High HSPA8 expression in many pan-cancer, including COAD.
The data of RNA-seq was download from UCSC XENA in TCGA and GTEx formats uniformly processed. The expression level of HSPA8 increased significantly in 21 types of cancer (Figure 1A), colon cancer (COAD) included (Figure 1B), as for paired tumor and adjacent samples, HSPA8 was found clearly high expressed as well (Figure 1C), the differences above all was statistically significant (P<0.001).
To further validate the expression of protein, we explored the patterns of HSPA8 in COAD by Human Protein Atlas and UALCAN. As was shown in Figure 1D, HSPA8 protein was not detected in normal colon tissues, while medium expressions were observed in COAD samples. We next analyzed the relationship of HSPA8 protein expression in colon cancer between normal and primary tumor by UALCAN (Figure 2A), and found that protein expression of tumor group surpassed normal tissues obviously (P<0.001).
Clinical Characteristics
The clinical data of 478 patients of COAD were obtained from TCGA, including age, race, gender, primary therapy outcome, CEA level, TNM stage, pathological stage, perineural invasion, lymphatic invasion (Table 1). Among them, HSPA8 expression was 50% high and 50% low in 478 cases, according the cut-off valve, which was the median HSPA8 expression. Correlation analysis elucidated that HSPA8 expression was significantly correlated with age (P=0.050), race (P=0.035) N stage (P<0.001), M stage (P=0.027), pathological stage (P=0.002), OS event (P=0.045).
Table 1 clinical characteristic of the COAD patients form TCGA.
Characteristic
|
Low expression of HSPA8
|
High expression of HSPA8
|
p
|
n
|
239
|
239
|
|
T stage, n (%)
|
|
|
0.150
|
T1
|
7 (1.5%)
|
4 (0.8%)
|
|
T2
|
37 (7.8%)
|
46 (9.6%)
|
|
T3
|
157 (32.9%)
|
166 (34.8%)
|
|
T4
|
37 (7.8%)
|
23 (4.8%)
|
|
N stage, n (%)
|
|
|
< 0.001
|
N0
|
121 (25.3%)
|
163 (34.1%)
|
|
N1
|
66 (13.8%)
|
42 (8.8%)
|
|
N2
|
52 (10.9%)
|
34 (7.1%)
|
|
M stage, n (%)
|
|
|
0.027
|
M0
|
162 (39%)
|
187 (45.1%)
|
|
M1
|
41 (9.9%)
|
25 (6%)
|
|
Pathologic stage, n (%)
|
|
|
0.002
|
Stage I
|
40 (8.6%)
|
41 (8.8%)
|
|
Stage II
|
76 (16.3%)
|
111 (23.8%)
|
|
Stage III
|
78 (16.7%)
|
55 (11.8%)
|
|
Stage IV
|
41 (8.8%)
|
25 (5.4%)
|
|
Gender, n (%)
|
|
|
0.521
|
Female
|
109 (22.8%)
|
117 (24.5%)
|
|
Male
|
130 (27.2%)
|
122 (25.5%)
|
|
Race, n (%)
|
|
|
0.035
|
Asian
|
2 (0.7%)
|
9 (2.9%)
|
|
Black or African American
|
38 (12.4%)
|
25 (8.2%)
|
|
White
|
122 (39.9%)
|
110 (35.9%)
|
|
Age, n (%)
|
|
|
0.050
|
<=65
|
108 (22.6%)
|
86 (18%)
|
|
>65
|
131 (27.4%)
|
153 (32%)
|
|
CEA level, n (%)
|
|
|
0.101
|
<=5
|
91 (30%)
|
105 (34.7%)
|
|
>5
|
61 (20.1%)
|
46 (15.2%)
|
|
Perineural invasion, n (%)
|
|
|
0.289
|
NO
|
68 (37.6%)
|
67 (37%)
|
|
YES
|
28 (15.5%)
|
18 (9.9%)
|
|
Lymphatic invasion, n (%)
|
|
|
0.051
|
NO
|
122 (28.1%)
|
144 (33.2%)
|
|
YES
|
94 (21.7%)
|
74 (17.1%)
|
|
OS event, n (%)
|
|
|
0.045
|
Alive
|
178 (37.2%)
|
197 (41.2%)
|
|
Dead
|
61 (12.8%)
|
42 (8.8%)
|
|
To further verify the correlation between COAD clinicopathological factors and the HSPA8 high-low dichotomy, logistic analysis was performed in this study. As a result, high HSPA8 expression showed a significant negative correlation with N stage (N1&N2) (odd ratio, 0.478; P<0.001), M stage (M1) (odd ratio, 0.528; P=0.021), pathologic stage (Stage III & Stage IV) (odd ratio, 0.513; P<0.001), whereas positively correlated with age (>65) (odd ratio, 1.467; P=0.041) (Table 2). ROC curve analysis was adapted to evaluated the potential valve of HSPA8 in differentiating patients from population, the AUC was 0.889, the result showed that HSPA8 was a potential biomarker for the diagnosis of COAD (Figure 2B). Moreover, the Wilcoxon rank sum test was employed to contrast the HSPA8 expression in patients with distinct clinicopathological characters. As a result, the expression of HSPA8 was obviously negative associated with N stage (P<0.001), M stage (P=0.013), pathologic stage (P<0.001), lymphatic invasion (P=0.03) (Figure 2C-2F).
