Clinicopathological and prognostic significance of sialyl Lewis X overexpression in patients with cancer: a meta-analysis.

Many studies have shown that sialyl Lewis X (sLe(X)) is related to cancer prognosis and clinicopathology, but failed to provide conclusive results. We conducted the present meta-analysis to identify the association between sLe(X) overexpression and cancer prognosis. We searched studies in PubMed and Embase databases. Relative risk or hazard ratio with 95% confidence intervals were estimated with the Mantel-Haenszel random-effect method and 29 studies were included. Our meta-analysis showed that sLe(X) overexpression is significantly related to lymphatic invasion, venous invasion, T stage, N stage, M stage, tumor stage, recurrence, and overall survival. In subgroup analysis, we found that cancer type and ethnicity might be two major contributing factors to the possible presence of heterogeneity among the studies. In conclusion, sLe(X) overexpression is associated with tumor metastasis, recurrence, and overall survival in cancer patients, it plays an important role in cancer prognosis.

Introduction

As is known to all, cancer is a common life-threatening disease. According to recent studies, the incidence of cancer increases 1% per year in Europe.1 Among the adult population, a rising trend is reported for soft tissue sarcoma.2 Breast, colorectal, prostate, and lung cancers are the most common oncological cause for death among the European population.3 Cancer cannot be cured, as expected, due to the limited knowledge of iatrotechnique. So, exploration of more precise bio-indicators is valuable for early diagnosis of cancer and improving prognosis of patients.

Cell surface carbohydrates are involved in various biological processes such as cellular differentiation, maturation, proliferation, and malignant transformation.4 Dramatic changes of cell surface carbohydrates are associated with cancer occurrence, tumor invasiveness, and metastatic behavior.5 Sialyl Lewis X (sLeX) (NeuNAcα2,3Galβ1,4[Fucα1,3] GlcNAc), a carbohydrate antigen, is related to cell adhesion and our previous study showed that inhibition of sLeX synthesis leads to decreased adhesion of trophoblast cells to endometrial epithelial cells.6 Also, sLeX is frequently expressed in human cancer cells and primary tumors.7,8 As a ligand for E-selectin and L-selectin, sLeX is related to cell adhesion.9 It has been demonstrated that sLeX was involved in the adhesion of tumor cells to vascular endothelium.10 The potential role of sLeX in the tumor metastatic process has been supported by several clinical studies.11–14

Many studies have identified the relationship between sLeX and cancer prognosis, but individual studies of the influence of sLeX expression in cancer have failed to provide conclusive results. The present meta-analysis was conducted to further explore the relationship between sLeX expression and cancer prognosis and clinicopathology.

Materials and methods

Publication search

We searched published studies in the PubMed and Embase databases up to May 2014 with the following search terms: (slex OR sialyl lewis x) AND (cancer OR neoplasms OR carcinoma OR tumor) AND prognosis. Furthermore, reference lists of main reports and review articles were also reviewed to identify additional relevant publications. The study was conducted and reported following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.

Selection criteria

Two authors (YL and JXL) reviewed the retrieved titles and abstracts to discriminate the eligible studies for inclusion in our meta-analysis independently. Published studies were included based on the following criteria: 1) written and published in English; 2) patients with cancer diagnosis by pathology; 3) studies about sLeX expression in cancer tissues; 4) sLeX expression was measured by immunohistochemistry (IHC) method; 5) full length paper with sufficient data on sLeX expression and prognosis and prognosis-related factors; 6) we could find the full text. We excluded studies with the following criteria: 1) written and published in a language other than English; 2) studies about cell lines and animals; 3) studies about sLeX expression in serum; 4) review articles without original data; 5) a commentary, letter to the editor, or monograph.

Data extraction

Two authors (YL and WG) performed the data evaluation independently. The following data were extracted from each study: the first author’s last name; publication year; country; cancer source; number of patients; number of sLeX expressions (positive/negative); clinicopathological factors (age, sex, tumor size, histological differentiation, lymphatic invasion, venous invasion, T/N/M stage, tumor stage, and recurrence); survival analysis.

