Combination of CD47 and CD68 expression predicts survival in eastern-Asian patients with non-small cell lung cancer

Abstract

Objective

Recent studies have indicated that CD47, interacting with SIRP-α, conveys “don’t eat me” signal in evasion of tumor cells and serves as a potential target for cancer immunotherapy. The purpose of this study was to investigate the clinical correlation of CD47 and uncover prognostic implications of CD47 and CD68 in non-small cell lung cancer (NSCLC).

Methods

The specimens from 384 patients with completely resected NSCLC were collected for immunohistochemical assays of CD47 and CD68. Cox multivariate proportion hazard analyses were conducted to confirm the independent prognostic value of CD47 and CD68. TCGA database and GSE37745 were used to identify the association between CD47 and immune cells.

Results

In 186 pairs of lung cancer and adjacent tissues, the RNA of CD47 was overexpressed in lung cancer tissues (P < 0.001). High expression of CD47 was associated with worse recurrence-free survival in RNA and protein level (P = 0.032 and P < 0.001, respectively). High expression of CD47 was significantly associated with large tumor size (P = 0.004), advanced pathologic TNM stage (P < 0.001), and histology (P = 0.003). Further analyses demonstrated that CD47 and CD68 predicted outcomes of patients independently. In addition, the expression of CD47 correlated with neutrophils, and did not correlated with B cells and CD4 + T cells in the TCGA database and GSE37745.

Conclusion

Combined use of CD47 and CD68 exhibited excellent performance in predicting survival of patients with NSCLC. CD47 was a potential therapeutic target for immune therapy of lung cancer.

Electronic supplementary material

The online version of this article (10.1007/s00432-020-03477-3) contains supplementary material, which is available to authorized users.

Introduction

Despite high morbidity and mortality of malignancies, immune-checkpoint inhibitors have significantly improved survival of patients, suggesting that activating immune system helps to kill tumor cells. As a major component of the body’s immune system, the innate immune system, associated with the clearance of cancer cells through phagocytosis, serves as the first defense against infections and malignancies (Iwasaki and Medzhitov 2010). Macrophages, which were characterized by CD68, play an important role in the innate immune system (Biswas and Mantovani 2010). Macrophages could be divided into two distinct types (M1 and M2). To our knowledge, M1 macrophages were correlated with less tumor invasiveness and good survival, while M2 macrophages were associated with tumor growth and poor survival (Mony and Schuchert 2018).

CD47, interacting with SIRP-α, conveys “don’t eat me” signal in the evasion of tumor cells from macrophages (Liu et al. 2019) and serves as a target for cancer immunotherapy (Logtenberg et al. 2019). CD47 is overexpressed in solid tumors and acts as an oncogene in previous studies (Chan et al. 2009; Edris et al. 2012; Zhao et al. 2016). Moreover, high expression of CD47 is associated with worse survival in several kinds of tumors (Chao et al. 2010; Nagahara et al. 2010; Suzuki et al. 2012). However, there are few studies focusing on the prognostic role of CD47 in lung cancer and its association with CD68, particularly in eastern-Asian patients.

In this study, we evaluated the expression of CD47 and CD68 in specimens from 384 patients with non-small cell lung cancer (NSCLC) by immunohistochemical assays. The results provided extra evidences for the anti-CD47 immune therapy of lung cancer.

Materials and methods

Tissue samples

The cohort to evaluate CD47 RNA, previously described in detail (Chen et al. 2019), consisted of 186 lung adenocarcinoma (LUAD) cases. The cohort to perform immunohistochemical (IHC) assay consisted of 384 NSCLC cases from Fudan University Shanghai Cancer Center diagnosed between 2009 and 2015. Patients with neoadjuvant treatment were excluded. Clinicopathologic parameters are showed in Table 1. The study was approved by the Institutional Review Board of FUSCC (IRB#090,977–1). Informed consents of patients were waived due to the retrospective study design.

Table 1.

