The expression of HOXC10 is correlated with tumor-infiltrating immune cells in basal-like breast cancer and serves as a prognostic biomarker.

Background: Homeobox gene C10 (HOXC10) plays a vital role in the occurrence and development of several cancers, but its effects and underlying mechanism in the prognosis of different subtypes of breast cancer remain unclear.
Methods: First, we evaluated and compared the expression levels of HOXC10 cancer to normal tissues in the Oncomine and Tumor Immune Estimation Resource (TIMER) databases. Second, the correlation between HOXC10 and the survival of patients with different types of cancer, including breast cancer, was analyzed in the PrognoScan, Gene Expression Profiling Interactive Analysis (GEPIA), and Kaplan-Meier plotter databases. Finally, the relationship between HOXC10 and immune-infiltration levels or gene marker sets of immune cells in basal-like breast cancer (BLBC) was investigated in the TIMER and GEPIA databases.
Results: The expression of HOXC10 was elevated in breast cancer tissues. High HOXC10 expression indicated a poor prognosis for breast cancer patients, and expression affected the survival time of lymph-node positive or grade III breast cancer patients. In BLBC, the median overall survival (OS) of patients with high HOXC10 expression was significantly shorter than that of patients with low HOXC10 expression. HOXC10 was positively correlated with the immune infiltration of macrophages in BLBC. Breast cancer patients with low HOXC10 expression in different enriched immune-cell subgroups had a favorable prognosis. Conclusions: The level of HOXC10 expression increased significantly in breast cancer, and elevated HOXC10 was positively correlated with immune-cell infiltration and poor prognosis in BLBC. These findings shed light on the important role of HOXC10 in breast cancer. HOXC10 should be recognized as a prognostic biomarker for determining prognosis and immune infiltration in BLBC patients. 2022 Annals of Translational Medicine. All rights reserved.

Introduction

Breast cancer, which has the highest incidence of cancers in women and is the second leading cause of cancer-related deaths in women, is a heterogeneous disease with different biological and histological characteristics that lead to distinct prognoses and respond differently to specific therapies (1,2). Based on gene expression profiling, breast cancer can be divided into the following 5 intrinsic subtypes: normal-like, luminal A, luminal B, human epidermal growth factor receptor 2-enriched (Her-2+), and basal-like breast cancer (BLBC) (3). BLBC was so named because its gene expression profile is similar to that of basal cells (4,5). In clinical settings, the majority of BLBCs are triple-negative breast cancers, and estrogen receptor (ER), progesterone receptor (PR), and Her-2 are all negatively diagnosed by immunohistochemistry (6). As the most aggressive subtype of breast cancer, BLBC is characterized by early distant metastasis, a high resistant rate, and poor outcomes which comprises 8% to 20% of all breast cancer (7,8). However, as the molecular mechanisms underlying the aggressive characteristics of BLBC are not well understood, identifying new biomarkers for predicting patients’ prognoses and providing targets for effective targeted therapies remain challenging. BLBC has been classified as basal-like immune activated (BLIA), basal-like immune-suppressed (BLIS), luminal androgen receptor (LAR) and mesenchymal (MES), suggesting potential immune therapeutic targets for BLBC. Thus, it is necessary to clarify the relationship between BLBC and immune invasion, and find an immune related biomarker to indicate the prognosis of BLBC.

Homeobox genes (HOXs) make important contributions to cell differentiation and morphogenesis in multicellular organisms by encoding a family of evolutionarily conserved transcription factors that bind to the promoters of target genes (9). In mammals, 39-gene HOX clusters reside in 4 separate chromosomal linkage groups designated as homeobox gene A (HOXA), homeobox gene B (HOXB), homeobox gene C (HOXC), and homeobox gene D (HOXD) that share identical organization (10). It is confirmed that HOX genes act as oncogenes or tumor suppressors in the occurrence and progression of many different cancers (11-14). HOXC genes are located on chromosome 12q13.3 (15).

Homeobox gene C10 (HOXC10), a member of the HOXC clusters, participates in the assembly of replicative complexes, the development of the spinal cord, and the formation of neurons by encoding a transcription factor containing a highly conserved DNA binding homeodomain (16-18). There is increasing evidence that HOXC10 plays a vital role in the occurrence and development of several cancers by influencing cancer cell proliferation, apoptosis, invasion and avoidance of immune destruction (19-21). Zhai et al. (22) analyzed normal cervical epithelial tissue, cervical intraepithelial neoplasia, and invasive squamous cell carcinoma tissues by gene chip and found that HOXC10 expression was upregulated with the increase of cervical epithelial malignancy, which confirmed that HOXC10 protein was related to the progression of cervical cancer in vivo and in vitro. Researchers have also found that the high expression of HOXC10 protein is closely related to the poor prognosis of thyroid cancer (23). In breast cancer research, it has been shown that HOXC10 is highly expressed in breast cancer and also an estrogen-responsive gene which might be associated with hormone-induced cell differentiation, development and human diseases (18). However, the effects and mechanisms underlying HOXC10 in the prognoses of different subtypes of breast cancer, especially BLBC, remain unclear.

In the present study, the different expression levels of HOXC10 in different cancer and normal tissues were assessed by the Oncomine and Tumor Immune Estimation Resource (TIMER) databases. The correlations between HOXC10 and the prognoses of cancer patients, including those with different subtypes of breast cancer, were analyzed using the PrognoScan, Gene Expression Profiling Interactive Analysis (GEPIA) and Kaplan-Meier plotter databases. Further, the relationship between HOXC10 and the immune-infiltration levels of immune cells in different tumor microenvironments were investigated in the TIMER database. The findings of this study shed light on the predictive role of HOXC10 in the prognoses of patients with BLBC and also identify a potential mechanism related to immune infiltration.

