Background:

Gut microbial alterations have been linked to chronic liver disease and hepatocellular carcinoma (HCC). The role of the oral microbiome in liver cancer development has not been widely investigated.

Methods:

Bacterial 16S rRNA sequences were evaluated in oral samples from 90 HCC cases and 90 controls who were a part of a larger U.S. case–control study of HCC among patients diagnosed from 2011 to 2016.

Results:

The oral microbiome of HCC cases showed significantly reduced alpha diversity compared with controls (Shannon P = 0.002; Simpson P = 0.049), and beta diversity significantly differed (weighted Unifrac P = 0.004). The relative abundance of 30 taxa significantly varied including Cyanobacteria, which was enriched in cases compared with controls (P = 0.018). Cyanobacteria was positively associated with HCC [OR, 8.71; 95% confidence interval (CI), 1.22–62.00; P = 0.031] after adjustment for age, race, birthplace, education, smoking, alcohol, obesity, type 2 diabetes, Hepatitis C virus (HCV), Hepatitis B virus (HBV), fatty liver disease, aspirin use, other NSAID use, laboratory batch, and other significant taxa. When stratified by HCC risk factors, significant associations of Cyanobacteria with HCC were exclusively observed among individuals with negative histories of established risk factors as well as females and college graduates. Cyanobacterial genes positively associated with HCC were specific to taxa producing microcystin, the hepatotoxic tumor promotor, and other genes known to be upregulated with microcystin exposure.

Conclusions:

Our study provides novel evidence that oral Cyanobacteria may be an independent risk factor for HCC.

Impact:

These findings support future studies to further examine the causal relationship between oral Cyanobacteria and HCC risk.

The incidence of liver cancer has been steadily increasing in the United States with rates nearly tripling over the past three decades. (1, 2) Annually, nearly 42,000 U.S. residents are diagnosed with liver cancer, and over 80% of patients die of the disease within 5 years. (1) The rising incidence of hepatocellular carcinoma (HCC), the most common primary liver cancer, has been largely attributed to the growing prevalence of chronic hepatitis C virus (HCV) and, more recently, metabolic conditions including obesity, type 2 diabetes, and nonalcoholic fatty liver disease (NAFLD; ref. 3). Nonetheless, the growing burden of HCC suggests the influence of unrecognized factors.

Gut microbial alterations have been linked to chronic liver disease and HCC and may involve a number of mechanisms including the production of DNA-damaging metabolites, activation of toll-like receptor signaling and TNFα production, alteration of bile and choline availability, metabolism of ethanol to acetaldehyde, proinflammatory effects, and the promotion of insulin resistance (4–6). There is evidence that gut microbial changes observed in liver disease may originate in the oral cavity through regular passage of bacteria from the mouth to the intestine (4). Over half of gut bacteria enriched in patients with cirrhosis are typically found in the oral cavity (4), suggesting mouth-to-gut microbial translocation. Reductions in oral commensal bacteria and increases in pathogenic species have been observed in patients with decompensated cirrhosis (7). Passage of oral bacteria to distal organs may also be facilitated in periodontal disease, a major manifestation of oral dysbiosis, in which accumulated bacteria and bacterial metabolites can travel through the bloodstream to other sites (8). Perturbations of the oral microbiome, including specific pathogenic bacteria, have been associated with cancers of the head and neck, esophagus, and pancreas (9–11). Periodontal disease has been linked to chronic liver disease and liver cancer risk (12–14). However, the role of the oral microbiome in liver cancer development has not been widely investigated. The objective of our study was to evaluate the relationship of the oral bacterial microbiome with HCC.

The study population was comprised of 90 HCC cases and 90 controls with oral samples obtained from a larger population-based case–control study of 673 patients with HCC and 1,166 controls. Details of the study have been previously described (15). Patients with HCC included males and females aged 35 to 84 years diagnosed from 2011 to 2016 identified through the NCI Surveillance, Epidemiology, and End Results (SEER) cancer registries in Connecticut, operated by Yale Cancer Center, and in New Jersey, managed by the Rutgers Cancer Institute of New Jersey, and through the Liver Transplant Center at Columbia University Irving Medical Center. Control subjects were adult males and females without cancer selected from the general population through random digit dialing in Connecticut and New Jersey. Written informed consent was obtained from study subjects and the investigation was approved by the institutional review boards at each recruitment site. The study was conducted in accordance with the U.S. Common Rule.

Risk factor information was collected, including demographic data, anthropometry, cigarette smoking, alcohol use, medical history (viral hepatitis, type 2 diabetes, fatty liver disease, cirrhosis), and regular use of aspirin and other NSAIDs. A total of 501 cases and 989 controls provided self-collected saliva/oral fluid samples using the Saliva Self-Collection Kit (OG-250, DNA Genotek) and the OraSure oral specimen collection device (OraSure Technologies, Inc.) according to the manufacturers' instructions. Total DNA was extracted from saliva/oral fluid samples using commercial reagents. These samples were used for HCV assessment and genomic analyses. Residual DNA samples were utilized for the present study. Ninety HCC cases were selected to include approximately equal representation by race/ethnicity including (non-Hispanic) Whites, Blacks, and Asians, and Hispanics. Ninety controls were selected with approximate frequency matching to cases by sex and race/ethnicity.

16S rRNA sequencing

Oral DNA samples (5 ng/μL) were amplified in PCR reactions targeting the V3-V5 (16) region of the bacterial 16S rRNA gene. The initial PCR utilized custom primers ligated with overhang Illumina adapter sequences (Integrated DNA Technologies, Inc.). Amplicons were purified using Agencourt AMPure XP beads (Beckman Coulter). To allow for multiplexing, a second PCR was run to attach unique indices to each end of 16S amplicons using the Nextera XT Index Kit (Illumina). Bar-coded amplicons were purified, quantified by Nanodrop 2000 (Thermo Fisher), normalized to equal concentrations, pooled, and checked on agarose gels for library sizes. Bar-coded libraries (50ul of 100 ng/μL) were assayed by 2 × 300 paired-end sequencing using the 600-cycle v3 kit on the MiSeq platform (Illumina) at the University of Hawaii.

Bioinformatic and statistical analyses

Sequencing data were processed using QIIME2 v.2020.2 (17). Raw sequences were demultiplexed and trimmed of primers followed by denoising and removal of chimeras using DADA2 (18). Filtered sequences were clustered into amplicon sequence variants, and representative sequences were aligned to construct phylogenetic trees. Taxonomic classification was optimized by extracting reads from the Greengenes 13.8 reference database specific to V3-V5. The naive Bayes classifier was trained against this reference database and taxonomy assigned to representative sequences using the trained classifier. The microbial database of the National Center for Biotechnology Information (NCBI; ref. 19) was utilized for additional sequence characterization.