Data synthesis and statistical analysis

Expression of sLeX was analyzed as dichotomous variables, as positive expression versus negative expression. The clinicopathological factors were also conducted as dichotomous variables, as older age versus younger age for age; male versus female for sex; large versus small for tumor size; high versus low for histological differentiation; I and II versus III and IV for tumor stage; pT2 versus more than pT3 for depth of invasion (T stage); with versus without for lymphatic invasion, venous invasion, lymph node metastasis (N stage), distant metastasis (M stage), recurrence. Survival of sLeX expression was analyzed by Cox’s regression analysis conducted as hazard ratio (HR) and 95% confidence interval (95% CI). The data of expression of sLeX and clinicopathological factors or survival rate were extracted and calculated by initial data of studies. These data were analyzed with random-effect method, and were measured in relative risk (RR) with 95% CI. Statistical heterogeneity was estimated by means of Cochran’s Q test and I2 test. The I2 test represents the percentage of variation to heterogeneity, which is categorized as low (0%–40%), moderate (40%–60%), high (60%–90%), very high (>90%). Subgroup analyses were carried out based on cancer or country of the included studies if a significant heterogeneity was found in overall meta-analysis. To identify any potential publication bias, we used Begg’s test. All statistical analyses were performed with Review Manager 5.2 and STATA 12.0.

Results

Systematic review

We identified 178 studies that fit our search strategy, 41 studies were identified in our primary search (Figure 1). Finally, 29 studies published between 1993 and 2013 were included in our meta-analysis.11,12,14–40 Detailed characteristics of these studies are provided in Table 1.

Figure 1

The flow diagram of included/excluded studies.