We present the following article in accordance with the REMARK reporting checklist (available at https://atm.amegroups.com/article/view/10.21037/atm-21-6611/rc).

Methods

Analysis in the Oncomine database

The HOXC10 expression of various types of cancers was identified in the Oncomine database (https://www.oncomine.org/resource/login.html) (24). The threshold was determined according to the following values: a P value of 0.001, a fold change of 1.5, and the gene ranking of all. The data came from messenger ribonucleic acid (mRNA).

Survival analyses in the PrognoScan, GEPIA, and Kaplan-Meier plotter databases

To investigate the effects of HOXC10 in the prognosis of different types of cancer, the correlations between HOXC10 expression levels and the survival of cancer patients were analyzed using the PrognoScan (http://www.abren.net/PrognoScan/) (25), GEPIA (http://gepia.cancer-pku.cn/index.html) (26), and Kaplan-Meier plotter databases (http://kmplot.com/analysis/) (27). The Kaplan-Meier plotter database were analyzed by filtering the intrinsic subtypes of breast cancer including BLBC to evaluate the relationship between HOXC10 expression and the outcomes of breast cancer patients with different intrinsic subtypes. The threshold was determined according to the following values: a Cox P value <0.05 in the PrognoScan database, and a log-rank P value <0.05 in the GEPIA and Kaplan-Meier plotter databases. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Correlation analyses between HOXC10 and immune cells in the TIMER, GEPIA, and Kaplan-Meier plotter databases

To systematically analyze the levels of immune invasion in diverse types of cancers and differences in the gene expression of different cancers, we used the TIMER database (https://cistrome.shinyapps.io/timer/) to evaluate data from The Cancer Genome Atlas (TCGA), which included 10,897 samples across 32 cancer types (28,29). First, we confirmed the expression of HOXC10 across data from 32 cancer tissues compared to data from paired normal tissues using the TIMER database. Second, we investigated the correlations of HOXC10 with typical infiltrating immune cells [i.e., B cells, cluster of differentiation (CD)4+ T cells, CD8+ T cells, neutrophils, macrophages, and dendritic cells (DCs)] using the TIMER algorithm. Third, we further analyzed the correlations between HOXC10 and genetic markers of immune cells in breast invasive carcinoma (BRCA) and different intrinsic subtypes of BRCA via correlation modules in the TIMER database. The gene markers of tumor-infiltrating immune cells included the markers of B cells, CD8+ T cells, general T cells, neutrophils, natural killer (NK) cells, monocytes, tumor-associated macrophages (TAMs), M1 macrophages, M2 macrophages, DCs, T-helper 1 (Th1) cells, T-helper (Th2) cells, follicular helper T (Tfh) cells, Th17 cells, regulatory cells (Tregs), and T cell exhaustion as described in previous studies (30,31).

The comprehensive GEPIA database, which comprises 9,736 tumor samples and 8,587 normal samples from TCGA and the Genotype-Tissue Expression Project (GTEx) programs (26), was also used to further confirm the significantly correlated genes in TIMER, which analyzes RNA-sequencing expression. The Spearman’s correlation test was selected, and all other default settings were accepted before analyzing the gene correlations.

Finally, to confirm the correlations between HOXC10 and biomarkers of immune cells in BRCA, we used a pan-cancer analysis of the Kaplan-Meier plotter database to export the different outcomes of patients with BRCA based on different levels of immune cells. The automatic best performing threshold was used as the cutoff. The results displayed the false discovery rates in addition to the P values.

Statistical analysis

The results of the Oncomine analysis are displayed with their P values, fold changes, and ranks. The survival curves were generated by the PrognoScan, GEPIA, and Kaplan-Meier plotter databases, and were displayed with the hazard ratio (HR) and P or Cox P values from the log-rank tests. The Spearman’s correlation analysis and statistical significance results were used to evaluate correlations between HOXC10 and immune infiltration or type markers of immune cells. For all the analyses in the present study, a P value <0.05 was considered statistically significant.

Results

The mRNA expression levels of HOXC10 in different human cancers

To study changes in HOXC10 expression between normal and tumor tissues, the HOXC10 mRNA levels of different tumors and normal tissues of multiple cancer types were analyzed first using the Oncomine database, from which 147 data sets were selected, including 30,079 samples. The analysis showed that the expression levels of HOXC10 were significantly higher in brain and central nervous system (CNS), breast, cervical, colorectal, esophageal, gastric, lung and sarcoma cancers tissues than normal tissues (see ). Additionally, the expression levels of HOXC10 were lower in kidney, melanoma, and ovarian cancer tissues than normal tissues in some data sets. The detailed results of HOXC10 expression in different cancer types are summarized in Table S1.

Figure 1

The mRNA expression levels of HOXC10 in different types of cancer. (A) The differential expression level of HOXC10 in different cancer and normal tissues in the Oncomine database. (The threshold was determined according to the following values: a P value of 0.001, a fold change of 1.5, and the gene ranking of all); (B) HOXC10 expression levels of different tumor types in TCGA database were detected by TIMER, and were significantly higher in the BRCA, ESCA, GBM, HNSC, LUAD, LUSC, STAD, and THCA tissues than the adjacent normal tissues. **, P<0.01; ***, P<0.001. HOXC10, Homeobox gene C10; TCGA, The Cancer Genome Atlas; TIMER, Tumor Immune Estimation Resource; BRCA, breast invasive carcinoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; STAD, stomach adenocarcinoma; THCA, thyroid carcinoma.