Correlation between CD47 expression and clinicopathologic characteristics of patients with non-small cell lung cancer

Variables	CD47			P values
	All cases (N = 384)	Low expression (N = 197)	High expression (N = 187)
Age (years)				0.601
Median (IQR)	61 (55–67)	61 (55–67)	61 (55–68)
Mean ± SD	60.54 ± 9.88	60.28 ± 10.46	60.81 ± 9.24
Range	26–83	26–83	36–80
Gender				0.315
Male	220 (57.3)	108 (54.8)	112 (59.9)
Female	164 (42.7)	89 (45.2)	75 (40.1)
Smoking history				0.982
Ever	162 (42.2)	83 (42.1)	79 (42.2)
Never	222 (57.8)	114 (57.9)	108 (57.8)
Pathological size, cm				0.004
Mean ± SD	3.28 ± 1.98	3.00 ± 1.69	3.58 ± 2.21
p-TNM stage				< 0.001
I	203 (52.9)	121 (61.4)	82 (43.9)
II	69 (18.0)	39 (19.8)	30 (16.0)
III	112 (29.2)	37 (18.8)	75 (40.1)
Histology				0.003
MIA	9 (2.3)	9 (4.6)	0 (0)
LPA	25 (6.5)	18 (9.1)	7 (3.7)
APA	146 (38.0)	65 (33.0)	81 (43.3)
PPA	32 (8.3)	16 (8.1)	16 (8.6)
IMA	12 (3.1)	10 (5.1)	2 (1.1)
MPA	5 (1.3)	2 (1.0)	3 (1.6)
SPA	37 (9.6)	16 (8.1)	21 (11.2)
SQCC	118 (30.7)	61 (31.0)	57 (30.5)
Mutational status				0.160
EGFR( +)/KRAS(-)	215 (56.0)	101 (51.3)	114 (61.0)
KRAS( +)/EGFR(-)	23 (6.0)	13 (6.6)	10 (5.3)
EGFR(−)/KRAS(−)	146 (38.0)	83 (42.1)	63 (33.7)

The P values in bold indicated statistical significance (P < 0.05)

IQR interquartile range, SD standard deviation, LPA lepidic pattern-predominant adenocarcinoma, MIA minimally invasive adenocarcinoma APA acinar pattern-predominant adenocarcinoma, PPA papillary pattern-predominant adenocarcinoma, IMA invasive mucinous adenocarcinoma; MPA, micropapillary pattern-predominant adenocarcinoma SPA solid pattern-predominant adenocarcinoma, SQCC squamous cell carcinoma

Immunohistochemistry

Formalin-fixed paraffin-embedded primary specimens of NSCLC were obtained to construct 4-μm-thick sections. IHC assays for CD47 and CD68 were carried out using IHC kits (KIHC-5, Proteintech, USA) in accordance with the manufacture’s protocol under the following conditions: CD47 (20,305-1-AP, 1:200, Proteintech) and CD68 (D4B9C, #76,437, 1:400, Cell Signaling Technology). The total set of tissue specimens was processed and immunostained in one time to guarantee the objective comparison between samples.

Evaluation of immunohistochemical staining

The images of CD47 IHC staining were evaluated using the immunoreactivity score (IRS) (Kahlmeyer et al. 2019; Lindner et al. 2015). The intensity of staining was defined in four categories (no staining = 0, weak staining = 1, moderate staining = 2, and strong staining = 3) and multiplied by the proportion of positive cells (no = 0, < 10% = 1, 10%-50% = 2, 51%-80% = 3, and > 80% = 4). The images of CD68 IHC staining were assessed as the percentage of CD68-positive cells to all visible cells. The cut-offs for the expression of CD47 and CD68 were determined by their median values. The cutoff for IRS of CD47 was 3, and that for CD68 was 5%. Two independent persons, blinded to the outcomes of patients, reviewed pathologic images. For cases with different results, a consensus was reached through discussion.

Statistical analysis

Data were analyzed using IBM SPSS software (version 25.0). Recurrence-free survival (RFS) was defined as the time from date of diagnosis to date of recurrence attributed to lung cancer. Overall survival (OS) was defined as the time from date of diagnosis to data of death due to any causes. Paired Student t tests were conducted to compare CD47 expression in 186 pairs of NSCLC cases. Kaplan–Meier analyses, log-rank tests, and Cox proportion hazard regression were applied to perform survival analyses. Multivariable Cox analyses were used to identify independent factors related to survival. Spearman’s tests were carried out to evaluate the correlation between CD47 and CD68 or the infiltrations of immune cells. The analyses of immune cells infiltration were conducted by TIMER (https://cistrome.shinyapps.io/timer/) in the TCGA database and GSE37745 (Botling et al. 2013; Li et al. 2017). All tests were two-tailed, and statistical significance was set at P < 0.05.

