Prognostic value of CD117 in cancer: a meta-analysis

Abstract

Background: The prognostic value of CD117 expression in cancers has been evaluated for several years while the results remain controversial. We thus performed a systematic review and meta-analysis of studies assessing the impact of CD117 expression on overall survival (OS) and disease-free survival (DFS) to clarify this issue. Methods: We searched Pubmed, Embase and Web of Science to identify studies on the prognostic impact of CD117 expression in cancers. A total of 4,458 patients from 39 eligible studies were included in the analysis. Pooled risk ratios (RRs) with 95% confidence interval (95% CI) were calculated to estimate the effect using random-effects model. Results: The analysis indicated that CD117 had significant association with poor OS of osteosarcoma (OR=1.36, 95% CI=1.03-1.79, I2=0%, fixed model) and renal carcinoma (OR=4.86, 95% CI=2.72-8.67, I2=0%, fixed model).However, no significant association between CD117 and DFS was found in overall studies. Conclusions: CD117 expression might be a predictive factor of poor prognosis in some surgically treated cancers, particularly in renal carcinoma.

Introduction

Despite the recent reduction in incidence and mortality, cancer is still a worldwide health burden and leads to more deaths than heart disease in some regions [1]. Surgical resection can be performed to remove the tumor if neither lymph node nor distant metastasis were present, while recurrence rate after surgery remains high [2]. Moreover, numerous cancers at the time of diagnosis are at advanced stage and the treatment options are limited, resulting in the persistent high mortality of cancers.

A lot of efforts have been made to investigate the prognostic biomarkers in cancers, helping to identify high-risk cancer patients who might need adjuvant treatment after surgery.

CD117, encoded by the proto-oncogene c-kit, is a transmembrane protein belonging to the type III subfamily of the receptor tyrosine kinases [3]. It has extracellular, intramembranous and intracellular domains. By binding to its ligand, called stem cell factor (SCF), this molecule plays an important part in regulating cellular activities, such as apoptosis, cell differentiation, proliferation, and cell adhesion [4].

Although the impact of CD117 expression on prognosis of patients with cancer has been explored recently, the prognostic value of CD117 expression in different tumor types remains conflicting because heterogeneous results were reported in studies and some of them included a small number of patients. To elucidate this issue, we performed this systematic review and meta-analysis to assess the prognostic significance of CD117 expression in various types of cancer.

Materials and methods

Publication search

We searched Pubmed, Embase and Web of Science to identify studies that assessed the prognostic value of CD117 expression in patients with carcinomas who underwent surgical resection of a tumor. The search strategy was the following terms: “CD117”, “c-kit”, “cancer”, “carcinoma”, “prognosis”, “prognostic”, and “survival”. The search ended in August, 2013, and no lower date limit was applied. References cited in selected articles were also searched manually to identify other relevant studies. Although our search did not have language limits initially, for the full-text reading and final evaluation we only performed the review of the studies published in English language. Conference abstracts were not selected for our analysis due to the insufficient data reported in them. Letters to the editor, reviews, and articles published in a book or papers were excluded. Criteria that an eligible study has to meet were as follows: (a) to evaluate the relationship between CD117 and overall survival (OS) or disease-free survival (DFS) of patients with carcinoma; (b) to assess CD117 expression using immunohistochemistry (IHC); (c) to determine CD117 expression in surgically resected primary tumor tissues (not in normal tissues or in body fluids such as blood and sputum). Two reviewers independently judged if studies screened were eligible. Disagreements were resolved by discussion. If the results reported in identified studies have the possible overlap (e.g., same authors, institutions), only the most informative study was involved in the analysis. The following information was extracted from each publication and used as a supplement if available: author, publication year, patient’s country, tumor stage, number of patients, research technique used, definition of positivity (cutoff value), and survival data. If data from any of the above categories were not reported in the primary study, items were treated as “not document”. A lower limit of number of patients included in each study was not set for inclusion in the meta-analysis.