To further evaluate HOXC10 expression in human cancers, we analyzed the expression levels of HOXC10 between different tumor and normal tissues based on the RNA-sequencing data of multiple malignancies in TCGA database. shows the differential expression of HOXC10 between the tumor and adjacent normal tissues across all TCGA tumors. The expression levels were significantly higher in BRCA, esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), stomach adenocarcinoma (STAD), and thyroid carcinoma (THCA) tissues than adjacent normal tissues. However, HOXC10 expression was significantly lower in kidney chromophobe (KICH) and prostate adenocarcinoma (PRAD) tissues than adjacent normal tissues (see ).

Predictive effect of HOXC10 in the prognosis of patients with different cancers

Based on the differences in HOXC10 expression between tumor and normal tissues, we first analyzed the associations between HOXC10 and survival across different cancers using the PrognoScan database (see Table S2). Only Cox P values <0.05 are shown in . HOXC10 expression significantly affected the prognosis of 3 type cancers (i.e., breast, brain and ovarian cancer). GSE1378 and GSE1379, 2 cohorts of 60 patients with breast cancer, respectively, indicated that a high level of HOXC10 expression was significantly associated with a high risk of relapse in breast cancer patient [GSE1378, relapse-free survival (RFS) HR =1.27%, 95% confidence interval (CI): 1.03–1.56, Cox P=0.024856; GSE1379, RFS HR =1.25%, 95% CI: 1.01–1.54, Cox P=0.036064; see ]. Further, the high expression of HOXC10 in ovarian cancer and brain cancer was related to a poor prognosis (see ).

Figure 2

Prognostic potential of HOXC10 across different cancers in the PrognoScan databases. (A) A forest plot displaying the prognostic role of HOXC10 using the PrognoScan database (Cox P value <0.05). Orange represents the results for breast cancer; blue represents the results for ovarian cancer and purple represents the results for brain cancer; (B-F) the Kaplan-Meier survival curves between the high and low expression of HOXC10 in different types of cancer in PrognoScan. (B,C) The Kaplan-Meier survival curves of RFS in 2 breast cancer cohorts (GSE1378, n=60; GSE1379, n=60); (D) the Kaplan-Meier survival curve of OS in a brain cancer cohort (GSE4271-GPL96, n=77); (E,F) the Kaplan-Meier survival curves for OS and DFS in an ovarian cancer cohort (GSE26712, n=185). RFS, relapse-free survival; OS, overall survival.

Second, we used GEPIA database to analyze TCGA data to evaluate the predictive value of HOXC10 in human cancers. Notably, the high or low expression of HOXC10 made no significant difference in the disease-free survival (DFS) of BRCA patients. However, BRCA patients with a low expression of HOXC10 had better overall survival (OS) than those with a high expression of HOXC10 (see Figure S1A). In addition, a high expression of HOXC10 in adrenocortical carcinoma (ACC), colon adenocarcinoma (COAD), and stomach adenocarcinoma (STAD) was correlated to poor DFS and OS (see Figure S1B-D). However, the high expression of HOXC10 in kidney renal clear cell carcinoma (KIRC) indicated a favorable prognosis (see Figure S1E).

Third, we analyzed large databases, including the Gene Expression Omnibus (GEO), European Genome-Phenome Archive (EGA), and TCGA databases, using the Kaplan-Meier plotter database to explore the potential prognostic relationship between HOXC10 and different cancers (see Figure S2). In breast cancer, the median OS and post-progression survival (PPS) times of patients with lower levels of HOXC10 expression were 120 and 43.2 months, respectively, while that of patients with higher levels of HOXC10 expression were 85.2 and 26.8 months, respectively, and the differences were significant (P=0.00353 and P=0.03, respectively; see Figure S2A). The same phenomenon was observed in lung cancer and gastric cancer (see Figure S2C,D). Conversely, a higher expression of HOCX10 was related to better RFS in ovarian cancer patients (P=0.038; see Figure S2B).

Prognostic potential of HOXC10 in BLBC

The above-mentioned 3 analyses confirmed that the expression level of HOXC10 could be used to predict the prognosis of patients with breast cancer. To further evaluate the relationship between HOXC10 expression and the outcomes of breast cancer patients with different intrinsic subtypes, the Kaplan-Meier plotter database was analyzed by filtering the intrinsic subtype of breast cancer. There was no significant difference between patients with high HOXC10 expression and those with low HOXC10 expression in luminal A, luminal B, and Her-2 + breast cancer (see Figure S3). Notably, the median OS of patients with high and low HOXC10 expression in BLBC was 55.4 and 115 months, respectively, and the difference was significant (HR =1.97%, 95% CI: 1–3.87; P=0.045; see ). The same prognostic effect of HOXC10 was also found in relation to the PPS of BLBC patients (HR =2.14%, 95% CI: 1.02–4.47; P=0.04; see ). There is no significant difference in RFS and DMFS (See ). The analysis results showed that the increased expression of HOXC10 in BLBC was related to a poor prognosis, which indicates that HOXC10 can serve as a biomarker for the prognosis of BLBC patients.