Results

High expression of CD47 is found in lung cancer tissues and associated with worse survival

To investigate CD47 expression in tumor tissues, we compared the RNA expression of CD47 in 186 pairs of lung cancers and adjacent tissues from our previous study (Chen et al. 2019). The results demonstrated that CD47 was overexpressed in lung cancer tissues (P < 0.001, Fig. 1a). Among 95 patients with invasive adenocarcinoma, high expression of CD47 RNA was associated with worse RFS (P = 0.032, Fig. 1b).

Fig. 1.

The expression of CD47 in primary lung cancer. a The relative mRNA expression level of CD47 in 186 pairs of primary lung cancer and its adjacent tissues. b Kaplan–Meier Curve of recurrence-free survival according to mRNA expression of CD47 in patients with lung adenocarcinoma

Next, we evaluate the protein level of CD47 and CD68 in completely resected tissues using immunochemical assays. A total of 384 NSCLC patients, undergoing surgical resection in Fudan University Shanghai Cancer Center, were enrolled in the study. The represent IHC images of CD47 and CD68 are shown in Fig. 2. CD47 protein was mainly localized in the membrane of cancer cells, whereas CD68 in the membrane and cytoplasm of macrophages. Further analyses demonstrated that the high expression of CD47 was associated with worse RFS and OS (RFS, P < 0.001, Fig. 3a; OS, P = 0.009, Fig. 3b).

Fig. 2.

Representative images of immunohistochemical staining for high and low expressions of CD47 and CD68 in lung cancer specimens. The cutoffs for the expression of CD47 and CD68 were determined by their median values

Fig. 3.

CD47 and CD68 were adverse predictor in non-small cell lung cancer (NSCLC). a–b Kaplan–Meier Curve of recurrence-free survival (a) and overall survival (b) according to protein expression of CD47 in patients with NSCLC. c–d Kaplan–Meier Curve of recurrence-free survival (c) and overall survival (d) according to protein expression of CD47 and CD68 in patients with NSCLC. The cut-offs for the expression of CD47 and CD68 were determined by their median values

Patients characteristics according to CD47 expression

Next, we investigated baseline of patients stratified by CD47 expression. The results are showed in Table 1. Among 384 enrolled patients, 187 (187/384, 48.7%) showed high expression of CD47, and 197 (197/384, 51.3%) showed low expression. High expression of CD47 was significantly associated with large tumor size (P = 0.004), advanced pathologic TNM stage (P < 0.001), and histology (P = 0.003). Additionally, there were no differences in age (P = 0.601), gender (P = 0.315), smoking history (P = 0.982), or mutational status (P = 0.160) between the CD47^high and CD47^low groups.

CD47 and CD68 predict survival of patients independently

Previous studies reported that CD47, the immune-suppressive marker expressed in the membrane of tumor cell, has a key role in tumor-mediated immune escape by interacting with tumor-associated macrophages (TAMs) characterized by CD68 (Feng et al. 2019), suggesting that CD47 expression might affect the amount of TAMs. Therefore, we explored the relationship between CD47 and CD68 by Spearman correlation analyses in the TCGA database and our cohort. The results demonstrated that CD68 was positively correlated with CD68 in both the TCGA database (r = 0.18, P = 2.3E-8, Supplementary Fig. 1a) and our cohort (r = 0.2642, P < 0.001, Supplementary Fig. 1b). Additionally, we investigated the combined use of CD47 and CD68 as prognostic factors. The survival analyses revealed that the RFS and OS in the CD47^low CD68^low group were the best, those in the CD47^high CD68^high group were the worst, and those in the CD47^low CD68^high and CD47^high CD68^low groups were intermediate (RFS, P < 0.001, Fig. 3c; OS, P < 0.001, Fig. 3d). Furthermore, Cox proportion analyses demonstrated that the expressions of CD47 and CD68 independently predicted RFS [CD47, hazard ration (HR) 2.090, 95% confident interval (CI) 1.475–2.962, P < 0.001; CD68, HR 2.550, 95% CI 1.847–3.521, P < 0.001) for operable NSCLC (Table 2).

Table 2.