SAS version 9.4 (SAS Institute) was used for statistical analyses of data. Tests with P < 0.05 were considered to be significant. The oral microbiomes of HCC cases and controls were compared by diversity indices and taxa composition. Taxa composition was evaluated by relative abundance, which was estimated based on the number of 16S sequences assigned to a taxon as a proportion of the total sequences in a sample. Alpha diversity was measured by Shannon, Simpson, Chao1, Faith's Phylogenetic Diversity Whole Tree, and Operational Taxonomic Unit indices. Beta diversity was evaluated by principal coordinate analysis and UniFrac distances with comparisons between groups measured by PERMANOVA.

Bacterial taxa relative abundance was evaluated as both categorical and continuous variables using the χ2, Wilcoxon two-sample and Kruskal–Wallis tests. Low/high levels of taxa were defined using cutpoints based on the mean relative abundance of controls. Univariate logistic regression with Firth bias correction (20) was used to model the association of individual taxa (low/high levels) with HCC status. Taxa significantly associated with HCC were each individually evaluated in multivariate models which included established liver cancer risk factors and other factors significantly differing between HCC cases and controls. Taxa remaining significant in multivariate analyses were jointly evaluated in a final multivariate model (Cirrhosis was not included as a covariate in any models as it is an immediate clinical precursor to HCC. Household income, which was missing from more than 10% of the study population, was excluded from the analyses).

The bacterial functional gene composition was inferred using PICRUSt2 (21). Metagenomic comparisons were made utilizing the Kruskal–Wallis tests and the Benjamini–Hochberg procedure to account for FDR due to multiple testing. Correlation of bacterial taxa and gene composition employed the Spearman correlation coefficient.

Data sharing

The complete 16S sequencing dataset is available through the NCBI (biosample accession number SAMN18915539).

Compared with controls, HCC cases were older and comprised larger proportions of non-Whites, non–US-born individuals, and those with a high school education or lower (Table 1). HCV was the major risk factor among HCC cases with nearly 58% affected. In addition to HCV, other liver cancer risk factors more predominant in cases compared with controls included hepatitis B virus (HBV), cigarette smoking, type 2 diabetes, fatty liver disease, and cirrhosis. A history of regular aspirin and other NSAID use were less prevalent in cases compared with controls. Obesity (body mass index ≥30 kg/m2) and alcohol consumption did not significantly vary between cases and controls.

Table 1.

Characteristics of HCC cases and controls.

Cases (n = 90)Controls (n = 90)Cases (n = 90)Controls (n = 90)
N%N%PaN%N%Pa
Sex  BMI  
 Male 66 73.3 55 61.1 0.081  <30 kg/m2 72 84.7 64 72.3 0.055 
 Female 24 26.7 35 38.9   ≥30 kg/m2 13 15.3 24 27.3  
Age group  Cigarette smoking  
 <60 31 34.4 45 51.7 0.02  No 27 30 55 62.5 <0.0001 
 60+ 59 65.6 42 48.3   Yes 63 70 33 37.5  
Race/Ethnicity  Alcohol consumption  
 White 20 22.2 41 45.6 0.003  No 28 31.1 38 43.2 0.096 
 Black 20 22.2 21 23.3   Yes 62 68.9 50 56.8  
 Hispanic 20 22.2 13 14.4  Type 2 diabetes  
 Other 30 33.3 15 16.7   No 50 56.2 71 80.7 0.001 
Birthplace  Yes 39 43.8 17 19.3  
 US 53 58.9 64 75.3 0.021 Fatty liver disease  
 Non- U.S. 37 41.1 21 24.7   No 50 56.2 71 80.7 0.001 
Marital status  Yes 39 43.8 17 19.3  
 Single/never married 15 17.2 10.3 0.29 Cirrhosis  
 Married 50 57.5 50 57.5   No 70 79.6 88 100 <0.0001 
 Separated/divorced 17 19.5 17 19.5   Yes 18 20.4  
 Widowed 5.8 11 12.6  Aspirin use  
Education  No 64 75.3 53 60.2 0.034 
 High school or lower 41 46.1 9.3 <0.0001  Yes 21 24.7 35 39.8  
 College 37 41.6 48 55.8  Other NSAID use  
 Graduate school 11 12.4 30 34.9   No 80 94.1 72 81.8 0.013 
Viral hepatitis   Yes 5.9 16 19.2  
 Negative 25 27.8 86 95.6 <0.0001       
 HBV 13 14.4        
 HCV 48 53.3 4.4        
 HBV and HCV 4.4        
Cases (n = 90)Controls (n = 90)Cases (n = 90)Controls (n = 90)
N%N%PaN%N%Pa
Sex  BMI  
 Male 66 73.3 55 61.1 0.081  <30 kg/m2 72 84.7 64 72.3 0.055 
 Female 24 26.7 35 38.9   ≥30 kg/m2 13 15.3 24 27.3  
Age group  Cigarette smoking  
 <60 31 34.4 45 51.7 0.02  No 27 30 55 62.5 <0.0001 
 60+ 59 65.6 42 48.3   Yes 63 70 33 37.5  
Race/Ethnicity  Alcohol consumption  
 White 20 22.2 41 45.6 0.003  No 28 31.1 38 43.2 0.096 
 Black 20 22.2 21 23.3   Yes 62 68.9 50 56.8  
 Hispanic 20 22.2 13 14.4  Type 2 diabetes  
 Other 30 33.3 15 16.7   No 50 56.2 71 80.7 0.001 
Birthplace  Yes 39 43.8 17 19.3  
 US 53 58.9 64 75.3 0.021 Fatty liver disease  
 Non- U.S. 37 41.1 21 24.7   No 50 56.2 71 80.7 0.001 
Marital status  Yes 39 43.8 17 19.3  
 Single/never married 15 17.2 10.3 0.29 Cirrhosis  
 Married 50 57.5 50 57.5   No 70 79.6 88 100 <0.0001 
 Separated/divorced 17 19.5 17 19.5   Yes 18 20.4  
 Widowed 5.8 11 12.6  Aspirin use  
Education  No 64 75.3 53 60.2 0.034 
 High school or lower 41 46.1 9.3 <0.0001  Yes 21 24.7 35 39.8  
 College 37 41.6 48 55.8  Other NSAID use  
 Graduate school 11 12.4 30 34.9   No 80 94.1 72 81.8 0.013 
Viral hepatitis   Yes 5.9 16 19.2  
 Negative 25 27.8 86 95.6 <0.0001       
 HBV 13 14.4        
 HCV 48 53.3 4.4        
 HBV and HCV 4.4        

Note: Missing values: age (3 controls); birthplace (5 controls); marital status (3 cases, 3 controls); education (1 case; 4 controls); BMI (5 cases, 2 controls; smoking (2 controls); alcohol (2 controls); type 2 diabetes (1 case, 2 controls); fatty liver disease (2 cases, 2 controls); cirrhosis (2 controls); aspirin (5 cases, 2 controls); other NSAIDs (5 cases, 2 controls).

