Heating and subsequent vaporization of bioactive cannabis compounds followed by inhalation by individuals is a predominant method of cannabis use (National Institute on Drug Abuse, 2020; Lim et al., 2022). Thus, the aim of the study was to gain a better understanding of the impact of heating on microorganisms in cannabis material placed in vaporizers. Initial investigations to determine the microbial levels in all samples showed that 10 g samples from the placebo had no growth on TSA, while 2.18 X 10⁶ CFU/g were recovered from the low potency and 1.76 X 10⁶ CFU/g were recovered from the high potency samples ( Table 1 ). Additionally, when reduced sample sizes (70 mg) were used, rarely (~8%) was any microbial growth detected among the plated placebo replicates and those with growth had a single colony. With this minimal microbial background, the placebo was chosen for the spiking experiments to optimize and assess the microbial recovery from cannabis materials. This lack of microorganisms limited enumeration challenges that may occur if there was a high resident microbial population to account for. E. coli was used for spiking experiments since it is a marker for fecal contamination and may be encountered in outdoor growing settings and with product handling (Thunberg et al., 2002; Simpson et al., 2002). The comparison of the amount of E. coli recovered from the spiked product versus the E. coli reference suspension used for the spiking showed that the recovery was 101.7% (SD: 20.4%) of the spiked amount, indicating that the recovery technique was efficient for extracting microorganisms for the cannabis materials.

Once methods were developed to efficiently recover the spiked E. coli, the experiments focused on investigating the impact of heating on microbial populations. Figure 1 shows the recoveries of E. coli from spiked placebo that was heated for 30 seconds and 70 seconds. The average reduction after heating for 30 seconds (preheating cycle) was 50.4% and after heating for 70 seconds (preheating, plus typical time to fill the vaporizer bag) was 46.1%. There was some experiment-to-experiment variability, especially in the second experiment where the first replicate appeared to have more E. coli following treatment and the other two replicates showing significant reductions (samples 2A, 2B and 2C in Figure 1 ).

Tables 2 and 3 show the genera of bacteria and fungi that were isolated from the cannabis samples with different potencies and their presence before or after heating during the final set of experiments examining the impact of heating on microbial counts. Quantitative results of the impact of heating on the resident microbial populations of placebo, low potency, and high potency samples are shown in Figures 2 and 3 . Overall, few bacteria or fungi were cultured from the placebo samples, regardless of whether heated or not. Bacterial genera identified in the placebo included Bacillus, Staphylococcus and Acinetobacter. For low potency samples, there was some variability across the experiments in the number of colonies on the TSA plates. For example, in experiments 1 and 2 examining the impact of heating on microbial counts, there was an increasing trend in the number of organisms detected with increasing heat exposure times. In experiment 3, the results were the opposite, indicating that there was a reduction in the number of organisms detected with heating. These findings may be an aberration driven by variability in the no-heat samples in experiment 3, that displayed intraexperimental variability, with one set of replicates matching the results of experiments 1 and 2. Bacteria identified included Microbacterium, Bacillus, Terribacillus, Enterobacter, Kosakonia, Pantoea, Siccibacter, and Pseudomonas. For high potency samples, trends in microbial numbers on TSA plates were consistent across all experiments, with a drop in numbers at 30 seconds, but an increase at 70 seconds ( Figure 2 ). Bacteria identified in high potency samples included Enterococcus, Enterobacter, Kosakonia, Pantoea, Siccibacter, Pseudomonas, and Stenotrophomonas species. When the taxonomy data was viewed across the sample types, all bacterial taxa, with the exception of Staphylococcus, that were identified in at least one no-heat sample, were also collected from a sample following heating ( Table 2 ). These data are indicative that heating under the conditions used were generally not sufficient to kill the bacteria present in the products.

Table 4 shows the bacterial taxa detected in the bulk material (pre-plating suspension) from each sample type, while Supplemental Figure S1 shows the principal coordinates analysis (PCoA) plots of the β diversity analysis for the samples. Analysis of the β diversity of the microbial populations present in the cannabis materials showed that the samples from different potency materials tended to group together, regardless of whether they were exposed to heating or not ( Supplemental Figure S1A ). DNA from bacterial taxa including Pantoea and Pseudomonas were more abundantly detected in the low and high potency samples and Pedobacter, Psychrobacillus, Paenibacillus, and Xanthomonas more abundant in the placebo ( Table 4 and Supplemental Table S1 ). Isolates from the genus Pantoea and Pseudomonas were also identified among the cultured organisms from low and high potency samples ( Table 2 ). Conversely, the more abundant taxa detected in the placebo subsamples (Pedobacter, Psychrobacillus, Paenibacillus, and Xanthomonas) were not cultured from the placebo; however, Acinetobacter species which was cultured from the samples accounted for about 2.5% of sequencing reads (i.e. sequencing output) from the placebo. Likewise, Bacillus species and members of the Enterobacteriaceae family which were also cultured also were detected in the sequencing results of the placebo samples, although they demonstrated low abundance with less than 0.1% of sequencing reads ( Supplemental Table S1 ). The fact that the taxa with the highest numbers of sequencing reads from the cannabis samples were not cultured, likely indicated that they were killed during the preparation of the placebo material, yet their DNA remained present in the products.

