total lung capacity and higher DPMs of emphysema (DPM_emphysema) than NS (figure 3b, d). Cannabis-smoking participants with high joint-years also had thicker airway wall measurements (Pi10) than NS (figure 3e). Unadjusted %LAA856, DPM_fSAD and Pi10 showed significant increasing linear trends amongst groups, but became nonsignificant after adjusting for age, sex, BMI, vaping and cigarette smoking status (supplementary table E6).

FIGURE 3.

Cannabis-smoking participants with high joint-years have more abnormalities on computed tomography (CT) and magnetic resonance imaging (MRI). a) Air trapping on quantitative CT (% of low attenuation area (LAA) with density <−856 Hounsfield units (%LAA856) at residual volume) in never-smoker (NS) and cannabis-smoking participants. Cannabis-smoking participants had higher %LAA856 than the NS group (median 1.5%, IQR 0.5–2.7%) in the high (median 5.3%, IQR 2.3–11.4%, p=0.004) and moderate (median 4.2%, IQR 2.5–8.4%, p=0.018) joint-year groups. Linear trend test across ordered groups: p=0.004 (unadjusted), p=0.77 (adjusted). b) Emphysema on quantitative CT (%LAA950 at total lung capacity) in NS and cannabis-smoking participants. The high joint-year group had a higher %LAA950 (median 3.3%, IQR 1.0–4.6%, p=0.042) than the NS group (median 1.0%, IQR 0.2–3.7%). Linear trend test across ordered groups: p=0.29 (unadjusted), p=0.82 (adjusted). c) Disease probability measure of functional small airways disease (DPM_fSAD) in NS and cannabis-smoking participants. The moderate joint-year group had a higher DPM_fSAD (median 14.6, IQR 8.2–27.4, p=0.042) than the NS group (median 7.9, IQR 4.4–15.9). Linear trend test across ordered groups: p=0.04 (unadjusted), p=0.33 (adjusted). d) Disease probability measure of emphysema (DPM_emphysema) in NS and cannabis-smoking participants. The high joint-year group had a higher DPM_emphysema (median 1.1, IQR 0.3–4.3, p=0.006) than the NS group (median 0.2, IQR 0.03–0.9). Linear trend test across ordered groups: p=0.05 (unadjusted), p=0.75 (adjusted). e) Pi10 (square root of the wall area of a hypothetical airway with a luminal perimeter of 10 mm) in NS and cannabis-smoking participants. The high joint-year group had higher Pi10 (median 4.0, IQR 3.9–4.1, p=0.007) than the NS group (median 3.9, IQR 3.9–4.0). Linear trend test across ordered groups: p=0.007 (unadjusted), p=0.22 (adjusted). f) Magnetic resonance imaging (MRI)-determined areas of low ventilation in NS and cannabis-smoking participants. The high joint-year group had a higher percentage of low ventilation regions (median 16.9%, IQR 1.8–26.5%, p=0.010) than the NS group (median 12.7%, IQR 8.3–16.3%). Linear trend test across ordered groups: p<0.001 (unadjusted), p=0.03 (adjusted). g) MRI-determined areas with ventilation defects in NS and cannabis-smoking participants. The high joint-year group had a higher percentage of ventilation defect regions (median 2.49%, IQR 1.43–6.12%, p=0.004) than the NS group (median 1.34%, IQR 0.95–2.16%). Linear trend test across ordered groups: p<0.001 (unadjusted), p=0.13 (adjusted). Adjusted models accounted for age, sex, body mass index, vaping status and cigarette smoking status. *: p<0.05 compared with NS; ^#: p<0.05 in adjusted linear trend test.

On ¹²⁹XeMRI, high joint-year cannabis-smoking participants had greater percentages of areas with low ventilation (figure 3f) and areas with ventilation defects (figure 3g). Across the groups, the unadjusted increasing linear trends in the percentages of ¹²⁹XeMRI low ventilation areas and ventilation defect areas were significant. After adjusting for age, sex, BMI, vaping and cigarette smoking status, the increasing linear trend for low ventilation % remained significant, whilst the increasing trend for ventilation defect % did not (supplementary table E6).

In exposure subgroup analyses, the cannabis+cigarette subgroup showed higher %LAA856, %LAA950, DPM_fSAD and DPM_emphysema than the cannabis-only group (supplementary figure E4a–d and table E5). There were no differences in Pi10 or ¹²⁹XeMRI ventilation measurements amongst exposure subgroups (supplementary figure E4e–g).