Figure 3

The expression of HOXC10 affected the outcomes of patients with BLBC in the Kaplan-Meier plotter database. (A) OS curve (n=241); (B) PPS survival curve (n=64); (C) DMFS survival curve (n=232); (D) RFS survival curve (n=618). The OS and PPS of HOXC10 patients in the over-expression group with BLBC were significantly longer than those of patients in the under-expression group (P=0.045 and P=0.04, respectively). BLBC, basal-like breast cancer; OS, overall survival; PPS, post-progression survival; DMFS, distant metastasis-free survival; RFS, relapse-free survival.

HOXC10 expression affects the survival time of lymph-node positive or grade III breast cancer patients

In addition to the different intrinsic subtypes that could affect the prognosis, many clinicopathological factors are related to the survival of patients with breast cancer. We used the Kaplan-Meier plotter database to examine the relationships between HOXC10 and the prognosis of breast cancer with different clinicopathological characteristics (see ). Notably, the elevated expression of HOXC10 had adverse effects on both the OS and PPS of lymph-node positive breast cancer patients (OS: HR =1.79%, 95% CI: 1.09–2.94; P=0.0206; PPS: HR =1.39%, 95% CI: 1.09–1.77; P=0.0072). Moreover, high HOXC10 expression negatively affected the OS of grade III breast cancer patients (HR =1.55%, 95% CI: 1.03–2.33; P=0.0034). Thus, HOXC10 appears to affect the prognosis of lymph-node positive or advanced-stage (i.e., grade III) breast cancer patients.

Table 1

The associations between HOXC10 mRNA expression and the outcomes of breast cancer patients with different clinicopathological characteristics using the Kaplan-Meier plotter database

Clinicopathological characteristics	Overall survival (n=1,402)				Progression-free survival (n=3,955)
	N	Hazard ratio	P value		N	Hazard ratio	P value
ER status
ER positive	548	1.06 (0.71–1.58)	0.7903		2061	1.03 (0.85–1.25)	0.7426
ER negative	251	1.31 (0.77–2.23)	0.3143		801	1.17 (0.9–1.52)	0.2398
PR status
PR positive	83	0.61 (0.15–2.46)	0.4857		589	1.02 (0.69–1.51)	0.9309
PR negative	89	0.86 (0.31–2.42)	0.7761		549	1.32 (0.93–1.87)	0.121
Her-2 status
Her-2 positive	129	0.79 (0.38–1.64)	0.5245		252	0.9 (0.55–1.45)	0.6581
Her-2 negative	130	0.78 (0.31–1.94)	0.5876		800	1.17 (0.86–1.6)	0.3078
Lymph-node status
LN positive	313	1.79 (1.09–2.94)	0.0206		1133	1.39 (1.09–1.77)	0.0072
LN negative	594	0.89 (0.59–1.34)	0.5738		2020	1.06 (0.87–1.29)	0.5466
Grade
I	161	0.61 (0.23–1.58)	0.3049		345	1 (0.55–1.83)	0.99
II	387	1.05 (0.64–1.7)	0.85		901	1.13 (0.85–1.5)	0.4049
III	503	1.55 (1.03–2.33)	0.034		903	1.19 (0.92–1.54)	0.1792
TP53 status
Mutated	111	1.5 (0.57–3.97)	0.4106		188	1.72 (0.92–3.2)	0.0851
Wild type	187	0.65 (0.32–1.29)	0.2102		273	1.09 (0.66–1.79)	0.7383

ER, estrogen receptor; PR, progesterone receptor; Her-2, human epidermal growth factor receptor 2; LN, lymph node; TP53, tumor protein p53.

The transcription of HOXC10 is correlated with the immune-infiltration level and immune marker set in BLBC

Tumor-infiltrating lymphocytes are independent predictors of the sentinel lymph-node status and survival of cancer patients (31). Previous research has shown that tumor infiltration is related to the prognosis of breast cancer patients (32-34). The effect of HOXC10 transcription on the recruitment of immune cells and the cancer microenvironment remains unclear. Thus, we investigated the relationship between the immune-infiltration and transcription levels of HOXC10 in different types of cancer (see Figure S4), including different subtypes of breast cancer (see ) using the TIMER database to understand the mechanism underlying HOXC10 and the outcomes of breast cancer patients. There were no significant correlations between immune-infiltration of B cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophils, DCs and transcription levels of HOXC10 in BRCA (see ) and luminal breast cancer (see ). However, HOXC10 was positively correlated with the immune infiltration of macrophages in BLBC (Cor =0.207; P=0.0198; see ), but was negatively correlated with the immune infiltration of macrophages in Her-2+ breast cancer (Cor = –0.276; P=0.0357; see ).

Figure 4

Correlated relationship between immune infiltration and HOXC10 expression in different subtypes of breast cancer using the TIMER database. (A) BRCA; (B) BLBC; (C) Her-2+ breast cancer; and (D) luminal breast cancer. TIMER, Tumor Immune Estimation Resource; BRCA, breast invasive carcinoma; BLBC, basal-like breast cancer; Her-2, human epidermal growth factor receptor 2.

Further, the relationships between HOXC10 expression levels and the genetic markers of different immune cells, including B cells, CD8+ T cells, general T cells, neutrophils, NK cells, monocytes, TAMs, M1 macrophages, M2 macrophages, DCs, Th1 cells, Th2 cells, Tfh cells, Th17 cells, Treg, and T cell exhaustion in BRCA were explored using the TIMER database, which showed that there were no significant correlations between HOXC10 and these biomarkers (see and ). Next, the GEPIA database was used to explore the relationship between the biomarkers of monocytes, TAMs, M1 and M2 macrophage and HOXC10 expression in BRCA (see Figure S5). The results showed that the CD86, macrophage-colony stimulating factor 1 receptor (CSF1R) of monocytes, the chemokine (C-C motif) ligand 2 (CCL2) of TAMs, the CD163, v-set and immunoglobulin domain–containing 4 (VSIG4), and membrane spanning 4-domains A4A (MS4A4A) of M2 macrophages were negatively correlated with HOXC10 expression in BRCA.