Abbreviation: BMI, body mass index.

aχ2 test.

In order to ascertain biases in the study population, their characteristics were compared with that of the overall case–control study population previously described (15). The study population subset differed from the overall study population with respect to race/ethnicity, marital status, and birthplace. Compared with the overall case study population, the case subset included larger proportions of non-Whites (28% vs. 78%, P < 0.0001), and those born outside of the United States (17% vs. 41%, P < 0.0001). Relative to the overall control study population, the control subset included larger proportions of non-Whites (7% vs. 54%, P < 0.0001), those born outside of the United States (8% vs. 25%, P < 0.0001), and a smaller proportion of married individuals (67% vs. 58%, P = 0.03). Liver cancer risk factors and other medical history (including fatty liver, cirrhosis, aspirin, other NSAID) did not significantly differ in the subset of 90 cases and 90 controls compared with the overall study population. An exception was viral hepatitis status– relative to the overall case study population, the case subset included larger proportions of individuals with HBV (3% vs. 14%, P < 0.0001); compared with the overall control study population, the control subset included a larger proportion of individuals with a history of HCV (1% vs. 4%, P = 0.0035).

Oral DNA samples yielded a total of 32,972,048 demultiplexed bacterial sequences with an average of 184,201 sequences per sample. After quality filtering, 10,687 representative sequences were yielded with a mean length of 561 base pairs per sequence. Taxonomic assignment identified a total of 13 phyla, 122 genera, and 83 species in oral samples.

Compared with controls, HCC cases showed significantly reduced alpha diversity based on Shannon diversity (5.84 and 6.23, respectively; P = 0.002), Simpson diversity (0.96 and 0.97, respectively; P = 0.049; Fig. 1), and observed Operational Taxonomic Units (OTU; 175.68 and 213.58, respectively; P < 0.0001). Beta diversity significantly differed between cases and controls (unweighted and weighted Unifrac; P = 0.01 and P = 0.004; Fig. 2).

Figure 1.

Oral microbiome showed significantly reduced alpha diversity in HCC cases compared with controls. Shannon diversity index 5.84 and 6.23, respectively, P = 0.002; Simpson diversity index 0.96 and 0.97, respectively, P = 0.049.

Figure 1.

Oral microbiome showed significantly reduced alpha diversity in HCC cases compared with controls. Shannon diversity index 5.84 and 6.23, respectively, P = 0.002; Simpson diversity index 0.96 and 0.97, respectively, P = 0.049.

Close modal
Figure 2.

Oral microbiome beta diversity in cases (red) and controls (blue) based on weighted Unifrac (P = 0.004).

Figure 2.

Oral microbiome beta diversity in cases (red) and controls (blue) based on weighted Unifrac (P = 0.004).

Close modal

Thirty taxa were significantly enriched or reduced in HCC. At the phylum level, Cyanobacteria, was 2.3-fold higher in HCC cases and Tenericutes 0.74-fold lower compared with controls. Seven genera and 1 species were enriched (ranging from 1.06–35–fold higher) in HCC cases relative to controls and 12 genera and 8 species were reduced (ranging from 0.27–0.93–fold).

The 30 taxa differentially enriched or reduced in HCC were individually evaluated in multivariate models which included sex, age group (<60, ≥60 years), race/ethnicity (White, Black, Hispanic, Asian, and Others), education (high school or lower, college, or higher), birthplace (United States, non–United States), HBV, HCV, alcohol, smoking, obesity, type 2 diabetes, fatty liver disease, aspirin use, other NSAID use, and laboratory batch. Five taxa remained significant in multivariate analyses including Cyanobacteria, Chryseobacterium, Neisseria elongata, Shuttleworthia satelles, Veillonella dispar. These 5 taxa were jointly evaluated in the final model. In the final model, only Cyanobacteria remained independently, positively associated with HCC [adjusted odds ratio (OR) 8.71; 95% CI, 1.22–62.00; P = 0.031) after adjustment for age, race, birthplace, education, smoking, alcohol, obesity, type 2 diabetes, HCV, HBV, fatty liver disease, aspirin use, other NSAID use, laboratory batch, and other significant taxa (Table 2 and Fig. 3). HCV, type 2 diabetes, and a ≤ high school education were also each independently, positively associated with HCC.

Table 2.

Association of risk factors with HCC.

Adjusted ORa (95% CI)P
Cyanobacteria (high vs. low)b 8.71 (1.22–62.00) 0.031 
Chryseobacterium (high vs. low)b 17.27 (0.36–838.81) 0.151 
Neisseria elongata (high vs. low)b 0.19 (0.02–1.69) 0.136 
Shuttleworthia satelles (high vs. low)b 0.23 (0.05–1.12) 0.070 
Veillonella dispar (high vs. low)b 0.30 (0.08–1.09) 0.067 
Sex: Male (vs. Female) 1.77 (0.41–7.68) 0.447 
Age: ≥60 (vs.<60 years) 2.28 (0.62–8.41) 0.216 
Education: high school or lower (vs. college or higher) 8.66 (1.75–42.77) 0.008 
Race: Black (vs. White) 5.27 (0.07–397.84) 0.637 
Hispanic (vs. White) 3.27 (0.37–28.88) 0.939 
Other race (vs. White) 0.23 (0.003–19.76) 0.748 
Birthplace: US (vs. non- U.S.) 0.32 (0.05–2.29) 0.256 
Fatty liver disease (yes vs. no) 15.95 (0.93–274.29) 0.056 
Aspirin use (yes vs. no) 0.52 (0.12–2.25) 0.384 
Nonsteroidal anti-inflammatory drug use (yes vs. no) 2.02 (0.29–14.19) 0.481 
Cigarette smoking (yes vs. no) 2.85 (0.70–11.66) 0.146 
Excess alcohol consumption (yes vs. no) 0.71 (0.17–2.88) 0.629 
BMI (≥30 kg/m2 vs. <30 kg/m20.31 (0.07–1.39) 0.124 
Type 2 diabetes (yes vs. no) 5.18 (1.21–22.28) 0.027 
HCV (yes vs. no) 55.14 (10.78–282.07) <0.0001 
HBV (yes vs. no) 8.43 (0.54–131.22) 0.128 
Laboratory batch 0.24 (0.00–17.78) 0.514 
Adjusted ORa (95% CI)P
Cyanobacteria (high vs. low)b 8.71 (1.22–62.00) 0.031 
Chryseobacterium (high vs. low)b 17.27 (0.36–838.81) 0.151 
Neisseria elongata (high vs. low)b 0.19 (0.02–1.69) 0.136 
Shuttleworthia satelles (high vs. low)b 0.23 (0.05–1.12) 0.070 
Veillonella dispar (high vs. low)b 0.30 (0.08–1.09) 0.067 
Sex: Male (vs. Female) 1.77 (0.41–7.68) 0.447 
Age: ≥60 (vs.<60 years) 2.28 (0.62–8.41) 0.216 
Education: high school or lower (vs. college or higher) 8.66 (1.75–42.77) 0.008 
Race: Black (vs. White) 5.27 (0.07–397.84) 0.637 
Hispanic (vs. White) 3.27 (0.37–28.88) 0.939 
Other race (vs. White) 0.23 (0.003–19.76) 0.748 
Birthplace: US (vs. non- U.S.) 0.32 (0.05–2.29) 0.256 
Fatty liver disease (yes vs. no) 15.95 (0.93–274.29) 0.056 
Aspirin use (yes vs. no) 0.52 (0.12–2.25) 0.384 
Nonsteroidal anti-inflammatory drug use (yes vs. no) 2.02 (0.29–14.19) 0.481 
Cigarette smoking (yes vs. no) 2.85 (0.70–11.66) 0.146 
Excess alcohol consumption (yes vs. no) 0.71 (0.17–2.88) 0.629 
BMI (≥30 kg/m2 vs. <30 kg/m20.31 (0.07–1.39) 0.124 
Type 2 diabetes (yes vs. no) 5.18 (1.21–22.28) 0.027 
HCV (yes vs. no) 55.14 (10.78–282.07) <0.0001 
HBV (yes vs. no) 8.43 (0.54–131.22) 0.128 
Laboratory batch 0.24 (0.00–17.78) 0.514 