Statistical analysis

Included studies were divided into two groups for analysis: those with data regarding OS and those regarding DFS. For the quantitative aggregation of the survival results, we measured the impact of CD117 expression on survival by risk ratio (RR) between the two survival distributions. RRs and 95% confidence intervals (CIs) were used to combine as the effective value. For those RRs that were not given directly in the published articles, the published data and figures from original papers were used to assess the RR according to the methods described by Parmar et al [5]. The heterogeneity assumption was calculated by the Q-test and P-values greater than 0.05 indicated a lack of heterogeneity among studies, so the RR was calculated by a fixed-effects model (the Mantel-Haenszel method and chi-squared tests). Otherwise, a random-effects model (the DerSimonian-Laird method) was used. Evidence of publication bias was sought using the methods of Egger et al. [6] and of Begg et al [7]. Intercept significance was determined by the t test suggested by Egger (P<0.05 was considered representative of statistically significant publication bias). Kaplan-Meier curves were read by GetData Graph Digitizer 2.24. All the statistical analyses were performed using Stata 12.0 for Windows (Stata Corporation, College Station, TX, USA).

Results

Characteristics of the studies

Using the search criteria (Figure 1), a total of 39 publications met the criteria for this analysis [8-46]. Of them, eight were on lung cancer, two were on colorectal cancer, three were on ovarian cancer, three were on malignant melanoma, two were on uterine sarcomas, two were on Merkel cell carcinoma two were on osteosarcoma, two were on renal cancer, one was on endometrial adenocarcinoma, one was on cervical cancer, one was on vulvar cancer, one was on breast cancer, one was on adenoid cystic carcinoma, one was on germ cell tumors, one was on extrapulmonary small cell carcinoma, one was on prostate cancer, one was on thymic epithelial tumors, and one was on acute myeloid leukemia. The total number of patients included was 4,458 ranging from 26 to 522 patients per study. Among the 39 studies, seven studies were performed in Asian populations, and the remaining 32 studies followed non-Asian patients. All patients in the eligible studies were determined by pathological stage. All of the studies reported the prognostic value of CD117 status for survival in patients with cancer. Of the 39 studies, 6 directly reported survival data, while the other thirty-three studies provided survival curves. Estimation using survival curves were segregated according to either DFS or OS. A RR on DFS or OS could be extracted for all enrolled studies. Eleven of the 39 studies identified CD117 expression as an indicator of poor DFS or OS, three studies demonstrated that CD117 was a favorable prognostic factor, and the other twenty-five studies showed no statistically significant impact of CD117 overexpression on DFS or OS. Of the 39 studies, 2 studies detected the CD117 expression by flow cytometry, other 37 studies performed by immunohistochemistry. The characteristics of the included studies are listed in Table 1.

Figure 1.

Flow chart of study inclusion.

Table 1.