Our metagenomics analyses are consistent with analyses of others. McKernan and colleagues also used both metagenomics and culture-based approaches to assess microbial populations in cannabis materials (McKernan et al., 2016; McKernan et al., 2021) and detected several species of medical importance including Acinetobacter baumannii, E. coli, Pseudomonas aeruginosa, Salmonella, Staphylococcus aureus, and Streptococcus pneumoniae. When they sequenced DNA from bacterial colonies isolated on agar plates they identified Pseudomonas, Bacillus, Enterobacter and Escherichia strains (McKernan et al., 2021). Our findings identified similar taxa, including Acinetobacter, Escherichia, Pseudomonas, Staphylococcus, Bacillus, Enterobacter, Enterococcus and Stenotrophomonas identified in cannabis material either from colonies growing on agar plates or in the pre-plating suspension ( Tables 2 and 4 , respectively). Members of these genera have been associated with human infections, which is a concern because people who are most susceptible to infections are those with weakened immune systems and such individuals may use cannabis in an effort to relieve effects associated with their preexisting health condition, such as cancer chemotherapy or HIV (Sexton et al., 2016; Rosenthal and Pipitone, 2020). As cannabis is often used as an inhaled product, it is notable that members of genera including Acinetobacter, Bacillus, Pseudomonas and Stenotrophomonas, have been associated with causing opportunistic respiratory infections including pneumonia (Hartzell et al., 2007; Brooke, 2012; Shimoyama et al., 2017; Kumar et al., 2018; Ishida et al., 2019).

For the experiments examining fungal growth, samples were plated on SDA ( Figure 3A ) and PDA ( Figure 3B ). The placebo samples had very low counts, averaging <1 CFU/g on SDA, and fungi identified from the placebo samples included members of the genus Cladosporium, Alternaria, Aspergillus, Irpex, Talaromyces and Gibberella. On the PDA media, there was an observed, though not significant, increase in fungi recovered from the placebo following heating (0.30 CFU/g in no heat samples vs. 1.22 CFU/g in 70-second heated samples). For low potency samples, there was variability in results across the experiments. On the SDA media, the most common result was a slight increase in fungal numbers with 30-second heating, followed by a reduction in 70-second samples to levels close to no-heat levels ( Figure 3A ). Fungal genera identified in low potency samples included Cladosporium, Septoria, Didymella, Epicoccum, Alternaria, Lectera, Gibberella, Paramyrothecium, Bullera and Hannaella. For the PDA media, the general trend was reduction in the total counts with heating (experiment 3 was an exception, Figure 3B ). High potency samples had overall higher fungal counts and more interexperimental variability in counts on SDA media. Alternaria, Diaporthe, Lectera, and Gibberella were detected in these samples. The general trend with the high potency samples seemed to indicate a reduction in fungi with heating, although there were exceptions ( Figure 3A ). When plated on PDA, there was a consistent reduction with the 30-second heating and with experiments 1 and 3 there was a reduction with 70-second heating too ( Figure 3B ). For all types of cannabis materials, the overall differences observed were not significant, indicating that heating did not effectively reduce total fungal numbers in the cannabis materials. When the fungal taxonomy data was assessed among the different sample types and heating conditions, only two genera, Talaromyces and Bullera, were found in at least one of the no-heat samples, but not in any of the samples following heating ( Table 3 ).

With few exceptions the viable fungi identified on a culture plate ( Table 3 ), were also detected in the microbial suspensions that were used for the dilutions and subsequent plating. One notable exception is with members of the genus Filobasidium which were detected in pre-plating suspension but did not appear to be cultured on any of the samples tested; thus, these members of this taxa may not have been viable in the subsamples where its DNA was isolated. Table 5 provides an overview of the fungi that were detected in each of the samples and the PCoA plots displaying the β diversity of the samples is shown in Supplemental Figure S1B . When a taxon was identified on the culture plates, but not in the pre-plating suspension, there was typically a closely related (same family) taxon detected in the suspension. These apparent observed discrepancies may be due to challenges in analyses of metagenomic data that rely on matching to microbial databases and slight variations in the sequencing that can lead to some inconsistency in the identifications, such that the same organism sequenced in separate experiments being identified as closely related organisms. Overall, the taxa identified on the culture plates appeared to represent well what was detected in the suspensions used for the plating. Detailed results of the ITS sequencing experiments in provided in Supplemental Table S2 .