Figure 5

The heatmap of the correlations between HOXC10 and the genetic markers of the infiltrating immune cells (including B cells, CD8+ T cells, general T cells, neutrophils, NK cells, Monocytes, TAMs, M1 macrophages, M2 macrophages, DCs, Th1 cells, Th2 cells, Tfh cells, Th17 cells, Treg, and T cell exhaustion) detected by the TIMER database in BRCA and different intrinsic subtypes of BRCA. CD, cluster of differentiation; NK, natural killer; TAM, tumor-associated macrophage; DC, dendritic cell; Th, T-helper; Tfh, follicular helper T; Treg, regulatory cells; TIMER, Tumor Immune Estimation Resource; BRCA, breast invasive carcinoma.

Table 2

Correlations between HOXC10 and the genetic markers of infiltrating immune cells in breast cancer and BLBC patients

Description	Gene markers	Breast cancer						BLBC
		None			Purity			None			Purity
		COR	P		COR	P		COR	P		COR	P
B cells	CD19	0.0465	0.12308		0.0503	0.11279		–0.172	0.04211		–0.1652	0.06136
	CD79A	0.0138	0.64678		0.0227	0.47361		–0.1649	0.05163		–0.1511	0.08733
	FCRL2	0.0207	0.49323		0.0292	0.35728		–0.161	0.05743		–0.1554	0.07873
	MS4A1	0.0188	0.53424		0.0286	0.36739		–0.2399	0.0043		–0.2414	0.00585
CD8+ T cells	CD8A	0.0242	0.42181		0.0309	0.33097		–0.2799	0.00085		–0.2709	0.00196
	CD8B	0.0363	0.22867		0.0387	0.22265		–0.1258	0.13842		–0.1072	0.22644
T cells (general)	CD3D	0.0117	0.69912		0.016	0.61489		–0.2878	0.00059		–0.2829	0.00121
	CD3E	0.021	0.48641		0.0258	0.41581		–0.3114	0.00019		–0.3066	0.00043
	CD2	0.0252	0.40456		0.0291	0.35933		–0.2909	0.00051		–0.2858	0.00107
Neutrophils	FCGR3B	–0.034	0.26644		–0.0377	0.23494		–0.1739	0.03987		–0.1804	0.04076
	CEACAM3	0.0183	0.54328		0.026	0.41242		–0.2023	0.01651		–0.2066	0.01885
	SIGLEC5	–0.012	0.6846		–0.0172	0.5872		–0.2804	0.00083		–0.2746	0.00169
	FPR1	–0.036	0.23721		–0.0355	0.26384		–0.2003	0.01777		–0.1853	0.03568
	CSF3R	0.0216	0.47348		0.0183	0.56378		–0.1501	0.07673		–0.1374	0.12035
	S100A12	0.0048	0.87368		0.0098	0.75699		–0.1344	0.11347		–0.12	0.17561
	CEACAM8	–0.005	0.86922		0.0104	0.7435		0.1001	0.23947		0.10664	0.22904
	ITGAM	–0.007	0.82108		–0.0196	0.53738		–0.3327	6.49E–05		–0.3285	0.00016
	CCR7	0.0193	0.52247		0.0213	0.50218		–0.3262	9.14E–05		–0.3249	0.00019
Natural killer cells	KIR2DL1	–0.061	0.04302		–0.0382	0.22844		–0.1815	0.03183		–0.1982	0.02435
	KIR2DL3	0.0073	0.8086		0.024	0.44993		–0.125	0.14118		–0.1431	0.10572
	KIR2DL4	–0.035	0.24025		–0.0307	0.33266		–0.1852	0.02849		–0.1901	0.03092
	KIR3DL1	–0.027	0.37454		–0.0272	0.39137		–0.2063	0.01448		–0.2099	0.01696
	KIR3DL2	–0.01	0.74839		0.0031	0.92324		–0.1862	0.02758		–0.1939	0.0277
	KIR3DL3	–0.057	0.00702		–0.0469	0.0364		–0.0763	0.20286		–0.0786	0.2083
	KIR2DS4	–0.044	0.14704		–0.0237	0.45493		–0.2407	0.00417		–0.2528	0.00385
	NCR1	–0.038	0.20267		–0.0369	0.24511		–0.1494	0.07804		–0.1478	0.09458
Monocytes	C3AR1	–0.039	0.19612		–0.0422	0.18389		–0.2644	0.00165		–0.2638	0.0026
	CD86	–0.028	0.34912		–0.0299	0.34683		–0.2717	0.00121		–0.2669	0.0023