Main characteristic of the included studies

Study	Patient’s country	Year	Tumor type	Tumor stage (UICC)	Technique	Location	Number of patients	Cut off (IHC)	RR (95% CI)
Rehfeld	Germany	2006	Small cell lung cancer	I-IV	IHC	Cytoplasm	195	≥1%	OS 1.03 (0.91-1.17)
Camps	Spain	2006	Small cell lung cancer	I-IV	IHC	ND	85	>1%	OS 0.89 (0.78-1.00)
									DFS 1.00 (0.88-11.14)
Erler	USA	2010	Small cell lung cancer	I-IV	IHC	Cytoplasm	97	>25%	OS 1.00 (0.94-1.07)
L’opez-Martin	Spain	2007	Small cell lung cancer	ND	IHC	membrane	204	>1%	OS 0.96 (0.88-1.06)
									DFS 0.96 (0.88-1.04)
Potti	USA	2003	Small cell lung cancer	ND	IHC	ND	193	>10%	OS 0.96 (0.90-1.02)
Rohr	Germany	2004	Small cell lung cancer	I-IV	IHC	ND	203	>25%	OS 0.98 (0.92-1.04)
Rossi	Italy	2003	Small cell lung cancer	ND	IHC	Cytoplasm and membrane	27	>25%	OS 1.14 (0.72-1.80)
Pelosi	Italy	2004	Lung cancer	I-III	IHC	Cytoplasm	66	>5%	OS 3.00 (0.54-16.77)
Scobie	USA	2003	Endometrial adenocarcinomas	I-IV	IHC	ND	72	>10%	DFS 1.90 (1.04-3.49)
Erdogan	Turkey	2007	Ovarian cancer	I-IV	IHC	Cytoplasm	68	>10%	OS 1.12 (0.44-2.85)
Tonary	Canada	2000	Ovarian cancer	I-IV	IHC	Cytoplasm and membrane	50	ND	OS 0.49 (0.30-0.82)
									DFS 0.71 (0.52-0.96)
Lassus	Finland	2004	Ovarian cancer	I-IV	IHC	Cytoplasm and membrane	522	>1%	OS 1.34 (1.11-1.61)
Huh	USA	2010	Uterine sarcomas	II-IV	IHC	ND	26	>25%	OS 1.00 (0.80-1.26)
									DFS 1.00 (0.80-1.26)
Winter	USA	2003	cancer of uterine corpus	I-IV	IHC	ND	39	>30%	OS 0.40 (0.18-0.92)
									DFS 0.30 (0.09-1.02)
Sukpan	Thiland	2011	Cervical cancer	I-IV	IHC	Cytoplasm	100	>5%	DFS 0.90 (0.34-2.38)
de Melo Maia1	Brazil	2012	Vulvar cancer	ND	IHC	Cytoplasm or membrane	139	≥score 2	OS 0.31 (0.16-0.59)
									DFS 1.04 (0.46-2.34)
Tanaka	Japan	2006	Gallbladder cancer	I-IV	IHC	Cytoplasm	47	>20%	OS 0.93 (0.49-1.77)
Fan	China	2013	Oesophageal cancer	I-IV	IHC	Cytoplasm and membrane	157	>1%	OS 1.76 (1.08-2.85)
									DFS 2.16 (1.18-3.96)
Friederichs	Germany	2010	Colorectal cancer	I-IV	IHC	Cytoplasm	263	>1%	OS 1.48 (1.04-2.11)
Medinger	Germany	2010	Colorectal carcinoma (1); Breast cancer (2)	ND	IHC	Cytoplasm or membrane	282	>1%	OS (1) 1.46 (0.94-2.28)
									OS (2) 1.17 (0.94-1.45)
Murakami	Japan	2011	Malignant melanoma	ND	IHC	Cytoplasm	39	>10%	OS 0.78 (0.56-1.08)
Potti	USA	2004	Malignant melanoma	ND	IHC	ND	202	>10%	OS 0.57 (0.07-4.58)
Bataille	France	2008	Malignant melanoma	I-IV	FC	/	62	≥10%	OS 2.62 (1.74-3.95)
Wei	China	2008	Osteosarcoma	ND	IHC	Cytoplasm	40	>1%	OS 1.24 (0.99-1.56)
Miiji	Brazil	2011	Osteosarcoma	ND	IHC	Cytoplasm	52	>10%	OS 1.69 (0.75-3.81)
Waltari	Finland	2010	Merkel cell carcinoma	I-IV	IHC	Cytoplasm	207	≥score 1	OS 1.18 (0.88-1.58)
									DFS 1.26 (0.93-1.71)
Aleodor	UK	2010	Merkel cell carcinoma	I-IV	IHC	Cytoplasm	40	>30%	OS 1.27 (0.60-2.70)
Schwarz	Germany	2008	Salivary gland carcinomas	I-IV	IHC	membrane	101	>10%	DFS 1.09 (0.68-1.73)
Bar-Sela	Israel	2003	Nasopharyngeal carcinoma	I-IV	IHC	Cytoplasm and membrane	49	≥10%	OS 0.79 (0.34-1.84)
Zhang	China	2009	Renal cancer	I-IV	IHC	membrane	119	>10%	OS 4.08 (2.02-8.22)
Jones	UK	2007	Pediatric renal tumour	II-III	IHC	Cytoplasm	226	>10%	OS 5.80 (2.29-14.67)
									DFS 3.05 (2.09-4.45)
Aslan	USA	2005	Adenoid cystic carcinoma	ND	IHC	Cytoplasm and membrane	45	>25%	DFS 1.27 (0.45-3.60)
Durán	Canada	2010	Germ cell tumors	ND	IHC	Cytoplasm and membrane	84	>1%	DFS 0.94 (0.46-1.94)
Kurt	Turkey	2005	Extrapulmonary small cell carcinoma	ND	IHC	ND	28	>5%	OS 1.00 (0.87-1.15)
Di Lorenzo	Italy	2004	Prostate cancer	ND	IHC	Cytoplasm or membrane	94	>1%	DFS 3.66 (1.28-10.51)
Petrini	Italy	2010	Thymic epithelial tumors	I-IV	IHC	membrane	120	>1%	DFS 4.23 (1.90-9.44)
Sharawat	India	2013	Acute myeloid leukemia	ND	FC	/	115	/	OS 16.50 (3.11-87.47)