aAdjusted for all variables listed.

bDefined using cut-off points based on the mean relative abundance in controls.

Figure 3.

Enrichment of Cyanobacteria in HCC cases (0.42%) compared with controls (0.19%; P = 0.018). Mean relative abundance depicted in horizontal line.

Figure 3.

Enrichment of Cyanobacteria in HCC cases (0.42%) compared with controls (0.19%; P = 0.018). Mean relative abundance depicted in horizontal line.

Close modal

The relationship of Cyanobacteria with HCC was further examined stratified by individual risk factors. The association of Cyanobacteria with HCC varied by diabetes, alcohol, obesity, HCV, HBV, fatty liver disease, aspirin use, and other NSAID use with significant associations observed for those with a negative history of each risk factor but not among those with a positive history (Fig. 4). The relationships also varied by sex and education with significant associations limited to females and college graduates. The joint effects of Cyanobacteria and these risk factors were evaluated in separate models using interaction terms and none were significant with the exception of sex (P = 0.035). In the full, multivariate model, the joint effect of Cyanobacteria and sex with HCC risk was not significant (P = 0.062).

Figure 4.

Association of Cyanobacteria with HCC risk by etiology. Significant associations observed for females, college graduates, and those with a negative history of HBV, HCV, type 2 diabetes, alcohol, obesity, fatty liver disease, aspirin use, and other NSAID use (*, No observation).

Figure 4.

Association of Cyanobacteria with HCC risk by etiology. Significant associations observed for females, college graduates, and those with a negative history of HBV, HCV, type 2 diabetes, alcohol, obesity, fatty liver disease, aspirin use, and other NSAID use (*, No observation).

Close modal

Interrogation of individual 16S sequences classified as Cyanobacteria in the NCBI database yielded 114 candidate species from 65 genera representing marine, freshwater, and terrestrial Cyanobacteria. Inferred metagenomic analyses identified bacterial genes correlated with Cyanobacteria (Table 3). Four of these bacterial genes were highly significantly correlated with Cyanobacteria (r = 0.99, P < 0.0001), indicating that these genes were specifically from Cyanobacteria, and also significantly enriched in HCC cases (FDR p < 0.05) all-trans-8′-apo-beta-carotenal 15,15′-oxygenase, coenzyme F420 hydrogenase, magnesium-protoporphyrin IX monomethyl ester (oxidative) cyclase, and plastoquinol–plastocyanin reductase. Nitric oxide dioxygenase was also significantly enriched in HCC cases (FDR p < 0.05) and correlated with Cyanobacteria (r = 0.62, p < 0.0001). Stearoyl-CoA 9-desaturase (SCD1), which did not differ between cases and controls, was correlated with Cyanobacteria (r = 0.57, p < 0.0001).

Table 3.

Bacterial genes: Correlation with Cyanobacteria and enrichment in HCC cases.

Correlation with CyanobacteriaEnriched in HCC Cases (FDR, P <0.0.05)
Bacterial geneaRbPCases
All-trans-8′-apo-beta-carotenal 15,15′-oxygenase 0.99 <0.0001 
Coenzyme F420 hydrogenase 0.99 <0.0001 
Magnesium-protoporphyrin IX monomethyl ester (oxidative) cyclase 0.99 <0.0001 
Plastoquinol–plastocyanin reductase 0.99 <0.0001 
Nitric oxide dioxygenase 0.62 <0.0001 
Stearoyl-CoA 9-desaturase 0.57 <0.0001  
S-(hydroxymethyl)glutathione dehydrogenase 0.36 <0.0001  
Catalase peroxidase 0.34 <0.0001  
3-hydroxyisobutyrate dehydrogenase 0.33 <0.0001  
4-hydroxyphenylpyruvate dioxygenase 0.32 <0.0001  
Sorbitol-6-phosphate 2-dehydrogenase 0.30 <0.0001  
3-alpha(or 20-beta)-hydroxysteroid dehydrogenase 0.30 <0.0001  
Cholesterol oxidase 0.29 0.0001  
L-lysine N(6)-monooxygenase (NADPH) 0.28 0.0001 
Pyrimidine monooxygenase 0.27 0.0002 
Cyclohexanone monooxygenase 0.26 0.0004 
Alkanesulfonate monooxygenase 0.25 0.0006  
Precorrin-3B synthase 0.25 0.0007  
4-hydroxymandelate oxidase 0.25 0.0009 
Protocatechuate 3,4-dioxygenase 0.24 0.001 
Decaprenylphospho-beta-D-erythro-pentofuranosid-2-ulose 2-reductase 0.24 0.0012  
Gluconate 2-dehydrogenase (acceptor) 0.24 0.0012  
Alcohol dehydrogenase (azurin) 0.24 0.0012  
Decaprenylphospho-beta-D-ribofuranose 2-dehydrogenase 0.23 0.0016  
D-galactose 1-dehydrogenase 0.23 0.0016 
Alcohol dehydrogenase 0.23 0.0023  
5,6-dimethylbenzimidazole synthase 0.23 0.0023  
Quinoprotein glucose dehydrogenase (PQQ, quinone) 0.23 0.0023 
4-phosphoerythronate dehydrogenase −0.23 0.0021  
Correlation with CyanobacteriaEnriched in HCC Cases (FDR, P <0.0.05)
Bacterial geneaRbPCases
All-trans-8′-apo-beta-carotenal 15,15′-oxygenase 0.99 <0.0001 
Coenzyme F420 hydrogenase 0.99 <0.0001 
Magnesium-protoporphyrin IX monomethyl ester (oxidative) cyclase 0.99 <0.0001 
Plastoquinol–plastocyanin reductase 0.99 <0.0001 
Nitric oxide dioxygenase 0.62 <0.0001 
Stearoyl-CoA 9-desaturase 0.57 <0.0001  
S-(hydroxymethyl)glutathione dehydrogenase 0.36 <0.0001  
Catalase peroxidase 0.34 <0.0001  
3-hydroxyisobutyrate dehydrogenase 0.33 <0.0001  
4-hydroxyphenylpyruvate dioxygenase 0.32 <0.0001  
Sorbitol-6-phosphate 2-dehydrogenase 0.30 <0.0001  
3-alpha(or 20-beta)-hydroxysteroid dehydrogenase 0.30 <0.0001  
Cholesterol oxidase 0.29 0.0001  
L-lysine N(6)-monooxygenase (NADPH) 0.28 0.0001 
Pyrimidine monooxygenase 0.27 0.0002 
Cyclohexanone monooxygenase 0.26 0.0004 
Alkanesulfonate monooxygenase 0.25 0.0006  
Precorrin-3B synthase 0.25 0.0007  
4-hydroxymandelate oxidase 0.25 0.0009 
Protocatechuate 3,4-dioxygenase 0.24 0.001 
Decaprenylphospho-beta-D-erythro-pentofuranosid-2-ulose 2-reductase 0.24 0.0012  
Gluconate 2-dehydrogenase (acceptor) 0.24 0.0012  
Alcohol dehydrogenase (azurin) 0.24 0.0012  
Decaprenylphospho-beta-D-ribofuranose 2-dehydrogenase 0.23 0.0016  
D-galactose 1-dehydrogenase 0.23 0.0016 
Alcohol dehydrogenase 0.23 0.0023  
5,6-dimethylbenzimidazole synthase 0.23 0.0023  
Quinoprotein glucose dehydrogenase (PQQ, quinone) 0.23 0.0023 
4-phosphoerythronate dehydrogenase −0.23 0.0021  

