The Expression Profiles and Deregulation of UDP-Glycosyltransferase (UGT) Genes in Human Cancers and Their Association with Clinical Outcomes

Simple Summary The human UDP-glycosyltransferase (UGT) superfamily plays a critical role in the metabolism of numerous endogenous and exogenous small lipophilic compounds, including carcinogens, drugs, and bioactive molecules with pro- or anti-cancer activity. Previous studies have documented the expression of UGT genes in several cancers derived from drug-metabolizing organs (e.g., liver, colon, kidney). The present study represents the first to comprehensively assess the expression profiles of UGT genes and their impact on patient survival in nearly 30 different cancers primarily derived from non-drug-metabolizing organs. Briefly, our comprehensive analysis of the transcriptomic (RNAseq) and clinical datasets of 9514 patients from 33 different cancers shows the widespread expression of UGT genes, indicative of active drug metabolism within the tumor through the UGT conjugation pathway. We further identified the UGT genes whose intratumoral expression was associated with patient survival, highlighting the potential of UGT genes as prognostic biomarkers and therapeutic targets in various cancers. Abstract The human UDP-glycosyltransferase (UGTs) superfamily has 22 functional enzymes that play a critical role in the metabolism of small lipophilic compounds, including carcinogens, drugs, steroids, lipids, fatty acids, and bile acids. The expression profiles of UGT genes in human cancers and their impact on cancer patient survival remains to be systematically investigated. In the present study, a comprehensive analysis of the RNAseq and clinical datasets of 9514 patients from 33 different TCGA (the Genome Cancer Atlas) cancers demonstrated cancer-specific UGT expression profiles with high interindividual variability among and within individual cancers. Notably, cancers derived from drug metabolizing tissues (liver, kidney, gut, pancreas) expressed the largest number of UGT genes (COAD, KIRC, KIRP, LIHC, PAAD); six UGT genes (1A6, 1A9, 1A10, 2A3, 2B7, UGT8) showed high expression in five or more different cancers. Kaplan–Meier plots and logrank tests revealed that six UGT genes were significantly associated with increased overall survival (OS) rates [UGT1A1 (LUSC), UGT1A6 (ACC), UGT1A7 (ACC), UGT2A3 (KIRC), UGT2B15 (BLCA, SKCM)] or decreased OS rates [UGT2B15 (LGG), UGT8 (UVM)] in specific cancers. Finally, differential expression analysis of 611 patients from 12 TCGA cancers identified 16 UGT genes (1A1, 1A3, 1A6, 1A7, 1A8, 1A9, 1A10, 2A1, 2A3, 2B4, 2B7, 2B11, 2B15, 3A1, 3A2, UGT8) that were up/downregulated in at least one cancer relative to normal tissues. In conclusion, our data show widespread expression of UGT genes in cancers, highlighting the capacity for intratumoural drug metabolism through the UGT conjugation pathway. The data also suggests the potentials for specific UGT genes to serve as prognostic biomarkers or therapeutic targets in cancers.

The expression profiles of UGT genes in human tissues have been investigated at RNA and protein levels using multiple approaches. Due to the lack of specific antibodies for most UGT enzymes, there is the only analysis of protein expression in human tissues for a subset of UGT genes (e.g., 1A1, 1A6, 2B7, 2B15, 2B17, 2B28) using custom-developed antibodies via immunohistochemistry, immunoblotting, or tissue microarrays [6][7][8][9][10][11][12]. There are commercial UGT antibodies from several companies (e.g, Sigma, Abcam), but their specificities have not yet been vigorously validated and have been reported to recognize several highly homologous UGT enzymes [11,13]. To overcome this limitation, recent studies have used stable isotope-labeled peptide-based liquid chromatography-tandem mass spectrometry (LC-MS/MS) [14][15][16][17][18][19][20]. However, there is more extensive data on UGT mRNA expression in human tissues and cell lines that have been generated using isoform-specific reverse transcriptase quantitative real-time polymerase chain reaction (RT-qPCR) [21][22][23][24][25][26][27] and RNA sequencing technology (RNAseq) [28][29][30]. RNAseq studies are particularly powerful as they provide accurate isoform-specific and high-throughput quantification of UGT transcripts. Collectively, these studies have demonstrated the widespread expression of UGT genes in normal human tissues. Tissues involved in detoxification, particularly liver, and to a lesser degree, kidney and gut, express the widest range of UGT isoforms and typically the highest transcript levels. However, a subset of UGTs shows high expression in tissues that are not generally associated with drug metabolism and detoxification. These UGTs may be important for local control of endogenous metabolites (such as steroids) and could, therefore, mediate intra-tissular drug metabolism.