Meta-analysis results

The meta-analysis was performed for cancer types on which more than 2 studies were eligible for inclusion. For studies evaluating overall survival (OS), there appeared to have heterogeneity among studies for CD117 expression (I2=80%). Accordingly, a random model was applied to calculate a pooled RR and its 95% CI (Table 2). We found that CD117 significantly predicted poorer OS, with the pooled RR being 1.12 (95% CI: 1.02-1.22, Figure 2A). In the tumor type subgroup analysis, CD117 expression was significantly associated with poor OS in patients with osteosarcoma (OR=1.36, 95% CI=1.03-1.79, I2=0%, fixed model) and renal carcinoma (OR=4.86, 95% CI=2.72-8.67, I2=0%, fixed model) (Table 2).

Table 2.

Meta-analysis of effects of CD117 expression on OS and DFS of different cancer types

Cancer type	Studies	Odds ratio			Heterogeneity
		OR (95% CI)	POR	Model	I2 (%)	P
OS
Total studies	37	1.12 (1.02-1.22)	0.01	Random	80	<0.001
Lung cancer	9	0.99 (0.95-1.04)	0.75	Fixed	11	0.34
Colorectal cancer	2	1.28 (0.95-1.74)	0.11	Random	55	0.13
Ovarian cancer	3	0.91 (0.44-1.88)	0.79	Random	85	0.001
Malignant melanoma	3	1.23 (0.43-3.51)	0.7	Random	91	<0.001
Uterine sarcomas	2	0.66 (0.14-3.80)	0.59	Random	92	0.0003
Merkel cell carcinoma	2	1.20 (0.91-1.58)	0.21	Fixed	0	0.84
Osteosarcoma	2	1.36 (1.03-1.79)	0.03	Fixed	0	0.36
renal cancer	2	4.86 (2.72-8.67)	<0.001	Fixed	0	0.54
DFS
Total studies	17	1.22 (0.98-1.51)	0.08	Random	84	<0.001
Lung cancer	2	0.97 (0.90-1.04)	0.39	Fixed	0	0.59
Uterine sarcomas	2	0.57 (0.05-6.11)	0.64	Random	93	0.0001

Figure 2.

Meta-analysis of the effects of CD117 expression on OS (A) and DFS (B).

For studies evaluating DFS, a random model was also applied due to the heterogeneity among studies (Table 2). The expression of CD117 was not significantly correlated with DFS, with 1.22 (95% CI: 0.98-1.51, Figure 2B) as the pooled RR.

We also performed subgroup analysis by tumor histological type, the region of reports or cut off value in lung cancer. There was still no significant association detected in all stratified analysis (Table 3). Due to the limited number of eligible studies (n=2 to 3) included in the analysis of cancer types other than lung cancer, we did not conduct the subgroup analysis for these carcinomas.

Table 3.

Subgroup analysis of the studies reporting the prognostic value of CD117 expression on OS of lung cancer

Stratified analysis	Studies	Odds ratio			Heterogeneity
		OR (95% CI)	POR	Model	I2 (%)	P
OS
Lung cancer overall	9	0.99 (0.95-1.04)	0.75	Fixed	11	0.34
SCLC	8	0.98 (0.94-1.02)	0.89	Fixed	0	0.5
LCNEC	1	1.38 (0.79-2.38)			55	0.13
Region
Germany	2	1.00 (0.94-1.07)	0.93	Fixed	0	0.34
Spain	2	0.94 (0.87-1.01)	0.1	Fixed	7	0.3
USA	2	0.97 (0.93-1.02)	0.19	Fixed	0	0.32
Italy	3	1.44 (0.97-2.14)	0.07	Fixed	0	0.42
Cut off
>25%	3	1.00 (0.94-1.06)	0.95	Fixed	0	0.68
<25%	6	0.99 (0.94-1.05)	0.75	Fixed	38	0.15
DFS
Lung cancer overall	2	0.97 (0.90-1.04)	0.39	Fixed	0	0.59

SCLC: Small cell lung cancer. LCNEC: large-cell neuroendocrine carcinomas.

