Patient Characteristics
A total of 1083 BC patients with RNA-sequencing data and clinical-pathologic information were obtained in this study, of which 112 patients had matched normal tissue samples retrieved from TCGA. According to the median expression level of OCRL, BC patients with related clinical-pathologic information were categorized into high and low groups (Table 1).
The expression levels of OCRL gene in BC and other 32 cancers
The pan-cancer analysis revealed that OCRL was significantly elevated in 14 types of cancers, and showed no statistical difference in 4 types of cancers, but were decreased in 12 types of cancers (Fig. 1A). The OCRL expression was dramatically higher in BC tissue than non-matched normal tissues (p<0.001) (Fig. 1B). In addition, the expression of matched samples also showed that the OCRL was significantly elevated in nearly half of cancers (7/18), and showed no statistical difference in 9 types of cancers, but were decreased in only 2 types of cancers (Fig. 1C). Notably, the expression of OCRL was significantly overexpressed in BC samples than 112 matched normal samples (Fig. 1D). Furthermore, we found that high expression of OCRL was significantly associated with the age (p<0.001), PR status (p=0.005), ER status (p=0 .002) and PAM50 (p<0.001) (Table.1). Subsequently, the logistic regression analysis also showed that these clinicopathological differences between the high and low expression of OCRL groups, including age (odds ratio [OR]=1.564, 95%CI 1.229-1.992, p<0.001), PR status (OR=1.464, 95%CI=1.129-1.903, p=0.004), ER status (OR=1.559, 95%CI=1.165-2.093, p=0.003) (Table 2).
In order to verifying the expression of OCRL, IHC staining from HPA database was carried out to reveal the over-expressed OCRL in BC tissues compared with paired normal tissues (Fig. 2A). In addition, our results also revealed that the expression of ORCL in 80% (4/5) of BC tissues compared with related paracancerous tissues was upregulated by qRCR and western blot (Fig.2 B-D). There were 5 females with a mean age of 53 years (range 30–68 years) involved in our cohort.
The interaction proteins and co-expression netowrk with OCRL in BC
For better understanding the OCRL gene participating in biological functions in BC, we showed the potentially related genes and constructed the protein-protein interaction (PPI) network by STRING website (Fig. 3A). In addition, the co-expression genes of OCRL were investigated by the LinkFinder module in the LinkedOmics web portal from TCGA-BC. A total of more than 20000 genes was provided by the volcano plot, and dark red dots showed the positive association, and dark green dots showed the negative association (Fig. 3B). Among these related genes, we respectively identified the top 50 genes positively and negatively as associated with OCRL in heat maps (Fig. 3C, D). Surprisingly, the top 50 positively related genes almost were unfavourable genes with high HR in BC. Meanwhile, the top 50 negatively related genes were low-risk markers, of which 40/50 genes showed protective hazard ratio (HR). (Fig. 3E). These results showed the wide impact of OCRL co-expression network on the prognosis of BC patients.
Functional Enrichment Analysis
The analysis of KEGG pathway revealed co-expressed genes of OCRL were mainly concentrated in Ubiquitin mediated proteolysis, Inositol phosphate metabolism, Homologous recombination, Fanconi anemia pathway, RNA transport, etc (Fig. 4A). GO term annotation revealed co-expressed genes of OCRL were mainly concentrated in microtubule cytoskeleton organization involved in mitosis, DNA strand elongation, protein polyubiquitination, spindle organization, DNA replication, etc (Fig. 4B). Above results showed the potential function and mechanisms of OCRL on the prognosis in BC.
Correlation between methylation and expression of OCRL
To sutdy the underlying mechanisms of OCRL overexpression in BC tissues, we further explored the association between OCRL expression and methylation status using the UALCAN database (p<0.001) (Fig.5A). Surprisingly, we found that the DNA methylation at the promoter of OCRL was significantly higher in BC tissues; however, the DNA methylation in higher pathological stage exhibited a lower level of than that in the lower pathological stage (p<0.001) (Fig.5B). In addition, we observed that the nearly half of methylation sites in the DNA sequences of OCRL were hypomethylated in BC (Fig.5C). Finally, several methylation sites were indicative of better prognosis, including cg05032353, cg05147108, and cg16716035 (Fig.5D–F).