Cannabis smoking is associated with changes in airway epithelial immune responses

We investigated biological differences associated with cannabis exposure using airway epithelial brushings from 61 participants (25 non-cannabis-smoking participants, 36 cannabis-smoking participants) who underwent bronchoscopy. The groups had similar age and sex demographics, but the prevalence of vaping and cigarette smoking was higher in the cannabis-smoking group (supplementary table E7). We found 121 significantly DEGs (FDR <0.05) in the cannabis-smoking group compared with the non-cannabis group (figure 4a). Of these, many were involved in mucin (MUC5AC), tissue repair and remodelling (SMAD family member 9 (SMAD9), fibromodulin (FMOD), aldehyde dehydrogenase 3 family member A1 (ALDH3A1), pirin (PIR)) and the anti-inflammatory response (calcitonin related polypeptide-ɑ (CALCA), azurocidin 1 (AZU1)) (supplementary table E8). A full list of DEGs can be found in the supplementary file. We identified seven depleted KEGG pathways involved in adaptive immune responses, including Th17 cell differentiation (figure 4b). Four downregulated Reactome pathways were identified, including regulation of the complement cascade and class A/1 rhodopsin-like receptors and its family member peptide-ligand binding receptors (figure 4c). We found 60 DEGs significantly correlated with joint-year exposure. The top 10 correlated genes ranked by FDR were associated with the inflammatory and immune response (C-C motif chemokine ligand 17 (CCL17), complement C3 (C3), C-X3-C motif chemokine ligand 1 (CX3CL1), SAMHD1) and oxidative stress (alcohol dehydrogenase 7 (ADH7), NAD(P)H quinone dehydrogenase 1 (NQO1), aldo-keto reductase family 1 member B10 (AKR1B10)) (supplementary table E9).

FIGURE 4.

Differentially expressed genes (DEGs) in airway epithelial brushings of cannabis smokers compared to non-cannabis smokers. a) The volcano plot has 121 DEGs with a false discovery rate (FDR) <0.05 highlighted. Blue indicates genes that are downregulated amongst cannabis smokers (compared with non-cannabis smokers). Red indicates genes that are upregulated amongst cannabis smokers (compared with non-cannabis smokers). Labelled genes indicate those with a log fold change (FC) >1. b, c) Gene set enrichment analysis results using Kyoto Encyclopedia of Genes and Genomes (KEGG) (b) and Reactome (c) pathway gene sets are shown. Negative normalised enrichment scores indicate downregulation of pathways in cannabis smokers compared to non-cannabis smokers.

Increased airway epithelial type 2 immune response and mucin gene expression in cannabis smoking

Results from our differential gene expression led us to evaluate specific immune responses in epithelial cells. In peripheral blood, the high joint-year group had higher eosinophil counts than the non-cannabis group (supplementary figure E5). The high joint-year group demonstrated higher airway epithelial type 2 gene signature scores (figure 5a) and inversely demonstrated lower type 17 gene signature scores (figure 5b) than the non-cannabis group. After adjusting for age, sex, BMI, vaping and cigarette smoking status, the type 17 signature score linear trend remained significant, whilst the type 2 signature score linear trend did not (supplementary table E6). Type 1 gene signature scores were similar across groups (supplementary figure E6). Relevant to type 2 gene signature scores, the moderate and high joint-year groups showed significantly higher expression of MUC5AC and MUC5AC:MUC5B ratio than the non-cannabis group (figure 5c, e), and linear trends remained significant following adjustment for age, sex, BMI, vaping and cigarette smoking status (supplementary table E6). The high joint-year group also expressed lower MUC5B than the non-cannabis group (figure 5d).

FIGURE 5.

Increased expression of type 2 immune response and mucin genes in the airway epithelium of high joint-year cannabis-smoking participants. a) The high joint-year group had higher expression (mean±sd 4.6±0.5, p=0.001) of type 2 gene signature by bulk RNA-sequencing than the never-smoker (NS) group (mean±sd 3.7±0.6). Linear trend test across ordered groups: p=0.006 (unadjusted); p=0.14 (adjusted). b) Cannabis-smoking participants had lower expression of type 17 gene signature by bulk RNA-sequencing than the NS group (mean±sd 4.0±1.2) in the moderate (mean±sd 3.1±0.9, p=0.011) and high (mean±sd 2.5±0.9, p<0.001) joint-year groups. Linear trend test across ordered groups: p=0.008 (unadjusted); p=0.02 (adjusted). c) Cannabis-smoking participants had higher expression of mucin 5AC (MUC5AC) by bulk RNA-sequencing than the NS group (mean±sd 8.0±2.3) in the moderate (mean±sd 10.7±1.5, p=0.006) and high (mean±sd 11.8±0.9, p<0.001) joint-year groups. Linear trend test across ordered groups: p<0.001 (unadjusted); p=0.01 (adjusted). d) The high joint-year group had lower expression of mucin 5B (MUC5B) (mean±sd 8.4±1.3, p=0.002) by bulk RNA-sequencing than the NS group (mean±sd 10.0±1.1). Linear trend test across ordered groups: p=0.017 (unadjusted); p=0.05 (adjusted). e) Cannabis-smoking participants had a higher MUC5AC:MUC5B gene expression ratio than the NS group (mean±sd 0.8±0.2) in the moderate (mean±sd 11±0.3, p=0.010) and high (mean±sd 1.4±0.3, p<0.0001) joint-year groups. Linear trend test across ordered groups: p<0.001 (unadjusted); p=0.005 (adjusted). Adjusted models accounted for age, sex, body mass index, vaping status and cigarette smoking status. CPM: counts per million; *: p<0.05 compared with never-smokers; ^#: p<0.05 in adjusted linear trend test.