Several studies have assessed the expression profiles of UGT genes and their deregulation in human cancers that are derived from drug-metabolizing tissues/organs, including liver cancer [31,32], kidney cancer [20], colon cancer [7,33,34], and gastric cancers [35,36]. However, little is known about the expression profiles of UGT genes and their deregulation in human cancers that are derived from non-drug-metabolizing tissues.
As recently reviewed [37], case-control studies have shown that a large number of genetic polymorphisms of UGT genes are associated with cancer development and progression. This is believed to be related to the critical roles of UGT enzymes in the systemic metabolism and clearance of carcinogens, cancer-modulating molecules, and anticancer drugs. The expression and activity of UGT genes within the tumor may also impact cancer development and progression through intratumoral inactivation of carcinogens and anticancer drugs. For example, high UGT2B17 expression has recently been shown to be associated with poor prognosis in chronic lymphocytic leukaemia (CLL), partly due to enhanced local inactivation of anti-leukaemic drugs (e.g., fludarabine) within CLL cells [38,39]. Therefore, it is necessary to systematically assess whether the intratumoral expression of UGT genes could be associated with clinical outcomes in different human cancers.
The Cancer Genome Atlas (TCGA) project analysed over 20,000 primary cancer patients from 33 different cancer types and provides freely-accessible databases for genomewide molecular profiles (e.g., RNAseq datasets) as well as clinicopathological data (e.g., survival times) for cancer patients (https://gdc.cancer.gov, accessed on 20 June 2021) [40]. Using the RNAseq and clinical datasets from the TCGA project, we recently reported the expression profiles of core ADME genes and their association with patient survival in 21 different TCGA cancer types [41]. Through analyses of the RNAseq and clinical datasets from the TCGA project, the present study defines the expression profiles of all UGT genes in 33 different TCGA cancer types and reports a subset of UGT genes that are significantly associated with survival rates in specific cancers. We further identify the UGT genes that are deregulated in 12 cancer types.

Assessment of the Expression Profiles of UGT Genes in Human Cancers
The expression profiles of UGT genes in cancers were assessed using RNAseq data of 9514 tumor samples from 33 different TCGA cancer types [42] (Table 1). RSEM is one of the most frequently used methods for quantifying transcript abundances from RNASeq data [43]. The mRNA levels (RNASeqV2) of UGT genes in 5  There are 22 functional UGT genes in the human genome [44]. The RSEM values of 2 UGT genes (UGT2B17, UGT2A2) were not available from the Firehose and PanCanAtlas databases, and thus, these 2 UGT genes were excluded from our analysis in this study. The normalized RSEM values of the assessed 20 UGT genes in 9514 tumor samples are provided in Table S1. The expression levels (e.g., medians, means) of each UGT gene in each of the 33 different cancer types were calculated using GraphPad Prism (version 8.1.2, GraphPad Software Inc, San Diego, CA, USA) and are presented in Table S2. For simplicity, we defined high, low, and no expression of a UGT gene by a median normalized RSEM level of >32, between 1-32, and <1, respectively. The variable expression of UGT genes within and between cancers are presented using the box-and-whisker plots ( Figures S1-S4).