aNinety bacterial genes were significantly correlated with Cyanobacteria; those with r < 0.23 or < −0.23 are not shown.

bSpearman correlation.

To assess factors influencing microbial composition independent of liver cancer, controls were separately evaluated. Among controls, alpha diversity was higher in males than in females (Shannon diversity, P = 0.019; Simpson diversity, P = 0.01; Observed OTU, P = 0.006). Beta diversity (weighted Unifrac) varied between Whites and non-Whites (P = 0.037) and by HCV status (P = 0.018). Cyanobacteria relative abundance did not vary by HCC risk factors among controls.

Our study provides unique preliminary evidence linking HCC to oral dysbiosis, including reduced microbial richness and evenness and changes in the abundance of specific taxa. Most notably, we report for the first time that oral Cyanobacteria may play a role in the development of HCC. Cyanobacteria, which comprised less than 1% of the overall oral microbiome, was positively associated with HCC after controlling for established liver cancer risk factors and other taxa enriched or reduced in HCC. Interestingly, when further examined individually by key risk factors and other characteristics, the association of Cyanobacteria with HCC was observed exclusively among those with a negative history of diabetes, alcohol, obesity, HCV, HBV, and fatty liver disease, aspirin use, and other NSAID use as well as females and college graduates. Aspirin was previously found to be inversely associated with HCC in our larger study population (15) and in other studies (22, 23). Lower education, which was more prevalent in HCC cases in the larger study population, remained a significant risk factor for HCC in the present analyses. The incidence of HCC is 2- to 4-fold lower in females than males (1) and the basis of this disparity is not entirely understood. Among controls, oral bacterial diversity was significantly higher in males indicating underlying microbial differences between the sexes. Nonetheless, the relative abundance of Cyanobacteria did not vary by sex or by HCC risk factors when evaluated among controls, indicating that differences in Cyanobacteria levels alone do not account for the etiologic variation in the association of Cyanobacteria and HCC. The relationship of Cyanobacteria with HCC specifically in the absence of known risk factors underscores the possibility that oral Cyanobacteria may be an unrecognized independent risk factor for HCC. While the multifactorial etiology of HCC in the United States has been well characterized, there is evidence that a substantial proportion of patients diagnosed with HCC have no underlying risk factors. In a cohort of over 11,000 HCC cases diagnosed in the United States in 2000 to 2014, 24% had no known etiology (24).

Cyanobacteria are photosynthetic, gram-negative bacteria that occur in all terrestrial and aquatic ecosystems (25, 26). Cyanobacteria include over 2,600 species which produce a wide range of secondary metabolites, including toxins targeting the liver, nervous system, skin, and gastrointestinal tract (25, 26). Three cyanotoxins have recognized or putative tumor-promoting properties in the liver including microcystin, nodularin, and cylindrospermopsin (25). The candidate Cyanobacteria species identified in oral samples included known producers of all three of these cyanotoxins including of the genera Anabaena, Anabaenopsis, Aphanizomenon, Cylindrospermopsis, Dolichospermum, Fischerella, Lyngbya, Nodularia, Nostoc, Oscillatoria, Planktothrix, Raphidiopsis, and Synechococcus (25, 27, 28).

Microcystins, the most common and well characterized of the cyanotoxins, are cyclic peptides produced by multiple species and include more than 80 variants (26). Microcystins are classified as group 2B carcinogens, or possibly carcinogenic to humans (26). Microcystins function as tumor promotors, primarily through inhibition of protein phosphatases 1 and 2A, resulting in excessive phosphorylation of intermediate filaments and microfilaments and damage to the cytoskeleton of hepatocytes (29).

Importantly, our findings specifically implicate microcystins in the association of Cyanobacteria with HCC risk. Two of the bacterial genes highly correlated with Cyanobacteria, all-trans-8′-apo-beta-carotenal 15,15′-oxygenase, which is involved in retinal biosynthesis (30), and plastoquinol-plastocyanin reductase, a key enzyme in photosynthesis, were also significantly enriched in HCC. These genes are specifically found in microcystin-producing genera, including Nostoc and Synechococcus (28, 31), which were among the candidate Cyanobacteria identified in oral samples.

Cyanobacteria are particularly abundant in nutrient-rich bodies of water where they can form blooms, which increases the risk of human exposure (25). Blooms containing toxin-producing Cyanobacteria are found in fresh, brackish, and marine waters globally and are increasing in the United States and worldwide due to changes in global temperatures combined with eutrophication of water bodies (25). The evidence linking cyanotoxins to HCC in human populations have been largely limited to ecologic studies. Studies in China have observed increased risk of liver cancer in communities with microcystin-contaminated drinking water sources (32, 33). Using remote sensing, Cyanobacteria blooms were documented in water bodies in over 60% of US counties and significant clusters of NAFLD death were linked to bloom-rich coastal areas (34). Cyanobacteria blooms have also been linked to elevated liver cancer mortality in parts of Europe (35). In one of the only studies to directly evaluate the association of cyanotoxins as measured in human samples with liver cancer risk, a hospital-based case–control study in China observed serum microcystin to be an independent risk factor for HCC (36).