Potential publication bias

Publication bias of the included studies was evaluated by funnel plots and Egger’s tests. As shown in Figure 3, the funnel plots were symmetric. The results of Egger’s test showed there was no evidence of publication bias on DFS (P=0.433) and OS (P=0.680).

Figure 3.

Funnel plots for the evaluation of potential publication bias in the impact of CD117 expression on OS (A) and DFS (B). The Egger test for publication bias was not significant for OS (p=0.433) and DFS (p=0.680).

Discussion

Kit protein, a cell-surface transmembrane receptor for stem cell factor, have been identified as a key oncogenic driver in a variety of solid tumours [47]. KIT has been shown to plays a role in the proliferation of a number of cell types, including mast cells, melanocytes, germ cells, and hematopoietic stem cells [48]. Recently, KIT, as a biomarker of prognosis in malignancies, has generated much interest. But the conclusions for its prognostic value are controversial. In gastrointestinal stromal tumors, myeloid leukemias and mast cell disorders, for example, c-KIT gene gain-of-function mutations result in constitutive tyrosine kinase activity and are considered to play a central role in oncogenesis and sustained tumor growth [49]. Coexpression of c-KIT and its ligand in small-cell lung cancer, for example, appears to result in an autocrine growth loop sustaining tumor cell proliferation [50]. On the other hand, a markedly better outcome has already been demonstrated in tumors that expressed c-KIT compared with those that did not, such as nasopharyngeal carcinomas [20] and multiple myeloma [21]. Thus, we performed a meta-analysis to veritably evaluate the role of KIT in the prognosis of cancer.

To our knowledge, this meta-analysis is the first study to systematically assess the association between CD117 expression and prognosis of various cancer types. It is reported that CD117 is expressed in a high proportion of patients with SCLC (28%-100%) [33,51]. And between 79% and 88% of SCLC cell lines were found to express CD117 [52]. Recently, a number of studies have focused on the association of CD117 expression with the survival of SCLC cancer patients. In this study, Totally 8 studies with the follow-up time of more than three years are involved. The pooled analysis shows that positive expression of CD117 is not significantly associated with the poor OS and DFS of patients with SCLC. In the subgroup analysis by tumor histological type, the region of reports or cut off value, there was still no significant association detected in all stratified analysis.

Interestingly, pooled analysis of the included studies exhibited a significant correlation between CD117 expression and poor OS of patients with osteosarcoma and renal carcinoma, suggesting that CD117 might be an independent prognostic factor of poor survival in patients with osteosarcoma and renal carcinoma. However, considering the limited studies included in the meta-analysis, our results should be interpreted with caution. More studies with a larger sample size on those cancer types are thus needed to further elucidate the prognostic value of CD117 in them.

Some limitations of this meta-analysis should be acknowledged. First, we found the heterogeneity in our meta-analysis. The heterogeneity could be explained by the fact that the technique of detecting CD117 is not comparable among the studies. Most studies (77.4%) in the meta-analysis used IHC staining to study expressions of CD117. Although IHC staining is simple and cost-effective to perform, results are highly dependent on a variety of methodological factors, such as storage time, fixation method of paraffin-embedded tissues, different primary antibodies, the revelation protocols and different levels of positive (0, 10, 50%, different scores combining intensity and percentage, intensity only) [53]. Immunostaining cutoff point were arbitrarily selected and varied between studies. So, variability in protein expression assessment must be considered a potential source of bias. Second, potential publication bias was a concern. We restricted our review to articles published in English or Chinese language because other languages were not accessible to the readers. This selection could favor the positive studies that are more often published in English while the negative ones tend to be more often reported in native languages [54].

To sum up, CD117 expression exhibited the significant association with OS of osteosarcoma and renal carcinoma. Whereas, the large prospective clinical studies based on homogeneous series of patients are needed to further confirm the prognostic value of CD117 in different types of cancer.

Disclosure of conflict of interest

None.