Assessment of the Deregulation of UGT Genes in Human Cancers
Of the 33 TCGA cancer datasets, 12 had at least 20 patients with RNAseq data (RNAseqV2) available for both tumor tissues and matched adjacent non-cancerous tissues ( Table 1). Using RNAseq data from these patients, we assessed whether UGT genes were differentially expressed in cancer tissues as compared to matched adjacent non-cancerous tissues. Briefly, the RNAseq data (RNASeqV2) was downloaded as high-throughput sequencing (HT-seq) counts from the TCGA legacy database (the human genome hg19 reference assembly) using the "TCGAbiolinks" R package (https://bioconductor.org/pac kages/release/bioc/html/TCGAbiolinks.html, accessed on 10 May 2021). To remove the distributional differences between within-lane and between-lane read counts, the EDAseq package in R (https://bioconductor.org/packages/release/bioc/html/EDASeq.html, accessed on 10 May 2021) was used, and genes were filtered with a quantile threshold of 0.25 [45]. Table S3 provides HT-seq counts for UGT genes in matched tumor and adjacent non-cancerous tissues for 611 patients from the assessed 12 TCGA cancer types.
The differential gene expression analysis between matched cancerous and adjacent non-cancerous tissues were assessed using the DESeq2 program as recently reported [31]. DESeq2 is one of the most frequently used programs for differential gene expression analysis [46]. Briefly, DESeq2 uses a Wald test for statistically significant testing. The Wald test p-values from the genes that pass the independent filtering step are adjusted for multiple testings using the Benjamini-Hochberg test. An adjusted p-value of <0.01 was considered statistically significant. DESeq2 reports a log2FoldChange for the differential expression of each gene and the associated Standard Error for the log2FoldChange estimate (lfcSE). Differentially expressed genes were defined by a log2FoldChange of >2 (equivalent to an upregulation of >4-fold) or <−2 (equivalent to downregulation of >4-fold). Table S4 shows the results from DESeq2 analysis pertaining to UGT genes.

Assessment of Associations between the Intratumoral Expression Levels of UGT Genes and Overall Survival of Cancer Patients Using Kaplan-Meier Survival Analysis
The TCGA Pan-Cancer Clinical Data Resources (TCGA-CDR) demonstrated the values of the clinical survival data of 33 TCGA cancer types for reliable survival analyses [42]. The TCGA-CDR collected overall survival data and other clinicopathological parameters for 11,160 patients from 33 different TCGA cancer types. We downloaded these survival data (i.e., TCGA-CDR-Supplemental Table S1) from the PanCanAtlas database (https:// gdc.cancer.gov/about-data/publications/pancanatlas, accessed on 4 June 2021). Of these patients, only 9514 patients with RNAseq data (normalized RSEM values as described above) available for tumour samples were included in our survival analyses (Table 1 and  Table S1).
Overall survival (OS) time was defined as the time from the day at diagnosis to the date of death (dead patients) or the date of the last follow-up (censored patients). The Kaplan-Meier survival analysis is a common approach for clinical survival analysis [54]. Using GraphPad Prism (version 8.1.2), we performed Kaplan-Meier plots and logrank tests to assess the potential associations between intratumoral mRNA levels (normalized RSEM values) of UGT genes and OS rates for each of the 33 TCGA cancer types. For UGT genes that were expressed in over 50% of the tumor samples, we separated the patients by gene expression into a high-expression group (upper 50 percentile) and low/no-expression group (lower 50 percentile) and performed logrank tests. For UGT genes that were expressed in 10-50% of the tumor samples, we separated the patients by gene expression into expression group and no-expression group and performed logrank tests. UGT genes that were expressed in less than 10% of the tumor samples were excluded from survival analysis. As a varying number of UGT genes were expressed in different cancer types (Table S1), the number of independent logrank tests performed varied among different cancers, ranging from 20 tests in LIHC to 3 tests in UVM. To control false-positive discovery rates, we adjusted the logrank p-values for each cancer type using Bonferroni correction, the most stringent test for multiple testing correction as recently reported [41]. A Bonferroni-corrected cutoff logrank p-value of <0.05 was considered statistically significant. Table S2 lists both logrank and Bonferroni-corrected p-values and the associated hazard ratios (HR) and 95% confidence intervals (CI) for all independent logrank tests that assessed the potential associations between intratumoral expression levels of UGT genes and overall survival rates in 33 different TCGA cancer types.