	CSF1R	–0.112	0.00021		–0.1096	0.0005		–0.2974	0.00038		–0.2959	0.0007
TAMs	CCL2	–0.027	0.36706		–0.0231	0.46604		–0.1282	0.13095		–0.1297	0.14264
	CD68	0.0105	0.72878		0.0033	0.91686		–0.2297	0.00643		–0.2253	0.01038
	IL10	0.0428	0.15599		0.0439	0.16637		–0.2085	0.01357		–0.2019	0.0219
M1 macrophages	NOS2	–0.084	0.00551		–0.0871	0.006		0.0824	0.33338		0.0881	0.32081
	IRF5	–0.072	0.01708		–0.0742	0.0192		–0.1965	0.02011		–0.1702	0.05396
	PTGS2	–0.038	0.21377		–0.041	0.196		0.1884	0.02594		0.18034	0.04097
M2 macrophages	CD163	–0.058	0.05648		–0.0558	0.07849		–0.2163	0.0104		–0.2144	0.01482
	VSIG4	–0.097	0.00123		–0.0978	0.002		–0.2077	0.01395		–0.2056	0.01959
	MS4A4A	–0.019	0.5244		–0.0183	0.56427		–0.1845	0.02926		–0.1854	0.0356
Dendritic cells	HLA–DPB1	–0.037	0.2167		–0.0404	0.20339		–0.2307	0.00621		–0.2245	0.01067
	HLA–DQB1	–0.025	0.41401		–0.0405	0.20181		–0.2494	0.00304		–0.2451	0.00521
	HLA–DRA	0.0105	0.72911		0.0062	0.84426		–0.2521	0.00273		–0.2544	0.00372
	HLA–DPA1	–0.015	0.62853		–0.016	0.61391		–0.2102	0.0128		–0.2063	0.01914
	CD1C	–0.025	0.40064		–0.0197	0.53471		–0.2507	0.00282		–0.2557	0.00345
	NRP1	0.0331	0.27264		0.0326	0.30492		–0.0458	0.59081		–0.0313	0.7245
	ITGAX	–0.009	0.76693		–0.0067	0.83264		–0.3097	0.00021		–0.3043	0.00048
	CD209	–0.043	0.1503		–0.0448	0.15788		–0.0325	0.70262		–0.038	0.66897
Th1 cells	TBX21	0.0062	0.83834		0.0108	0.73436		–0.2319	0.00593		–0.2268	0.00987
	STAT4	0.0121	0.68963		0.0168	0.59651		–0.2586	0.0021		–0.2569	0.00338
	STAT1	0.0311	0.302		0.0318	0.31629		–0.1806	0.03288		–0.1666	0.05929
	IFNG	–0.019	0.51824		–0.0212	0.50352		–0.228	0.00674		–0.229	0.00903
	TNF	–0.03	0.31225		–0.0202	0.52431		–0.227	0.0071		–0.2341	0.0077
Th2 cells	GATA3	0.0266	0.37862		0.0412	0.19378		–0.1227	0.14842		–0.1324	0.13453
	STAT6	0.0116	0.70183		0.0208	0.51217		–0.1279	0.1319		–0.0999	0.25956
	STAT5A	–0.09	0.00288		–0.0804	0.0112		–0.1911	0.02387		–0.1723	0.05092
	IL13	0.0287	0.342		0.0292	0.35764		–0.1848	0.0288		–0.1868	0.03401
Tfh cells	BCL6	–0.131	1.26E–05		–0.12	0.0001		–0.1123	0.18628		–0.1259	0.15487
	IL21	–0.015	0.61721		–0.0136	0.66722		–0.2584	0.00206		–0.2468	0.00481
Th17 cells	STAT3	0.0293	0.33211		0.0473	0.13584		–0.0419	0.62252		–0.0566	0.52339
	IL17A	–0.017	0.56991		–0.0114	0.71881		–0.0662	0.43707		–0.0541	0.5426
Tregs	FOXP3	0.0801	0.00785		0.0874	0.0058		–0.2933	0.00046		–0.2842	0.00114
	CCR8	0.0756	0.01218		0.085	0.0073		–0.2948	0.00041		–0.2825	0.00118
	STAT5B	0.0232	0.44267		0.0398	0.20948		0.0554	0.51494		0.06682	0.45131
	TGFB1	–0.04	0.18603		–0.0551	0.08212		–0.1915	0.02354		–0.1844	0.03652
T-cell exhaustion	PDCD1	0.016	0.59556		0.0218	0.49221		–0.2002	0.01772		–0.1935	0.028
	CTLA4	0.0043	0.88702		0.0106	0.73839		–0.3091	0.00022		–0.3061	0.00044
	LAG3	0.0032	0.91633		–0.001	0.97483		–0.2495	0.00303		–0.2439	0.00546
	HAVCR2	–0.024	0.42419		–0.0263	0.4076		–0.3194	0.00013		–0.3172	0.00027
	GZMB	–0.017	0.56991		–0.0136	0.66762		–0.2446	0.00359		–0.2426	0.0056