Fungi, including members of the genus Aspergillus, are a concern in inhaled products including cannabis and tobacco products (Verweij et al., 2000; Pauly and Paszkiewicz, 2011; Remington et al., 2015). Several aspergilli have been associated with allergic-type reaction and infections, which are a major concern for immunocompromised patients (Stevens and Melikian, 2011; Ruchlemer et al., 2015; Benedict et al., 2020). Fungi are also a concern due to the ability of certain species to produce mycotoxins including aflatoxins, ochratoxin A, patulin, and sterignatocystin (Varma et al., 1991; Pauly and Paszkiewicz, 2011; McKernan et al., 2015). Aflatoxin B1 can lead to an increased inflammatory response in the respiratory system and potentially lead to other pathologies (Pauly and Paszkiewicz, 2011). We detected members of the genus Aspergillus in samples, including after 70-second heating. Most of the other fungal taxa identified through the sequencing experiments ( Tables 3 and Supplemental Table S2 ) are plant-associated fungi; however, some have been associated with causing allergic reactions and/or infections in highly immunocompromised individuals, thus they may be a concern if inhaled (Yilmaz et al., 2014; Gabriel et al., 2016; Caillaud et al., 2021; Anees-Hill et al., 2021).

An early challenge of the project was optimizing microbial recovery techniques. Our initial efforts were done following the general methodologies described USP <61> methods (U. S. Pharmacopeial Convention, 2019a), which is the compendial chapter for determining microbial counts for non-sterile products used in human drug products. A challenge with this method, is the recommended large sample volume (10 g), which proved impractical in the evaluation of cannabis material used in a vaporizer that holds only approximately 70 mg of sample. Therefore, for most experiments we utilized a smaller 70 mg sample of product that is consistent with the amount that fits in the vaporizer sample holder. This change was not without issues; we initially reduced the DPBS volume to 0.63 ml (maintaining the original w/v ratio), but the volume was too low for efficient solubilization of microorganisms from the cannabis material. The DPBS volume was subsequently increased to 4 ml which greatly improved extraction of bacteria during E. coli spiking experiments and was used in all of the subsequent studies. The high potency samples were particularly difficult to suspend and extract the microorganisms due to the stickiness of the samples from the resin-like compounds. Both low and high potency materials tended to clump together and stick to tubes and pipet tips, which may have led to some variability observed in the recovery experiments. Our research team found similar challenges with smokeless tobacco products, which had a range of moisture and viscosity traits (Han et al., 2016). The increased diluent volumes employed helped to minimize many of these challenges.

The studies described were foundational, but they do have some limitations. Only a limited number of cannabis plant material samples were tested; however, the samples had a range of characteristics that provided valuable data on the impact of heating on microbial levels. The sequencing studies were performed on samples from the final experiment examining the impact of heating on microbial counts after observing a diversity of colony morphologies in the first two experiments that warranted a more in-depth characterization of the microbial populations. The sequencing data from individual TSA and SDA culture plates were aggregated for each cannabis type and heating status in Tables 2 and 3 to provide a more comprehensive picture of the organisms present in each sample and whether those taxa were present following the heating process. With the limited exceptions noted above, bacteria and fungi taxa grew following heating. It should be noted that even though some taxa were not detected following heating, it does not necessarily indicate that heating specifically killed or inactivated these organisms, as they may have been diluted out prior to plating.

Overall, the study demonstrated that heating of the cannabis materials at 190°C in the vaporizer for 70 seconds (settings suggested by the manufacturer to volatize the bioactive compounds) did not lead to significant reductions in microorganisms present in the cannabis materials and in some cases, the heating seemed to enhance the numbers of organisms recovered. This latter observation may be due to the heat volatilizing some residual compounds that may have facilitated “sticking” of microorganisms to the plant material. Finally, it is unknown if the vaporizer airflow could facilitate the transfer of microorganisms from the cannabis material into the vapor collection bag that the end-user inhales. However, the findings demonstrate that heating with a vaporizer may not be considered an effective means to limit microbial hazards associated with inhaled cannabis use. Therefore, other microbial reduction methodologies, such as gamma irradiation, ultraviolet germicidal irradiation or pasteurization may be needed to limit microbial hazards in cannabis material (Hazekamp, 2016; Boyar et al., 2021).