Page 85 - GPD-2-1

Gene & Protein in Disease A pan-cancer analysis of HMGB1

genes. The P-values and correlation coefficient (R) were However, compared with the control group, HMGB1
calculated and displayed in figures. In addition, TIMER2 expression in kidney chromophobe (KICH), lung
was applied to obtain heatmap data of selected genes, adenocarcinoma (LUAD) (P < 0.001), and prostate
including P-values and partial correlation (cor) in the adenocarcinoma (PRAD) (P < 0.05) were relatively low
purity-adjusted Spearman’s rank correlation test. (Figure 1A).

Jvenn was used for cross-analysis to compare the We used the GTEx dataset to obtain normal tissue data
[16]
HMGB1-binding and interacted genes. We integrated as a control, and further evaluated the HMGB1 expression
these two sets of data for Kyoto Encyclopedia of Genes differences between the tumor and normal tissues. The
and Genomes (KEGG) pathway analysis. The “tidyr” results showed that HMGB1 was highly expressed in CHOL,
and “ggplot2” R packages were applied to show the COAD, lymphoid neoplasm diffuse large B-cell lymphoma
enrichment pathway. Besides, the “clusterProfiler” R (DLBC), GBM, brain lower grade glioma (LGG), READ,
packages were applied to conduct Gene Ontology (GO) pancreatic adenocarcinoma (PAAD), STAD, and thymoma
enrichment analysis. The R language software (R-4.1.0, (THYM) (Figure 1B, P < 0.05).
64-bit; https://www.r-project.org/) was applied to this In addition to transcription, CPTAC was used to
analysis. Two-tailed P < 0.05 was considered statistically evaluate HMGB1 at the protein level. As displayed in
significant . A simple list of methods and corresponding Figure 1C, we found that the total protein expression of
[17]
tools is shown in Table 1.
HMGB1 in COAD, GBM, LIHC, ovarian cancer (OV)
3. Results was significantly higher than that in normal tissues
(all P < 0.001). However, the HMGB1 protein expression
3.1. Gene expression analysis in the breast cancer (BRCA) (P < 0.01), uterine corpus
TIMER2 was used to study HMGB1 expression among endometrial carcinoma (UCEC), LUAD, and PAAD
different TCGA tumor types. The results demonstrated (P < 0.001) was decreased.
that HMGB1 expression was significantly higher in By using GEPIA2 tool, we probed the relationship
cholangiocarcinoma (CHOL), colon adenocarcinoma between HMGB1 expression and tumor pathological stage,
(COAD), esophageal carcinoma (ESCA), head and including adrenocortical carcinoma (ACC), LIHC, skin
neck squamous cell carcinoma (HNSC), lung squamous cutaneous melanoma (SKCM), and THCA (Figure 1D, all
cell carcinoma (LUSC), stomach adenocarcinoma P < 0.05).
(STAD), liver hepatocellular carcinoma (LIHC), rectum
adenocarcinoma (READ) (P < 0.001), bladder urothelial 3.2. Survival analysis
carcinoma (BLCA), and glioblastoma multiforme (GBM) On the basis of the expression level of HMGB1, cancer
(P < 0.01) than in the control group. Obviously, there cases were divided into low expression group and high
were also some tumor types showing undifferentiated expression group. Then, we used TCGA and GEO datasets
expression (e.g., kidney renal clear cell carcinoma [KIRC], to explore the correlation between HMGB1 expression
cervical squamous cell carcinoma and endocervical and prognosis of patients with different tumor types. As
adenocarcinoma [CESC], thyroid carcinoma [THCA], displayed in Figure 2A, the high expression of HMGB1 was
and pheochromocytoma and paraganglioma [PCPG]). related to poor OS in ACC (P < 0.01) and LUAD (P < 0.05)
cancers. On the contrary, the low expression of HMGB1
Table 1. Methods and corresponding tools was related to poor OS in KIRC (P < 0.05) and THYM
Methods Tools (P < 0.05). DFS analysis displayed that the high expression
Gene expression analysis TIMER2, GEPIA2, UALCAN, TCGA, of HMGB1 was associated with the poor prognosis in ACC
GTEx, CPTAC (P < 0.001), CESC (P < 0.01), HNSC (P < 0.05), LUAD
Survival analysis GEPIA2, TCGA, GEO (P < 0.05), and SARC (P < 0.05) (Figure 2B).
Genetic variation analysis cBioPortal, TCGA 3.3. Genetic variation analysis
Immune infiltration analysis TIMER2, TCGA Cancer in humans occurs as a result of an accumulation
Gene enrichment analysis STRING, TCGA, GEPIA2, TIMER2, of genetic changes. We researched the HMGB1 genetic
Jvenn, R language variations among different tumor samples in the TCGA
TCGA: The Cancer Genome Atlas; TIMER2: Tumor immune list. From our analysis, we found that the frequency
estimation resource, version 2; GTEx: Genotype-tissue expression; of HMGB1 variation (>4%) was the highest in DLBC
CPTAC: Clinical Proteomic Tumor Analysis Consortium;
GEPIA2: Gene Expression Profiling Interactive Analysis, version 2; tumors, mainly including “mutation” and “deep deletion”
GEO: Gene Expression Omnibus; Jvenn: An interactive Venn diagram viewer. types. Colorectal cases have the highest frequency in the

Volume 2 Issue 1 (2023) 3 https://doi.org/10.36922/gpd.301