Inter-Individual Variation in UGT Expression within Human Cancers
The expression of UGT genes is highly dynamic and influenced by multiple extrinsic stimuli and signaling pathways; as such, many UGTs show exceptionally wide interindividual variation in normal tissues, particularly in the liver [3,44]. Thus, it was of interest to determine the inter-individual variability of UGT expression within cancer cohorts. The variable expression of UGT genes within 33 different cancer types are presented using the box-and-whisker plots that show the distribution of the expression levels (minimum, first quartile, median, third quartile, and maximum) in each of these cancers (Figure 2, Figures S3 and S4). Overall, all UGT genes showed inter-individual variable expression in each of the 33 cancer types. For example, UGT1A1 was detected in 362 of the 365 LIHC tumors with RSEM ranging from 1.6 to 87,075. A total of 82 tumours had an expression level of >10,000, whereas 88 tumours had an expression level of <1000 (Table S1). This high interindividual variable expression for UGT1A1 and other UGT genes may result in intratumoral variabilities in the conjugation of UGT substrates and hence impact cancer progression and patient survival as assessed in Section 3.4.

Associations between Intratumoral UGT Expression Levels and Overall Survival of Cancer Patients
Using Kaplan-Meier plots and logrank tests followed by Bonferroni multiple corrections, we assessed the potential association of the intratumoral expression levels of UGT genes with overall survival (OS) rates in each of the 33 TCGA cancer types (Table S2). Overall, our results showed that the expression levels of 6 UGT genes (1A1, 1A6, 1A7, 2A3, 2B15, UGT8) were significantly associated with OS rates in at least one cancer type (Figure 4). Of these genes, four showed significant association with increased OS rates (favourable survival), namely (1) UGT1A1 in LUSC ( Figure 4A), (2) UGT1A6 in ACC ( Figure 4B), (3) UGT1A7 in ACC ( Figure 4C), and (4) UGT2A3 in KIRC ( Figure 4D). UGT8 was significantly associated with decreased OS rates (unfavourable survival) in UVM ( Figure 4H). UGT2B15 showed significant association with increased OS rates in BLCA and SKCM) ( Figure 4E,G) but decreased OS rates in LGG ( Figure 4F).   genes with overall survival rates in TCGA cancer types as indicated. For Kaplan-Meier survival analysis, the patients were separated into high-expression groups and low/no-expression group by gene expression levels as described in the Materials and Methods Section. The number of patients in each group for each Kaplan-Meier plot/logrank test was given in bracket following the p-value. Hazard Ratios (HR) and 95% confidence intervals (95% CI) are also shown. Logrank p-values were adjusted using Bonferroni correction as described in the Materials and Method Section. A Bonferroni-corrected cutoff logrank p-value of <0.05 indicates statistical significance.

Discussion
We recently analysed the RNAseq data from the Human Protein Atlas [55] (https: //www.proteinatlas.org, accessed on 10 January 2019) and the Genotype-Tissue Expression (GTEx) project [56] and reported the expression profiles of all 22 UGT genes in a panel of 43 normal human tissues [44]. This demonstrated that many UGT genes that have been traditionally associated with liver or other detoxifying tissues, in fact, have complex body-wide expression patterns. As an example, seven UGTs (1A1, 1A6, 1A10, 2B7, 2B15, 2B17, and UGT8) showed expression in more than 20 different tissues [44]. This widespread expression supports an important role for UGTs in intra-tissular drug/xenobiotic exposure, as well as controlling endogenous lipophilic metabolites and signaling molecules. These roles may also contribute to the systemic clearance of UGT substrates. UGT expression within tumors derived from various tissues could similarly affect intratumoral drug exposure, as well as levels of endogenous growth-regulatory chemicals. In the present study, we analysed the RNAseq data of 9514 cancer patients from the TCGA project, allowing us to characterize the expression profiles of 20 UGT genes in 33 different cancer types.