Cell cultures from cannabis smokers show increased MUC5AC protein in the airway epithelium

Viable airway epithelial cells from a subset of bronchoscopy participants were expanded in ALI cell cultures (supplementary table E10). Histological representation of MUC5AC protein expression in the cultured epithelium are shown in figure 6a. Quantification of the MUC5AC protein confirmed a significantly higher percentage of MUC5AC expression amongst cannabis-smoking participants (figure 6b). Higher MUC5AC expression in ALI cell cultures correlated with higher CAT scores and lower FEV₁/FVC ratios amongst cannabis-smoking participants (figure 6c, d). MUC5AC expression in ALI cell cultures also positively correlated with radiographic emphysema on CT (figure 6e, f) and ventilation abnormalities on ¹²⁹XeMRI (figure 6g, h).

FIGURE 6.

Mucin 5AC (MUC5AC) protein quantification in air–liquid interface (ALI) cell cultures correlates with clinical features in cannabis smokers. a) Representative cross-sectional images from ALI cell cultures stained for MUC5AC (red), with haematoxylin counterstain. Scale bars: 100 µm. b) Quantification of MUC5AC expression as percentage of total area in ALI cell cultures from cannabis smokers (median 29.7%, interquartile range (IQR) 16.2–40.4%) versus non-cannabis smokers (median 21.1%, IQR 8.2–29.0%, p=0.04). c) COPD Assessment Test (CAT) scores reported by cannabis-smoking participants are positively correlated with MUC5AC protein expression in ALI cell cultures (n=20, ρ=0.45, p=0.048). d) Pre-bronchodilator (BD) forced expiratory volume in 1 s (FEV₁)/forced vital capacity (FVC) % reported by cannabis-smoking participants is negatively correlated with MUC5AC protein expression in ALI cell cultures (n=21, ρ= −0.50, p=0.021). e) Percentage of low attenuation area (LAA) with density <−950 Hounsfield units (%LAA950) measured by computed tomography (CT) lung scans in cannabis-smoking participants is positively correlated with MUC5AC protein quantity in ALI cell cultures (n=21, ρ=0.62, p=0.004). f) Disease probability measure of emphysema (DPM_emphysema) measured by CT lung scans in cannabis-smoking participants is positively correlated with MUC5AC protein quantity in ALI cell cultures (n=13, ρ=0.64, p=0.021). g) Magnetic resonance imaging (MRI)-determined percentage area of low ventilation in cannabis-smoking participants is positively correlated with MUC5AC protein quantity in ALI cell cultures (n=21, ρ=0.47, p=0.031). h) MRI-determined percentage area of ventilation defect in cannabis-smoking participants is positively correlated with MUC5AC protein quantity in ALI cell cultures (n=21, ρ=0.64, p=0.002). *: p<0.05.

Discussion

In this study, a patient-to-cell approach uncovered clinical and biological abnormalities in the small airways associated with cannabis smoke exposure. Our cohort reflected real-world cannabis smoking practices in Canada, where more than 80% of participants combined cannabis smoking with cigarette smoking or vaping [16]. Notably, we found an exposure-dependent relationship between cannabis smoking and respiratory symptom scores, even at low joint-years and after adjusting for vaping and cigarette smoking. Vaping and cigarette smoking in addition to cannabis smoking was associated with higher CAT scores compared with cannabis smoking only, suggesting an additive effect of using multiple inhalational vehicles.

We also found that high joint-year cannabis smoking was associated with lower FEV₁/FVC and FEF_25–75%, although the clinical significance of these small differences (compared to NS participants) is unknown. Previous studies have reported conflicting data on the impact of cannabis smoking on lung function [17], possibly due to variability in study populations and levels of cannabis exposure. In the Dunedin study by Hancox et al. [9], cannabis-smoking participants had lower FEV₁/FVC ratios associated with higher FVC values. Similarly, in the National Health and Nutrition Examination Survey, Kempker et al. [18] reported lower FEV₁/FVC ratios attributed to increased FVC. In a population of older adults (>40 years) in the CanCOLD study, Tan et al. [10] found significant FEV₁ decline in those with >20 joint-year histories.