Univariate and multivariable Cox regression analyses of factors associated with recurrence-free survival (N = 384)

Variables	Univariate		Multivariate
	HR (95% CI)	P	HR (95% CI)	P
Age	1.012 (0.996, 1.029)	0.128
Sex		0.122
Female	Reference
Male	1.282 (0.935, 1.758)
Smoking status		0.088		0.337
Never	Reference		Reference
Ever	1.319 (0.959, 1.814)		1.179 (0.842, 1.651)
Histology		0.358
Adenocarcinoma	Reference
Squamous cell carcinoma	0.836 (0.571, 1.225)
p-TNM stage		< 0.001		< 0.001
I	Reference		Reference
II	2.585 (1.635, 4.086)	< 0.001	2.663 (1.649, 4.301)	< 0.001
III	5.804 (4.041, 8.337)	< 0.001	4.703 (3.241, 6.825)	< 0.001
CD47		< 0.001		< 0.001
Low expression	Reference		Reference
High expression	2.980 (2.124, 4.181)		2.090 (1.475, 2.962)
CD68		< 0.001		< 0.001
Low expression	Reference		Reference
High expression	2.763 (2.012, 3.793)		2.550 (1.847, 3.521)

The P values in bold indicated statistical significance (P < 0.05)

Then, we conducted subgroup analyses to investigate whether mutational status could influence the prognostic effect of CD47. Among patients with EGFR mutation, CD47 (HR 2.625, 95% CI 1.691–4.076, P < 0.001) and CD68 (HR 2.222, 95% CI 1.492–3.309, P < 0.001) were associated with RFS independently (Table 3). Among patients with KRAS mutation, there is no prognostic effect of CD47 or CD68. Among patients without EGFR and KRAS mutation, CD47 (HR 3.139, 95% CI 1.680–5.866, P < 0.001) and CD68 (HR 3.638, 95% CI 1.962–6.746, P < 0.001) were also prognostic factors (Table 3).

Table 3.

Cox regression analysis of CD47 and CD68 on recurrence-free survival in patients stratified by mutational status

Variables	Cox regression
	HR (95% CI)	P
EGFR(+)/KRAS(-)
CD47		< 0.001
Low expression	Reference
High expression	2.625 (1.691, 4.076)
CD68		< 0.001
Low expression	Reference
High expression	2.222 (1.492, 3.309)
KRAS( +)/EGFR(−)
CD47		0.115
Low expression	Reference
High expression	3.243 (0.752, 13.989)
CD68		0.370
Low expression	Reference
High expression	1.840 (0.485, 6.979)
EGFR(-)/KRAS(-)
CD47		< 0.001
Low expression	Reference
High expression	3.139 (1.680, 5.866)
CD68		< 0.001
Low expression	Reference
High expression	3.638 (1.962, 6.746)

The P values in bold indicated statistical significance (P < 0.05)

Correlation with between CD47 and immune cells

Since CD47 played a significant role in immune evasion of tumor cells (Liu et al. 2019; Logtenberg et al. 2019), we further used the TIMER (Li et al. 2017) to investigate the relationship between CD47 and immune infiltration in LUAD and lung squamous carcinoma (LUSC) from the TCGA database, and GSE37745 was used as a validation dataset (Botling et al. 2013). The results showed that the expression of CD47 in LUAD and LUSC was significantly correlated with the infiltration level of neutrophils in both datasets (Fig. 4 and Supplementary Fig. 2). Moreover, B cells and CD4 + T cells were not associated with CD47 in the two databases, supporting the conclusion that CD47 was a key factor in innate immunity rather than adaptive immunity.

Fig. 4.

The correlation between CD47 and infiltration of immune cells from innate (a) and adaptive (b) immune system in the TCGA database

Discussion

Immune-checkpoint inhibitors (ICIs) have been a major breakthrough in past decades. However, only ICIs (PD-1, PD-L1, or CTLA-4) targeting the lymphoid lineage, modulating the adaptive T-cell response, are currently used in clinical practice. CD47 is an emerging checkpoint for myeloid cells and the innate immune system, and serves as a potential target for cancer immune therapy. In several malignancies, the results of anti-CD47 therapy were promising (Chao et al. 2010, 2011; Lee et al. 2014). Nevertheless, in lung cancer, therapies of monoclonal antibodies targeting CD47 are still at early stage of developments. In this study, we reported that the high expression of CD47 was upregulated in lung cancer tissues. To our knowledge, for the first time, the study revealed the combine use of CD47 and CD68 could predict survival in eastern-Asian patients with NSCLC and the expression of CD47 correlated with the infiltration of cells from the innate immune system.