Overall, UGT genes exhibited cancer-specific expression profiles and high interindividual variabilities within cancer cohorts. Four (1A6, 1A10, 2B7, UGT8) of the aforementioned seven UGT genes that are widely expressed in normal tissues were also highly expressed in more than five different cancer types ( Figure 2). Notably, UGT8 and UGT1A6 showed high expression in 18 and 13 different cancer types, respectively. A total of 25 cancer types (BLCA, CESC, CHOL, COAD, DLBC, ESCA, GBM, HNSC, KICH, KIRC, KIRP, LGG, LIHC, LUAD, LUSC, OV, PAAD, READ, SKCM, STAD, TGCT, STAD, THCA, THYM, UCS) highly expressed at least 1 UGT gene. Previous studies have shown high expression of UGT genes in drug-metabolizing tissues (e.g., liver, kidney, colon) [20,22,23,27,30,33]. Consistently, we found abundant expression of similar UGT genes in cancers that are derived from drug-metabolizing tissues (LIHC, KIRC/KIRP, COAD) ( Figure 1). For example, 12 UGT genes were abundantly expressed in liver cancer (LIHC) in accordance with their high expression in the normal liver. The present study represents the first to comprehensively assess the expression profiles of UGT genes in nearly 30 different cancers derived from non-drug-metabolizing tissues. Collectively, our results demonstrate widespread expression profiles of UGT genes in human cancers, implying active metabolism of UGT substrates within the tumors that are derived from not only drug-metabolizing tissues but also non-drug-metabolizing tissues.
Given that intratumoural UGT expression could enhance the inactivation of anticancer drugs and cancer-growth modulating molecules (e.g., steroids, lipids, fatty acids), it may be predicted to impact cancer progression and patient survival. Indeed, clinical and preclinical studies have shown that overexpression of specific UGTs can promote resistance to anticancer drugs that they metabolize [57][58][59][60][61][62][63]. In the present study, Kaplan-Meier survival analysis of 9145 patients from 33 different TCGA cancer types identified significant associations of intratumoral expression levels of UGT genes with overall survival rates in seven different cancer types, namely LUSC (1A1), ACC (1A6, 1A7), BLCA (2B15), KIRC (2A3), LGG (2B15), SKCM (2B15), and UVM (UGT8) (Figure 4). Of note, the associations of UGT1A1, UGT2B15, and UGT2A3 with favourable OS rates in LUSC [41], BLCA [41], and KIRC [55], respectively, were recently reported. As discussed in detail below, the inactivation of anticancer drugs within the tumor through the UGT conjugation pathway can reduce therapeutic efficacy and patient survival. Analysis of drug regimens received by TCGA patients could test this hypothesis. Therefore, we obtained drug regimen data for these seven TCGA cancer types from NCI Genomic Data Commons (https://gdc.canc er.gov/, accessed on 10 June 2021) (Table S5). Unfortunately, because drug regimens were only available for a small proportion of patients for these seven cancer types, we were not able to assess whether the observed association of UGTs with OS rates could be related to intratumoral inactivation of anticancer drugs.
Interestingly, our data show that high UGT levels were frequently correlated with increased OS rates (Figure 4). These findings are consistent with recent reports for other drug-metabolizing enzymes (e.g., CYPs, NAT1) [41], and support a hypothesis that intratumoral expression of UGTs and other drug-metabolizing enzymes can impact cancer patient survival through not only drug metabolism but also metabolism of numerous endogenous bioactive molecules (e.g., steroid hormones, amino acids, fatty acids, bile acids) that can modulate cancer growth as briefly discussed below.