BLBC, basal-like breast cancer; COR, correlation coefficient; TAM, tumor-associated macrophage; Th, T-helper; Tfh, follicular helper T; Treg, regulatory cells; CD, cluster of differentiation; FCRL2, Fc receptor like 2; MS4A1, membrane spanning 4-domains A1; FCGR3B, Fc gamma receptor IIIb; CEACAM, CEA cell adhesion molecule; SIGLEC5, sialic acid binding Ig like lectin 5; FPR1, formyl peptide receptor 1; CSF3R, colony stimulating factor 3 receptor; S100A12, S100 calcium binding protein A12; ITGAM, ntegrin subunit alpha M; CCR, C-C motif chemokine receptor; KIR2DL, killer cell immunoglobulin like receptor, two Ig domains and long cytoplasmic tail; KIR3DL, killer cell immunoglobulin-like receptor with three domains and long cytoplasmic tail; KIR2DS4, killer cell immunoglobulin like receptor, two Ig domains and short cytoplasmic tail 4; NCR1, natural cytotoxicity triggering receptor 1; C3AR1, complement C3a receptor 1; CSF1R, colony stimulating factor 1 receptor; CCL2, C-C motif chemokine ligand 2; IL, interleukin; NOS2, nitric oxide synthase 2; IRF5, interferon regulatory factor 5; PTGS2, prostaglandin-endoperoxide synthase 2; VSIG4, V-set and immunoglobulin domain containing 4; MS4A4A, membrane spanning 4-domains A4A; HLA-DPB1, major histocompatibility complex, class II, DP beta 1; HLA-DQB1, major histocompatibility complex, class II, DQ beta 1; HLA-DRA, major histocompatibility complex, class II, DR alpha; HLA-DPA1, major histocompatibility complex, class II, DP alpha 1; NRP1, neuropilin 1; ITGAX, integrin subunit alpha X; TBX21, T-box transcription factor 21; STAT, signal transducer and activator of transcription; IFNG, interferon gamma; TNF, tumor necrosis factor; GATA3, GATA binding protein 3; BCL6, B cell leukemia transcription repressor; FOXP3, forkhead box P3; TGFB1, transforming growth factor beta 1; PDCD1, programmed cell death 1; CTLA4, cytotoxic T-lymphocyte associated protein 4; LAG3, lymphocyte activating 3; HAVCR2, hepatitis A virus cellular receptor 2; GZMB, granzyme B.

Further, we explored the correlations between HOXC10 and genetic markers in different subtypes of BRCA. As shows, in contrast to its positive correlation with HOXC10 expression in luminal breast cancer, the correlations between genetic markers of immune cells in BLBC and Her-2+ were mostly negative (see and Table S3). Specifically, after the correlation adjustment by purity, the complement C3a receptor 1 (C3AR1), CD86 and CSF1R of monocytes, the CD68 and IL10 of TAMs, the CD163, VSIG4 and MS4A4A of M2 macrophage, and the interferon regulatory factor 5 (IRF5) of M1 macrophages were found to be significantly correlated with the expression levels of HOXC10 in BLBC (see ). Thus, HOXC10 appears to be closely related to macrophage polarization in BLBC, which further confirms that expressions in BLBC are correlated to immune infiltration.

Prognostic effect of HOXC10 expressions in BRCA based on immune cells

Both the TIMER and GEPIA databases were used to analyze the relationship between HOXC10 and biomarkers of immune cells in BRCA, but the results differed. To confirm the correlation, we used the Kaplan-Meier plotter database to analyze the relationship between HOXC10 and the OS of patients with BRCA based on different levels of immune cells. The results showed that high HOXC10 expressions of BRCA in enriched CD4+ memory T cells (HR =1.6, P=0.024), enriched macrophages (HR =1.71, P=0.034), enriched NK cells (HR =2.07, P=0.032), and enriched Th2 cells (HR =1.83, P=0.029) was negatively correlated with OS time, while the high expression of HOXC10 had no significant correlation with OS in decreased subgroups of the immune-cells (see ). There was no significant correlation between high HOXC10 and the outcomes of BRCA patients with enriched B cells, regulatory T cells, and Th1 cells. However, compared to the low HOXC10 expression group, the high HOXC10 expression group had a poorer prognosis regardless of whether or not they had enriched or decreased CD8+ memory T cells, eosinophils, and MES stem cells (see Figure S6). We also analyzed the prognostic potential of RFS in BRCA based on the different immune-cell subgroup (see Figure S7). All the results indicated that a high level of HOXC10 affects the survival of BRCA patients, which may be due to the infiltration of some immune cells.

Figure 6

The correlations between HOXC10 expression and the OS of patients with breast cancer across different subgroups of immune cells based on an analysis of the Kaplan-Meier plotter database. (A,B) The OS curves in enriched CD4+ memory T-cell subgroup and decreased CD4+ memory T-cell subgroup; (C,D) the OS curves in the enriched macrophage subgroup and the decreased macrophage subgroup; (E,F) the OS curves in the enriched NK T-cell subgroup and the decreased NK T-cell subgroup; (G,H) the OS curves in the enriched Th2 cell subgroup and the decreased Th2 cell subgroup. OS, overall survival; CD, cluster of differentiation; NK, natural killer; Th, T-helper.

Discussion

HOXC10, which is known to regulate the rib formation of mammals (35,36), is a key regulator in the development, proliferation, invasion, and metastasis of cancer cells in lung cancer (19), glioma (20), gastric cancer (37), thyroid cancer (23), and squamous cell carcinomas (22). In the present study, we examined the abnormal expression and predictive effect of HOXC10 in different types of cancers, especially in different subtypes of breast cancer. We also evaluated the correlations between HOCX10 and cancer immune infiltrates to clarify whether HOXC10 affected the prognosis of breast cancer patients.