Association between OCRL expression and immune infiltration level
As shown in Fig. 6A, we observed that the expression of OCRL was notably negatively associated with various immune cells infiltration in BC from the TCGA database. It was worth noting that the enrichment scores of (pDCs), cytotoxic cells, dendritic cells (DCs), CD8+ T celss in the high-OCRL group were dramatically lower in high-RCRL group than those in the low-OCRL group (p < 0.001) (Fig. 6B-E). The negative associations between OCRL expression with plasmacytoid dendritic cells cells (pDCs) (r=-0.372, p<0.001), cytotoxic cells (r=-0.293, p<0.001), dendritic cells (DCs) (r=-0.267, p<0.001), CD8+ T celss (r=-0.259, p<0.001) were all proved by Pearson correlation analysis (Fig. 6F-I). These results showed that high OCRL expression markedly effected immune cells infiltration.
Association between OCRL with immune biomarkers
In order to further clarifying the underlying mechanisms of OCRL expression on influencing immune response in BC tissues, we also studied the correction between OCRL expression and various immune signatures from TISIDB database, including the immune-related biomarkers of 28 tumour-infiltrating lymphocytes (TILs) types, immunomodulators, chemokines and receptors (Fig. 7A-F).
In our study, OCRL expression was negatively corrected with TILs, such as Monocyte_abundance, Th1_abundance, Act_B_abundance, and Marophage_abundance (Fig. 7A). Furthermore, the expression of OCRL showed the markedly negative correlations with immunomodulators, which included major histocompatibility complex (MHC) molecules, immunoinhibitors, and immunostimulators (Fig. 7B-D). Fig. 7B showed the negatively correlations between the expression of OCRL and MHC molecules, such as HLA-F_exp, HLA-DPB1_exp, HLA-E_exp and HLA-DOB_exp. Fig. 5C showed the negative correlations between OCRL expression and immunostimulators, such as TNFRSF4_exp, C10orf54_exp, TNFRSF14_exp and TNFRSF8_exp, and. Fig. 5D showed the negative correlations between OCRL expression and immunoinhibitors, such as TGFB1_exp, LGALS9_exp, PDCD1_exp and CD244_exp. Fig. 5E showed negative correlations between OCRL expression and chemokines, such as CXCL2_exp, CCL14_exp, CCL19_exp, and CX3CL1_exp. Fig. 5F showed negative correlations between OCRL expression and receptors, such as CCR10_exp, CXCR5_exp, CXCR3_exp, and CCR7_exp. Therefore, we confirmed that OCRL played potential roles in regulating various immune molecules in BC to affect immune cells infiltration and function in the tumor microenvironment.
Prognostic value of OCRL in BC
In order to determining the role of OCRL in the prognosis, the OS were compared using the KM method in BC patients. The median value of OCRL expression was used as a cut-off score, and the patients were divided into high and low OCRL expression groups. Compared with the high OCRL expression group, the OS of the low OCRL expression group showed a significantly better prognosis (OS: hazard ratio [HR] = 1.73, 95% CI=1.24–2.40, p<0.001) (Fig.8A). Furthermore, the Cox proportional regression was used to analyze independent prognostic factors (Fig. 8B). The multivariate analysis indicated that OCRL expression was independently correlated with the worse OS (HR=1.835, 95% CI=1.108–3.040, p<0.001). Other characteristics, such as older (HR=2.850, 95% CI=1.717–4.729, p<0.001), advanced T stage (T1 & T2) (HR=2.489, 95% CI=1.259–4.921, p<0.001) and M1 status (HR=3.163, 95% CI=1.090–9.176, p<0.001), were associated with the worse OS.
In addition, we also observed that BC patients with high OCRL expression was significantly associated with unfavorable prognosis in several subgroups, including T1 & T2, N0 & N1, M0, stage I & II, age > 60 years, Luminal A, HER2-positive or HER2-negative, ER-positive and PR-positive subgroups (Figure 9).
Construction and validation of a nomogram
To predict the prognosis probability of patients with BC, a OCRL expression-based nomogram integrating independently prognostic factors was constructed (Fig. 10A). Every factor corresponded to a risk score, and the sum of the scores were accumulated on a range from 0 to 250. The 1-year, 5-year, 10-year survival probability of BC patients was respectively determined by drawing vertical lines to the outcome axis. Also, the corresponding calibration curves showed prediction efficacy, indicating that the nomogram was an appropriate prognostic model to guide clinical work (Fig. 10B-D).