The novelty of our approach to studying the effects of cannabis smoking lies in our use of advanced imaging techniques and bronchoscopy to characterise small airway abnormalities that may be overlooked by conventional diagnostic tools. DPM using paired inspiratory–expiratory CT images is a more sensitive method to detect functional small airway disease and emphysema than conventional quantitative CT [13]. Furthermore, hyperpolarised ¹²⁹XeMRI can detect sub-clinical functional ventilation defects with extraordinary temporal and spatial resolution, whereby the distribution of inhaled ¹²⁹Xe gas throughout the lung identifies areas of ventilatory heterogeneity [14]. In our cohort, these sensitive imaging methods identified ventilatory abnormalities amongst high joint-year cannabis-smoking participants. It is unclear whether these ventilatory abnormalities are reversible with ongoing cannabis smoke exposure or if they represent early precursor signs of airway disease.

We complement these findings with transcriptomic data of small airway epithelial brushings, in which cannabis smoking was associated with DEGs. Specifically, we provide novel transcriptomic evidence of an exposure-dependent, cannabis smoke-associated type 2 immune response and mucin dysregulation in the small airway epithelium. The upregulated type 2 immune response may be signalling an aberrant tissue repair and regeneration mechanism in the airway epithelium in the face of inhaled exposures [19]. Cannabis exposure has been shown in vitro to promote type 2 airway inflammation by activating interleukin-4 (IL-4) and reducing type 1 inflammation by suppressing interferon-γ (IFN-γ) and interleukin-12 (IL-12) in activated T-cells [20]. Type 2 cytokines help to regulate type 1- and type 17-driven inflammation [21]. Relevant to our findings, we observed a reduction in type 17 immune response with an increased type 2 immune response across joint-year groups. Type 2 immunity functions in the lung largely to protect against helminth infections and promote tissue repair, though persistent, dysregulated activation can lead to asthma and allergic inflammation [22]. Therefore, our findings raise the possibility of cannabis-associated changes in the lower airways that share similar biological mechanisms as those observed in patients with type 2-high asthma or airways disease.

The increase we observed in the MUC5AC:MUC5B ratio can likely be attributed to both cannabis smoke exposure and the type 2 immune response, because MUC5AC expression is stimulated by combustible toxicants [23] and the type 2-related cytokine interleukin-13 (IL-13) [24]. In healthy airways, MUC5B is primarily expressed, whilst elevated MUC5AC expression is associated with impaired mucociliary clearance and chronic airway diseases such as asthma or COPD [25]. Our findings are similar to those in cigarette smoke-exposed airways, where single-cell RNA-sequencing of secretory cells in the airway epithelium demonstrates similar increases in the MUC5AC:MUC5B ratio [23], suggesting that combustible smoke itself may be driving this particular feature. We further validated these findings using ALI cell culture models, showing not only that cannabis smoking is associated with higher MUC5AC expression, but also that this expression is significantly correlated with respiratory symptoms, lung function and imaging abnormalities.

There are several notable limitations to this study. First, in this observational study, cannabis smoking was quantified by participant questionnaires, which may be subject to recall bias. Second, whether cannabis smoking is as damaging as cigarette smoking is unanswered in this cohort because it was difficult to recruit a representative cigarette smoking-only group that did not have any underlying respiratory disease. Given the real-world pattern of cannabis smoking with concurrent use of other inhaled substances, it is challenging to isolate the effects of cannabis smoking. In our analyses, we adjusted for the effects of smoking and vaping status, but residual confounding cannot be fully excluded. Furthermore, the sample size of the bronchoscopy subset was limited and did not allow us to address whether combined inhalational exposures had a synergistic or additive effect when compared with cannabis smoking alone. Finally, as a cross-sectional study, we were unable to determine the effects of cannabis smoking cessation or changes in the severity of use on our clinical, physiological, imaging and molecular readouts. Therefore, future investigations should broaden airway epithelial sampling to participants with multiple inhalational exposures, evaluate cannabis-associated effects longitudinally and replicate findings in independent cohorts.

Notwithstanding these limitations, we provide evidence using multiple modes of respiratory structure and function assessments that cannabis smoking is associated with negative effects in the airways, even at low levels of exposure. With global cannabis consumption on the rise, identifying these harms is becoming increasingly important and caution should be exercised as a growing number of nations legalise its use.

Shareable PDF

This PDF extract can be shared freely online.

Shareable PDF

ERJ-01659-2025.Shareable

Supplementary material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material

ERJ-01659-2025.Supplement

Supplementary file

ERJ-01659-2025.File

Data availability

The airway epithelial gene expression dataset is available at GEO (#GSE307690).