To examine the effects of HOXC10 on different types of cancer, we used 2 independent online databases (i.e., Oncomine and TIMER) to analyze the differential expression of HOXC10 in tumor and adjacent normal tissues. The Oncomine analysis showed that the expression levels of HOXC10 were significantly higher in brain and CNS, breast, cervical, colorectal, esophageal, gastric, lung, and sarcoma cancer tissues than normal tissues (see ). Conversely, the expression of HOXC10 was observed to be lower in kidney, melanoma, and ovarian cancer tissues than normal tissues in some data sets. The results of the TIMER database analysis indicated that HOXC10 expression was significantly higher in BRCA, ESCA, GBM, HNSC, LUAD, LUSC, STAD, and THCA tissues than adjacent normal tissues, while HOXC10 expression was significantly lower in KICH and PRAD tissues than adjacent normal tissues (see ). To explore the systematic prognostic landscape of HOXC10 in patients with different types of cancer, we used the PrognoScan, GEPIA, and Kaplan-Meier plotter databases to compare the different survival times of patients with high or low expression levels of HOXC10. Our results suggested that HOXC10 may play opposite roles depending on the cancer cell type. Similarly, the overexpression of HOXC10, as a proto-oncogene, has been found to be correlated with the poor prognosis of patients with esophageal squamous cell carcinoma (ESCC) (38), and gastric cancer (39). We found that the higher expression of HOCX10 was related to better RFS in ovarian cancer patients (P=0.038; see Figure S2B).

In breast cancer research, it has been shown that HOXC10 is highly expressed in breast cancer patients (18,40). Similarly, all the analyses of the online Oncomine, TIMER, PrognoScan, GEPIA and Kaplan-Meier plotter databases indicated that HOXC10 acts as an oncogene in breast cancer, and the higher expression of this marker is related to a poor prognosis (see ). In drug-resistance research, HOXC10 has been reported to contribute to the chemotherapy resistance of breast cancer via DNA repair by binding to cyclin-dependent kinase 7 and activating the nuclear factor kappa B (NF-κB) pathway (40). Moreover, HOXC10, the promoter of which contains the ER response element, was also involved in the aromatase inhibitor-resistance of ER positive breast cancer via the regulation of ERα and ERβ (41). The activity of HOXC10 is upregulated through estrogen by recruiting the mixed-lineage leukemia (MLL) 3 and MLL4 to the estrogen response factor in the promoter region of HOXC10 (18). In addition, it has been confirmed that the high expression of the HOXC10 protein is related to the poor prognosis of ER-negative breast cancer patients (40). However, the specific role of HOXC10 in BLBC remains poorly understood. The Kaplan-Meier plotter database was analyzed by filtering the intrinsic subtype of breast cancer to evaluate the relationship between HOXC10 expression and the outcomes of patients with BLBC. The median OS of BLBC patients with high HOXC10 expression was significantly shorter than that of those with low expression (HR =1.97%, 95% CI: 1–3.87; P=0.045; see ). The prognostic effect of HOXC10 was also shown in relation to the PPS of BLBC patients (HR =2.14%, 95% CI: 1.02–4.47, P=0.04; see ), which indicates that HOXC10 could serve as a biomarker for predicting the prognosis of BLBC patients.

In our results, the analysis of the Kaplan-Meier plotter database based on different intrinsic subtypes of breast cancer showed that the elevated expression of HOXC10 had adverse effects on both OS and PPS in lymph-node positive breast cancer patients (P=0.0206 and P=0.0072, respectively), which indicates that tumor infiltration might be related to the prognosis of breast cancer. Some researchers have compared the expression levels of HOXC10 between lymph-node positive and lymph-node negative breast cancer and found that HOXC10 expression was upregulated in patients with positive lymph nodes (42). The related prognosis and possible immune mechanisms of HOXC10 in BRCA, especially in BLBC, are still ambiguous. Thus, we investigated whether HOXC10 was positively correlated to the immune infiltration of macrophages in BLBC (see ). The C3AR1, CD86 and CSF1R of monocytes, the CD68 and IL10 of TAMs, the CD163, VSIG4 and MS4A4A of M2 macrophages, and the IRF5 of M1 macrophages were significantly correlated with the expression levels of HOXC10 in BLBC (see ). Thus, HOXC10 appears to be closely related to macrophage polarization in BLBC, which further confirms that expressions in BLBC are correlated to immune infiltration. Recent studies have examined the possible mechanisms by which HOXC10 is related to the immune infiltration and the poor prognosis of patients with different types of cancer. In glioma cells, HOXC10 was found to induce the expression of immunosuppressive genes, including transforming growth factor-β (TGF-β) programmed death ligand 2, CCL2, and tryptophan 2,3-dioxygenase (20). When the expression of HOXC10 in gastric cancer cells was overexpressed by transfected overexpression plasmids in vitro and in vivo, the expression of interleukin-6, tumor necrosis factor-α (TNF-α), TGF-β, and the epidermal growth factor related to the tumor microenvironment were increased (43). In hepatocellular carcinoma, interleukin-1β was found to induce the expression of HOXC10, which in turn promoted metastasis by upregulating 3-phosphoinositide dependent protein kinase-1 and vasodilator-stimulated phosphoprotein expression (44). Li et al. found that HOXC10 interacted with its antisense transcript partner lncHOXC-AS3 to regulate the osteogenesis of MES stromal cells derived from the bone marrow of patients with multiple myeloma (45). In oral squamous cell carcinoma, HOXC10 was found to induce migration and invasion by regulating the drosophila wingless gene and mouse DNA integration site-1 gene family (Wnt)/epithelial-MES transition signaling pathway (46). Thus, interactions between HOXC10 and the tumor microenvironment could be related to correlations between HOXC10 expression and immune infiltration and the poor prognosis of BLBC patients.

In summary, increased HOXC10 expression was found to be correlated with poor prognosis and increased immune-infiltration levels of macrophages in BLBC. HOXC10 expression may contribute to the regulation of monocytes, TAMs, M1 macrophages, and M2 macrophages. Thus, HOXC10 appears to play an important role in immune-cell infiltration and could serve as a prognosis biomarker in patients with BLBC.

The article’s supplementary files as