We can only speculate on the source of oral Cyanobacteria in our study population comprised of residents of three Northeast U.S. coastal states. Human exposure to Cyanobacteria toxins primarily occurs through oral ingestion of contaminated drinking water and, to a lesser extent, dermal contact and inhalation of aerosolized toxin during recreational and other water activities (26, 37). The possibility of drinking water as a source of Cyanobacteria hepatotoxins in our study population cannot be ruled out. Cyanobacteria and cyanotoxins are not regulated in the United States although a few states have developed guidelines for microcystins in drinking water (37). The World Health Organization (WHO) has recommended an upper limit of 1 ng/mL for microcystins in drinking water designated for human use (38). Cyanotoxins have been detected in U.S. drinking water sources. In a survey of 33 U.S. lake and reservoir water sources, microcystins were detected in the majority of raw water samples, albeit nearly all at levels under the WHO limit. Cyanobacteria toxins can also affect irrigation sources, resulting in contamination of soil and agricultural products (39). Exposure may also occur through contaminated seafood including fish and crustaceans (28, 37).

The ability of oral bacteria and their metabolites to access the liver is consistent with evidence that oral bacteria can be translocated to the gut and other distal organs; moreover, this process may be exacerbated in individuals with advanced liver disease as well as those with periodontal disease (4, 7, 8). The premise that oral Cyanobacteria can lead to exposure of the liver to cyanotoxins is supported by our recent work showing that oral exposure to environmental sources of Cyanobacteria can result in both local and systemic levels of cyanotoxins (40). In Guam, a U.S. Pacific Island territory, Cyanobacteria was found to be the predominant bacteria in locally-grown Areca catechu nuts and Piper betle leaves (40), plants which are used in betel nut chewing practiced throughout parts of Asia and the Pacific (41, 42). Oral Cyanobacteria relative abundance was 90-fold higher in current chewers compared with former/never chewers (40). Importantly, we also detected microcystins/nodularins and cylindrospermopsin in both saliva and plant samples. In an independent sample, we detected these hepatotoxins at varying levels in saliva, sera, and urine of both chewers and nonchewers suggesting more widespread environmental exposure (43).

Our study findings provide some evidence of indirect effects of microcystin on HCC development through the dysregulation of lipid metabolism. Aberrant lipid metabolism, including alterations in fatty acid synthesis, has been linked to the pathogenesis of HCC (44). A number of animal studies have demonstrated that oral ingestion of microcystins, including chronic low-dose exposure, in the context of preexisting NAFLD, can exacerbate progression to the more severe form, nonalcoholic steatohepatitis (45–48). We observed the bacterial gene, SCD1, a key enzyme in fatty acid metabolism, to be highly correlated with Cyanobacteria. In animal models, microcystin exposure influences the expression of SCD1 and other lipogenic genes leading to hepatic lipid accumulation (49). SCD1 is overexpressed in a number of cancers including HCC (50). Nonetheless, SCD1 in oral bacteria did not significantly differ between cases and controls, underscoring the complexity of the interplay of bacterial and host genes.

Our findings are bolstered by the population-based design of the overall study, which covered three U.S. metropolitan areas. Nonetheless, our purposeful sample selection to achieve comparable numbers by race likely resulted in a study subset not entirely representative of the larger study population including greater proportions of non–U.S.-born individuals and those with a history of HBV— possibly reflecting birthplace in regions of the world with endemic HBV. Our study findings are also limited by potential reverse causality given the case–control study design with oral samples collected after cancer diagnosis. Presumably, exposure to tumor-promoting effects of cyanotoxins would have occurred in the years prior to liver cancer diagnosis. Ideally, our findings should be confirmed in larger prospective studies measuring local and systemic levels of cyanotoxins in prediagnostic biospecimens. No information was available on periodontal history, residential history, sources of drinking water, produce and seafood, or exposure to bodies of water. We were unable to definitively ascertain the oral Cyanobacteria species given the large number of candidate taxa.

Our study provides the first evidence that oral Cyanobacteria may be an independent risk factor for HCC, possibly through the direct tumor-promoting effects of microcystins and other hepatotoxins and their dysregulating influence on lipid metabolism. Our findings have important implications for current climate-related increases in Cyanobacteria overgrowth and its potential impact on liver cancer risk in the United States and worldwide. Future studies are necessary to further examine the causal relationship between oral Cyanobacteria and HCC risk.

B.Y. Hernandez reports grants from NCI during the conduct of the study. J.K. Lim reports research contracts to Yale University from Allergan, Eiger, Genfit, Gilead, Intercept, Pfizer, and Viking. L.L. Wong reports other support from Eisai outside the submitted work. No disclosures were reported by the other authors.

B.Y. Hernandez: Conceptualization, resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. X. Zhu: Data curation, formal analysis, validation, writing–review and editing. H.A. Risch: Resources, funding acquisition, methodology, writing–review and editing. L. Lu: Writing–review and editing. X. Ma: Writing–review and editing. M.L. Irwin: Writing–review and editing. J.K. Lim: Writing–review and editing. T.H. Taddei: Writing–review and editing. K.S. Pawlish: Writing–original draft. A.M. Stroup: Writing–review and editing. R. Brown: Writing–review and editing. Z. Wang: Data curation, writing–review and editing. L.L. Wong: Writing–review and editing. H. Yu: Conceptualization, funding acquisition, methodology, writing–review and editing.

H. Yu received grant R01CA138698. H. Yu, L. Mishra, Li, and K. Shetty received grant U01CA230690. A.M. Stroup received contract 75N91021D00009. The New Jersey Department of Health received cooperative agreement NU5U58DP006279-02-00. The authors acknowledge the Pacific Island Partnership for Cancer Health Equity (PIPCHE), which provided support for prior research which served as the basis for the methods and resources used in the current study. PIPCHE is supported by the NCI of the NIH under award numbers U54CA143727 and U54CA143728.