Androgens are known to promote bladder cancer (BLCA) development and progression [64][65][66]. The two most potent natural androgens (testosterone and dihydrotestosterone) are primarily inactivated by UGT2B15 and UGT2B17 [67]. Our observed association of high UGT2B15 levels with increased OS rates in BLCA ( Figure 4) might be attributable to its intratumoral glucuronidation of androgens within the tumour as recently suggested [41]. Unfortunately, UGT2B17 data were not available for analysis within the TCGA dataset. We also showed an association of UGT2B15 with increased OS rates in SKCM or decreased OS rates in LGG ( Figure 4); however, the underlying mechanisms remains unknown.
Adrenocortical carcinoma (ACC) is a very rare malignancy that originates in the cortex of the adrenal gland, and patients with ACC often have poor clinical outcomes [68]. Surgery remains the only curative treatment for patients with ACC, and mitotane is the most effective drug in adjuvant chemotherapy of ACC or in inoperable ACC [69]. There is no evidence that mitotane and its two activate metabolites [1,1-(o,p'-dichlorodiphenyl)-2,2 dichloroethene (o,p'-DDE), 1,1-(o,p'-dichlorodiphenyl) acetic acid (o,p'-DDA) ] are substrates of any UGT [70]. Some patients with ACC have increased levels of steroid hormones and mineralocorticoids, and they tend to show hypercortisolism and hyperandrogenism [68]. ACC showed moderate expression of UGT1A6 and UGT1A7 but low/no expression of other UGT genes ( Figure S2). Our observed association of high levels of UGT1A6 or UGT1A7 with increased OS rates might be related to their intratumoral inactivation and clearance of endogenous bioactive molecules, such as the aforementioned steroid hormones. This hypothesis remains to be investigated.
We showed a significant association of high UGT8 expression with poor OS rates in Uveal Melanoma (UVM) ( Figure 4H). UVM originates from melanocytes within the uveal tract and is the second most common melanoma subtype after cutaneous melanoma [71]. Primary UVM is treated with surgery and radiation; however, remote metastasis (most often in the liver) occurs in nearly 50% of the patients with a very poor prognosis [72]. A 10-gene signature (SIRT3, HMCES, SLC44A3, TCTN1, STPG1, POMGNT2, RNF208, ANXA2P2, ULBP1, CA12) that predicts prognosis for this disease has been recently reported [73]. Further studies are warranted to determine whether UGT8 may be a useful prognostic biomarker for UVM. The mechanism by which UGT8 could influence UVM is currently speculative. UGT8 galactosidates bile acids [74] and ceramide [75]. The later reaction generates galactosylceramide (GalCer), which can be further converted into sufatide [76]. Ceramides are critical regulators of survival and drug resistance in melanoma, hence UGT8 might control UVM outcomes via modulation of ceramide levels [77]. In support of this notion, UGT8 overexpression was recently shown to promote basal-like breast cancer (BLBC) cell proliferation and invasion through production of Gal-Cer and sufatide [78,79]. Although Cao et al. recently showed an association between high intratumoral UGT8 levels and poor survival in BLBC [79], no association was observed in our analysis of the TCGA BRCA dataset as a whole (1080 patients) or the basal subtype (179 patients) (data not shown).