Table 2 Univariate regression analysis of HSPA8 and other clinical characteristics with OS in COAD
Characteristics
|
Total(N)
|
Odds Ratio (OR)
|
P value
|
T stage (T3&T4 vs. T1&T2)
|
477
|
0.857 (0.544-1.347)
|
0.504
|
N stage (N1&N2 vs. N0)
|
478
|
0.478 (0.329-0.692)
|
<0.001
|
M stage (M1 vs. M0)
|
415
|
0.528 (0.305-0.901)
|
0.021
|
Pathologic stage (Stage III&Stage IV vs. Stage I&Stage II)
|
467
|
0.513 (0.353-0.743)
|
<0.001
|
Age (>65 vs. <=65)
|
478
|
1.467 (1.017-2.120)
|
0.041
|
Perineural invasion (YES vs. NO)
|
181
|
0.652 (0.326-1.281)
|
0.219
|
CEA level (>5 vs. <=5)
|
303
|
0.654 (0.405-1.049)
|
0.079
|
Performance of DEGs in COAD samples with high- and low expressed HSPA8
The gene expression profiles were divided into high- and low-expression groups according to the criteria of median HSPA8 mRNA expression (Additional file 1: S1). analyzed for differences in the median mRNA expression. A total of 3321 DEGs from gene expression RNA-seq-HTSeq-Counts, including 9 up-regulated and 3312 down-regulated, were identified statistically significant between the two groups (|log fold change (log FC) | > 1.5, P < 0.05) (Figure 3A). The top five up-regulated DEGs and top five down-regulated DEGs were shown by the heat map (Figure 3B).
GO/KEGG functional enrichment analysis and PPI network construction
In association with functional enrichment analysis via GO/KEGG, the correlation with the biological process (BP) consisted of sensory perception of bitter taste and corresponding detection of chemical stimulus involved; cellular component (CC) included nuclear nucleosome and muscle myosin complex; molecular function (MF) included bitter taste receptor activity, taste receptor activity; KEGG included taste transduction, olfactory transduction (Figure 4A, 4B).
We also employed STRING to explore the PPI network of HSPA8 protein to analysis the interaction in COAD, the top 10 gene names and corresponding protein, as well as the annotations and score are shown in Figure 5A, 5B. Moreover, with a threshold of 0.4, the network of HSPA8 and its potential co-expressed genes in HSPA8-related DEGs was constructed by SRING, a total of 114 protein coding were screened out with the threshold of |log fold change (log FC) | > 1.5, P < 0.05 (Figure 5C) (Additional file 1: S2).
Immune infiltration analysis in COAD
Spearman correlation analysis demonstrated that the HSPA8 expression level in the COAD tumor microenvironment was relevant to the immune infiltration quantified by ssGSEA, the expression level of HSPA8 had clearly positive correlation with infiltrating level of Th2 cells, but negative correlation with Plasmacytoid DC (Figure 6).
The expression of HSPA8 affected the prognosis of COAD in patients with different clinicopathological status
Kaplan-Meier survival analysis shown that high HSPA8 expression in COAD patients was significantly associated with the longer overall survival (OS) (hazard ratio [HR], 0.66 (0.45-0.98); P=0.041) (Figure 7A). It also presented that high expression of HSPA8 correlated with good prognosis in the subgroups, including male (P=0.038), high T stage (T2-4) (P=0.017), no lymphatic invasion (P=0.035), no perineural invasion (P=0.035) (Figure 7B-7E).