This work was supported by funding from the NCI (Grant Nos. R01CA138698 and U01CA230690); institutional funding provided to the University of Hawaii Cancer Center; the New Jersey State Cancer Registry supported by the National Program of Cancer Registries of the Centers for Disease Control and Prevention under cooperative agreement NU5U58DP006279-02-00 awarded to the New Jersey Department of Health; the Surveillance, Epidemiology, and End Results program of the NCI under contract 75N91021D00009 awarded to the Rutgers Cancer Institute of New Jersey, and the State of New Jersey.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Howlader
N
,
Noone
AM
,
Krapcho
M
,
Miller
D
,
Brest
A
,
Yu
M
, et al
editors
.
SEER cancer statistics review, 1975–2017
.
Bethesda, MD:
National Cancer Institute
;
based on November 2019 SEER data submission, posted to the SEER web site
,
April 2020. Available from:
https://seer.cancer.gov/csr/1975_2017/.
2.
Njei
B
,
Rotman
Y
,
Ditah
I
,
Lim
JK
. 
Emerging trends in hepatocellular carcinoma incidence and mortality
.
Hepatology
2015
;
61
:
191
9
.
3.
McGlynn
KA
,
Petrick
JL
,
El-Serag
HB
. 
Epidemiology of hepatocellular carcinoma
.
Hepatology
2021
;
73
Suppl 1:4–13
.
4.
Qin
N
,
Yang
F
,
Li
A
,
Prifti
E
,
Chen
Y
,
Shao
L
, et al
Alterations of the human gut microbiome in liver cirrhosis
.
Nature
2014
;
513
:
59
64
.
5.
Boursier
J
,
Diehl
AM
. 
Implication of gut microbiota in nonalcoholic fatty liver disease
.
PLoS Pathog
2015
;
11
:
e1004559
.
6.
Minemura
M
,
Shimizu
Y
. 
Gut microbiota and liver diseases
.
World J Gastroenterol
2015
;
21
:
1691
702
.
7.
Bajaj
JS
,
Betrapally
NS
,
Hylemon
PB
,
Heuman
DM
,
Daita
K
,
White
MB
, et al
Salivary microbiota reflects changes in gut microbiota in cirrhosis with hepatic encephalopathy
.
Hepatology
2015
;
62
:
1260
71
.
8.
Li
X
,
Kolltveit
KM
,
Tronstad
L
,
Olsen
I
. 
Systemic diseases caused by oral infection
.
Clin Microbiol Rev
2000
;
13
:
547
58
.
9.
Fan
X
,
Alekseyenko
AV
,
Wu
J
,
Peters
BA
,
Jacobs
EJ
,
Gapstur
SM
, et al
Human oral microbiome and prospective risk for pancreatic cancer: a population-based nested case-control study
.
Gut
2018
;
67
:
120
7
.
10.
Hayes
RB
,
Ahn
J
,
Fan
X
,
Peters
BA
,
Ma
Y
,
Yang
L
, et al
Association of oral microbiome with risk for incident head and neck squamous cell cancer
.
JAMA Oncol
2018
;
4
:
358
65
.
11.
Peters
BA
,
Wu
J
,
Pei
Z
,
Yang
L
,
Purdue
MP
,
Freedman
ND
, et al
Oral microbiome composition reflects prospective risk for esophageal cancers
.
Cancer Res
2017
;
77
:
6777
87
.
12.
Yang
B
,
Petrick
JL
,
Abnet
CC
,
Graubard
BI
,
Murphy
G
,
Weinstein
SJ
, et al
Tooth loss and liver cancer incidence in a Finnish cohort
.
Cancer Causes Control
2017
;
28
:
899
904
.
13.
Thistle
JE
,
Yang
B
,
Petrick
JL
,
Fan
JH
,
Qiao
YL
,
Abnet
CC
, et al
Association of tooth loss with liver cancer incidence and chronic liver disease mortality in a rural Chinese population
.
PLoS One
2018
;
13
:
e0203926
.
14.
Helenius-Hietala
J
,
Suominen
AL
,
Ruokonen
H
,
Knuuttila
M
,
Puukka
P
,
Jula
A
, et al
Periodontitis is associated with incident chronic liver disease-A population-based cohort study
.
Liver Int
2019
;
39
:
583
91
.
15.
Shen
Y
,
Risch
H
,
Lu
L
,
Ma
X
,
Irwin
ML
,
Lim
JK
, et al
Risk factors for hepatocellular carcinoma (HCC) in the northeast of the United States: results of a case-control study
.
Cancer Causes Control
2020
;
31
:
321
32
.
16.
Baker
GC
,
Smith
JJ
,
Cowan
DA
. 
Review and re-analysis of domain-specific 16S primers
.
J Microbiol Methods
2003
;
55
:
541
55
.
17.
Bolyen
E
,
Rideout
JR
,
Dillon
MR
,
Bokulich
NA
,
Abnet
CC
,
Al-Ghalith
GA
, et al
Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2
.
Nat Biotechnol
2019
;
37
:
852
7
.
18.
Callahan
BJ
,
McMurdie
PJ
,
Rosen
MJ
,
Han
AW
,
Johnson
AJ
,
Holmes
SP
. 
DADA2: High-resolution sample inference from Illumina amplicon data
.
Nat Methods
2016
;
13
:
581
3
.
19.
Federhen
S
. 
The NCBI taxonomy database
.
Nucleic Acids Res
2012
;
40
:
D136
43
.
20.
Heinze
G
,
Schemper
M
. 
A solution to the problem of separation in logistic regression
.
Stat Med
2002
;
21
:
2409
19
.
21.
Douglas
GM
,
Maffei
VJ
,
Zaneveld
JR
,
Yurgel
SN
,
Brown
JR
,
Taylor
CM
, et al
PICRUSt2 for prediction of metagenome functions
.
Nat Biotechnol
2020
;
38
:
685
8
.
22.
Petrick
JL
,
Sahasrabuddhe
VV
,
Chan
AT
,
Alavanja
MC
,
Beane-Freeman
LE
,
Buring
JE
, et al
NSAID use and risk of hepatocellular carcinoma and intrahepatic cholangiocarcinoma: the liver cancer pooling project
.
Cancer Prev Res
2015
;
8
:
1156
62
.
23.
Sahasrabuddhe
VV
,
Gunja
MZ
,
Graubard
BI
,
Trabert
B
,
Schwartz
LM
,
Park
Y
, et al
Nonsteroidal anti-inflammatory drug use, chronic liver disease, and hepatocellular carcinoma
.
J Natl Cancer Inst
2012
;
104
:
1808
14
.
24.
Brar
G
,
Greten
TF
,
Graubard
BI
,
McNeel
TS
,
Petrick
JL
,
McGlynn
KA
, et al
Hepatocellular carcinoma survival by etiology: a SEER-medicare database analysis
.
Hepatol Commun
2020
;
4
:
1541
51
.
25.
Buratti
FM
,
Manganelli
M
,
Vichi
S
,
Stefanelli
M
,
Scardala
S
,
Testai
E
, et al
Cyanotoxins: producing organisms, occurrence, toxicity, mechanism of action and human health toxicological risk evaluation
.
Arch Toxicol
2017
;
91
:
1049
130
.
26.
International Agency for Research on Cancer Working Group
.
Ingested nitrate and nitrite, and cyanobacterial peptide toxins. IARC monographs on the evaluation of carcinogenic risks to humans
.
Volume 94
.
Lyon France
:
International Agency for Research on Cancer
; 
2010
.
27.
Cires
S
,
Alvarez-Roa
C
,
Wood
SA
,
Puddick
J
,
Loza
V
,
Heimann
K
. 
First report of microcystin-producing Fischerella sp. (Stigonematales, Cyanobacteria) in tropical Australia
.
Toxicon
2014
;
88
:
62
6
.
28.
Jakubowska
N
,
Szelag-Wasielewska
E
. 
Toxic picoplanktonic cyanobacteria–review
.
Mar Drugs
2015
;
13
:
1497
518
.
29.
Fujiki
H
,
Suganuma
M
. 
Tumor promoters–microcystin-LR, nodularin and TNF-alpha and human cancer development
.
Anticancer Agents Med Chem
2011
;
11
:
4
18
.
30.
Kloer
DP
,
Ruch
S
,
Al-Babili
S
,
Beyer
P
,
Schulz
GE
. 
The structure of a retinal-forming carotenoid oxygenase
.
Science
2005
;
308
:
267
9
.
31.
Ahrazem
O
,
Gomez-Gomez
L
,
Rodrigo
MJ
,
Avalos
J
,
Limon
MC
. 
Carotenoid cleavage oxygenases from microbes and photosynthetic organisms: features and functions
.
Int J Mol Sci
2016
;
17
.
32.
Yu
SZ
. 
Primary prevention of hepatocellular carcinoma
.
J Gastroenterol Hepatol
1995
;
10
:
674
82
.
33.
Yu
S
,
Zhao
N
,
Zi
X
. 
[The relationship between cyanotoxin (microcystin, MC) in pond-ditch water and primary liver cancer in China]
.
Zhonghua Zhong Liu Za Zhi
2001
;
23
:
96
9
.
34.
Zhang
F
,
Lee
J
,
Liang
S
,
Shum
CK
. 
Cyanobacteria blooms and non-alcoholic liver disease: evidence from a county level ecological study in the United States
.
Environ Health
2015
;
14
:
41
.
35.
Svircev
Z
,
Drobac
D
,
Tokodi
N
,
Vidovic
M
,
Simeunovic
J
,
Miladinov-Mikov
M
, et al
Epidemiology of primary liver cancer in Serbia and possible connection with cyanobacterial blooms
.
J Environ Sci Health C Environ Carcinog Ecotoxicol Rev
2013
;
31
:
181
200
.
36.
Zheng
C
,
Zeng
H
,
Lin
H
,
Wang
J
,
Feng
X
,
Qiu
Z
, et al
Serum microcystin levels positively linked with risk of hepatocellular carcinoma: a case-control study in southwest China
.
Hepatology
2017
;
66
:
1519
28
.
37.
U.S. Environmental Protection Agency, Office of Water, Health and Ecological Criteria Division
. 
Drinking water health advisory for the cyanobacterial microcystin toxins
.
Washington, DC:
United States Environmental Protection Agency
; 
2015
.
38.
World Health Organization
.
Guidelines for drinking-water quality. 4th ed, incorporating the first addendum.
Geneva, Switzerland
:
World Health Organization
; 
2017
. p.
344
6
.
39.
Drobac
D
,
Tokodi
N
,
Kiprovski
B
,
Malencic
D
,
Vazic
T
,
Nybom
S
, et al
Microcystin accumulation and potential effects on antioxidant capacity of leaves and fruits of Capsicum annuum
.
J Toxicol Environ Health A
2017
;
80
:
145
54
.
40.
Hernandez
BY
,
Zhu
X
,
Sotto
P
,
Paulino
Y
. 
Oral exposure to environmental cyanobacteria toxins: implications for cancer risk
.
Environ Int
2021
;
148
:
106381
.
41.
Paulino
YC
,
Novotny
R
,
Miller
MJ
,
Murphy
SP
. 
Areca (Betel) nut chewing practices in Micronesian populations
.
Hawaii J Public Health
2011
;
3
:
19
29
.
42.
International Agency for Research on Cancer Working Group
.
Personal habits and indoor combustions. Volume 100 E, A review of human carcinogens. IARC monographs on the evaluation of carcinogenic risks to humans
.
Lyon, France
:
International Agency for Research on Cancer
; 
2009
.
43.
Hernandez
BY
,
Zhu
X
,
Sotto
P
,
Paulino
Y
. 
Betel nut chewing, oral Cyanobacteria, and exposure to cyanotoxins [abstract]
.
In:
Proceedings of the AACR Special Conference on the Microbiome, Viruses, and Cancer
; 
2020
Feb
21
24
;
Orlando, FL. Philadelphia (PA)
:
AACR
; 
2020
.
Abstract nr A11
.
44.
Sangineto
M
,
Villani
R
,
Cavallone
F
,
Romano
A
,
Loizzi
D
,
Serviddio
G
. 
Lipid metabolism in development and progression of hepatocellular carcinoma
.
Cancers (Basel)
2020
;
12
.
45.
He
J
,
Li
G
,
Chen
J
,
Lin
J
,
Zeng
C
,
Chen
J
, et al
Prolonged exposure to low-dose microcystin induces nonalcoholic steatohepatitis in mice: a systems toxicology study
.
Arch Toxicol
2017
;
91
:
465
80
.
46.
Lad
A
,
Su
RC
,
Breidenbach
JD
,
Stemmer
PM
,
Carruthers
NJ
,
Sanchez
NK
, et al
Chronic low dose oral exposure to microcystin-lr exacerbates hepatic injury in a murine model of non-alcoholic fatty liver disease
.
Toxins (Basel)
2019
;
11
:
486
.
47.
Clarke
JD
,
Dzierlenga
A
,
Arman
T
,
Toth
E
,
Li
H
,
Lynch
KD
, et al
Nonalcoholic fatty liver disease alters microcystin-LR toxicokinetics and acute toxicity
.
Toxicon
2019
;
162
:
1
8
.
48.
Albadrani
M
,
Seth
RK
,
Sarkar
S
,
Kimono
D
,
Mondal
A
,
Bose
D
, et al
Exogenous PP2A inhibitor exacerbates the progression of nonalcoholic fatty liver disease via NOX2-dependent activation of miR21
.
Am J Physiol Gastrointest Liver Physiol
2019
;
317
:
G408
28
.
49.
Lei
H
,
Song
Y
,
Dong
M
,
Chen
G
,
Cao
Z
,
Wu
F
, et al
Metabolomics safety assessments of microcystin exposure via drinking water in rats
.
Ecotoxicol Environ Saf
2021
;
212
:
111989
.
50.
Pope
ED
 III
,
Kimbrough
EO
,
Vemireddy
LP
,
Surapaneni
PK
,
Copland
JA
 III
,
Mody
K
. 
Aberrant lipid metabolism as a therapeutic target in liver cancer
.
Expert Opin Ther Targets
2019
;
23
:
473
83
.