The findings of the current study could provide impetus for future translational UGT research. The potential for such translation is supported by numerous clinical and preclinical studies. For example, our findings of high interindividual expression variability and deregulation of UGT genes within specific cancer types may be relevant to intratumoral exposure of anticancer drugs that are primarily metabolized through UGT conjugation. In support of this idea, pre-clinical and clinical studies have shown that high intratumoral expression of several UGT genes (e.g., 1A1, 1A6, 2B7, 2B17) contributed to de novo or acquired resistance to various anticancer drugs [39,60,80]. These findings, together with observations in the present study, suggest that assessing intratumoral UGT activity could help to achieve optimal personalized anticancer therapy. Moreover, our observed associations of intratumoral UGT expression with patient survival highlight their potential as prognostic biomarkers. In addition to the data shown here that assessed OS within 33 TCGA cancer types, other studies have reported the prognostic value of various UGTs in specific subsets of cancer patients [6,38,39,75]. For example, previous work shows that UGT2B17 and UGT2B28 are overexpressed in advanced and metastatic prostate cancer and associate with poor outcomes [6,[81][82][83]. Pre-clinical studies in mouse xenograft models show that UGT2B17 overexpression promotes androgen-independent tumor progression via a pathway that may involve tyrosine-protein kinase Src [82]. UGT2B17 is also prognostic in chronic lymphocytic leukaemia (CLL), where high expression is associated with shorter treatment-free and overall survival primarily through its intracellular inactivation of antileukaemic drugs such as fludarabine [39]. In addition, we have shown that UGT2B15 and UGT2B17 are prognostic in specific molecular subtypes of breast cancer as defined by the METABRIC (Molecular Taxonomy of Breast Cancer International Consortium) project [84]. Based on these previous studies and the new data presented herein, we suggest that future work could focus on developing selected UGTs as clinically actionable biomarkers and/or therapeutic targets for new drug discovery.

Conclusions
In conclusion, the present study reported the unique expression profiles of UGT genes in 33 TCGA cancer types and identified the patterns of UGT deregulation in 12 TCGA cancer types. We further identified the UGT genes whose intratumoral expression was significantly associated with overall survival. Collectively, our results provide compelling evidence for the active metabolism of UGT substrates within tumors, and support further interrogation of UGT genes as potential prognostic biomarkers and therapeutic targets. Intratumoral UGT activity can influence cancer progression and patient survival not only through drug metabolism but also through the inactivation of numerous endogenous bioactive molecules that can modulate cancer growth.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/1 0.3390/cancers13174491/s1, Figure S1: The expression profiles of UGT genes in 10 different types of TCGA cancers; Figure S2: The expression profiles of UGT genes in 17 different types of TCGA cancers; Figure S3: The expression profiles of 12 UGT genes in 33 different types of TCGA cancers; Figure S4: The expression profiles of UGT3A genes in 33 different types of TCGA cancers; Figure S5: Expression of UGT genes in matched cancerous (T) and adjacent non-cancerous (N) tissues in 8 different TCGA cancer types; Table S1: The expression levels (normalized RSEM values) of UGT genes and clinical datasets of 9514 patients from 33 different types of TCGA cancers; Table S2: The expression levels (means, medians, standard deviations) of UGT genes and assessment of their associations with overall survival rates in 33 different TCGA cancer types; Table S3: HT-seq counts for UGT genes in matched tumor and adjacent non-cancerous tissues for 611 patients from 12 different TCGA cancer types; Table S4: Differential gene expression analysis of 611 patients from 12 TCGA cancer types using DESeq2 revealed the UGT genes that were deregulated in these cancers; Table S5: Drug regimens of 7 TCGA cancer types.  Data Availability Statement: For analysis of gene expression profiles, RNASeq data (RSEM) were obtained from the TCGA Firehose database (http://gdac.broadinstitute.org) and the PanCanAtlas database (https://gdc.cancer.gov/about-data/publicatio ns/pancanatlase, accessed on 20 June 2021). For gene differential expression analysis, RNAseq data (HT-seq counts) were obtained from the TCGA legacy database using the "TCGAbiolinks" R package (https://bioconductor.org/packages/ release/bioc/html/TCGAbiolinks.html). The survival data of TCGA patients were obtained from the PanCanAtlas database (https://gdc.cancer.gov/about-data/publications/pancanatlas).

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