CD47 has been considered as an adverse biomarker for survival in several cancers (Zhao et al. 2018). Animal experiments revealed that CD47 blockade could strengthen the effect of anti-angiogenic therapy on NSCLC by enhancing macrophage infiltration and tumor cell destruction (Zhang et al. 2019). Furthermore, anti-CD47 siRNA could even suppress the growth of lung cancer cells in vitro (Wu et al. 2018). However, there are few papers studying the exact role of CD47 in the prognosis of lung cancer. Barrera et al. found that CD47 overexpression in white blood cells was associated with poor survival in patients with NSCLC (Barrera et al. 2017). Liu et al. showed that the mRNA level of CD47 could predict the survival of NSCLC in eastern Asians (Liu et al. 2017). Nevertheless, Arrieta et al. investigated the expression of CD47 in Mexicans and demonstrated that the prognostic role of CD47 was only shown in patients with EGFR mutations rather than the whole population. In our study, we performed immunohistochemical staining in primary NSCLC tissues from 384 Chinese patients and found CD47 predicted RFS and OS in protein level. The above studies indicated that the prognostic value of CD47 might be more apparent in eastern Asians. Moreover, the high expression of CD47 was associated with large pathological size and advanced TNM stage, implicating that CD47 played an important role in development and progression of lung cancer. These results suggested that anti-CD47 therapy may be more effective in advanced NSCLC.

In our study, we found CD47 did not have prognostic value in KRAS( +) patients, while it did in KRAS(−) patients. Ma et al. demonstrated patients with KRAS mutation had worse survival compared with those without (Ma et al. 2020). Perhaps, KRAS( +) patients had unfavorable outcomes, which covered the prognostic effect of CD47.

TAMs were a key part of microenvironment of cancer which affected tumor progression and metastasis (Karnevi et al. 2014; Mantovani et al. 2017). In general, high density of TAMs (characterized by CD68) indicated poor survival in lung cancer (Zhang et al. 2012). In this study, we found the prognostic value of CD68 was independent of CD47. The high expression of CD68 and CD47 simultaneously indicated the worst survival of patients with NSCLC. Furthermore, CD68 positively correlated CD47 in the TCGA database, implicating the close connection of CD47 and macrophages.

The immune cells played a crucial role in tumor-immune network and evasion of tumor cell (Bindea et al. 2013; Mlecnik et al. 2010). In fact, if we better understand the immune infiltration status, we could choose the patients who were most likely to benefit. In this study, we found that CD47 was closely correlated with neutrophils and not correlated with B cells and CD4 + T cells in LUAD and LUSC, supporting the conclusion that CD47 was a key factor in innate immunity rather than adaptive immunity.

It need be addressed that there are some limitations in the study. First, the study is based on the specimens from one institution, which may cause selection bias. Future validations from other centers are needed. Second, in this study, we analyzed the infiltration of immune cells only by bioinformatic methods based on the data from TCGA database rather than IHC assays. However, previous studies have demonstrated that data from RNA-seq could reliably predict immune cell infiltration (Aran et al. 2017; Gentles et al. 2015; Newman et al. 2019). Therefore, we believe that it is acceptable to use RNA-seq to evaluate the infiltration of immune cells.

In conclusion, our study showed that the high expression of CD47 was associated with poor survival in NSCLC. The combination of CD47 and CD68 could predict outcomes of NSCLC patients. The expression of CD47 correlated with neutrophils and did not correlate with B cells and CD8 + T cells, and CD47 was a therapeutic target for immune therapy of lung cancer.

Electronic supplementary material

Supplementary Figure 1. The association between CD47 and CD68 in the TCGA database (A) and our cohort (B). Spearman’s tests were carried out to investigate the correlation between CD47 and CD68.

Supplementary Figure 2. The correlation between CD47 and infiltration of immune cells from innate (A) and adaptive (B) immune system in GSE37745.

Below is the link to the electronic supplementary material.

Abbreviations

CI

Confidential interval

HR

Hazard ratio

ICIs

Immune-checkpoint inhibitors

IHC

Immunohistochemistry

IRS

Immunoreactivity score

LUAD

Lung adenocarcinoma

LUSC

Lung squamous carcinoma

NSCLC

Non-small-cell lung cancer

OS

Overall survival

RFS

Recurrence-free survival

TAMs

Tumor-associated macrophages

Funding

This work was supported by the National Natural Science Foundation of China (81930073 and 81772466), Shanghai Municipal Science and Technology Major Project (Grant No. 2017SHZDZX01, VBH1323001/026), Shanghai Municipal Key Clinical Specialty Project (SHSLCZDZK02104), and Pilot Project of Fudan University (IDF159034).

Compliance with ethical standards

Conflict of interest

None.