Table 1

Characteristics of the included studies

Study ID	Country	Cancer source	Number of patients	sLeX expression (positive/negative)	Clinicopathological factors	Survival analysis
Nakamori et al18 (1993)	Japan	Colorectal cancer	132	50/82	Sex, differentiation, T stage, N stage, lymphatic invasion, venous invasion, tumor stage, recurrence	NA
Yamaguchi et al19 (1994)	Japan	Colorectal cancer	170	56/114	Differentiation, T stage, N stage, lymphatic invasion, venous invasion, tumor stage, recurrence	NA
Idikio20 (1997)	Canada	Prostate cancer	38	30/8	Differentiation	NA
Nakamori et al21 (1997)	Japan	Colorectal cancer	159	58/101	Age, sex, differentiation, T stage, N stage, lymphatic invasion, venous invasion, tumor stage	NA
Shimodaira et al22 (1997)	Japan	Colorectal cancer	43	28/15	Tumor size, differentiation, T stage, N stage, lymphatic invasion, venous invasion, tumor stage	NA
Ura et al12 (1997)	Japan	Gastric cancer	110	91/19	T stage, N stage	NA
Baldus et al17 (1998)	Germany	Gastric cancer	127	85/42	Sex, tumor stage	NA
Farmer et al23 (1998)	United States	HNSCC	82	51/31	Age, sex, M stage, tumor stage	NA
Fukuoka et al11 (1998)	Japan	Lung cancer	52	34/18	N stage, M stage	NA
Tatsumi et al24 (1998)	Japan	Gastric cancer	87	41/46	Differentiation, T stage, N stage, M stage, lymphatic invasion, venous invasion	NA
Yamaguchi et al25 (1998)	Japan	Breast cancer	102	61/41	Age, tumor size, N stage	NA
Kurahara et al14 (1999)	Japan	OSCC	70	24/46	M stage	NA
Takao et al26 (1999)	Japan	EBDC	73	45/28	Age, sex, differentiation, T stage, N stage, M stage, lymphatic invasion, venous invasion, tumor stage	NA
Futamura et al27 (2000)	Japan	Gastric cancer	245	135/110	Age, sex, differentiation, T stage, N stage, M stage, venous invasion, tumor stage	NA
Grabowski et al28 (2000)	Germany	Colorectal cancer	182	103/79	Sex, differentiation, T stage, N stage, M stage, tumor stage	Multi
Nakagoe et al16 (2000)	Japan	Colorectal cancer	101	76/25	Tumor stage	Uni
Machida et al29 (2001)	Japan	Lung cancer	25	19/6	Tumor size, N stage, M stage, lymphatic invasion, venous invasion	NA
Takahashi et al30 (2001)	Japan	PDAC	23	15/8	NA	Multi
Baldus et al31 (2002)	Germany	Colorectal cancer	243	165/78	Differentiation, N stage, M stage, tumor stage	NA
Konno et al32 (2002)	Japan	Colorectal cancer	134	47/87	N stage, M stage, venous invasion	Multi
Nakagoe et al34 (2002)	Japan	Breast cancer	87	37/50	Age, differentiation, T stage, N stage, M stage, tumor stage	Multi
Nakagoe et al33,34 (2002)	Japan	Gastric cancer	101	31/70	Age, sex, tumor size, differentiation, T stage, N stage, lymphatic invasion, venous invasion	Multi
Kashiwagi et al35 (2004)	Japan	Gallbladder cancer	54	28/26	T stage, N stage, lymphatic invasion, venous invasion	NA
Yu et al36 (2005)	People’s Republic of China	Lung cancer	61	40/21	Age, sex, T stage, N stage, recurrence	Uni
Faried et al37 (2007)	Japan	ESCC	130	40/90	Sex, differentiation, T stage, N stage, M stage, lymphatic invasion, venous invasion, tumor stage	Multi
Croce et al38 (2008)	Argentina	HNSCC	125	29/96	Age, sex, differentiation, T stage, N stage, tumor stage	NA
Sozzani et al39 (2008)	Italy	Breast cancer	127	37/90	Differentiation, T stage, N stage, venous invasion	NA
Portela et al40 (2011)	Spain	Colorectal cancer	155	67/88	Age, sex, tumor size, differentiation, T stage, N stage, M stage, tumor stage	NA
Schiffmann et al15 (2012)	Germany	Colorectal cancer	215	102/113	Sex, differentiation, T stage, N stage, M stage	NA

Abbreviations: NA, not available; OSCC, oral squamous cell carcinoma; EBDC, extrahepatic bile duct carcinoma; PDAC, pancreatic ductal adenocarcinoma; Multi, Multivariate; Uni, Univariate; sLeX, sialyl Lewis X; ESCC, esophageal squamous cell carcinoma; HNSCC, head and neck squamous cell carcinoma.

Association of sLeX expression with cancer prognosis and clinicopathology

sLeX expression correlated with prognostic factors, including lymphatic invasion (lymphatic invasion versus non-lymphatic invasion) (pooled RR =1.36, 95% CI: 1.15–1.61, I2=62.3%), venous invasion (venous invasion versus non-venous invasion) (pooled RR =1.41, 95% CI: 1.18–1.67, I2=52.9%), T stage (pT3–4 stage versus pT2 stage) (pooled RR =1.14, 95% CI: 1.04–1.27, I2=59.6%), N stage (lymph node metastasis versus non-lymph node metastasis) (pooled RR =1.46, 95% CI: 1.29–1.66, I2=55.1%), M stage (distant metastasis versus non-distant metastasis) (pooled RR =1.76, 95% CI: 1.34–2.31, I2=42.1%), tumor stage (stage III/IV versus stage I/II) (pooled RR =1.42, 95% CI: 1.19–1.68, I2=69.9%), tumor recurrence (recurrence versus non-recurrence) (pooled RR =2.92, 95% CI: 2.02–4.23, I2=0.0%) (Figure 2A).

Figure 2

The association between sLeX and cancer prognostic factors.

Notes: (A) The cancer prognostic factors which were significantly related to sLeX overexpression. (a) Lymphatic invasion; (b) venous invasion; (c) T stage; (d) N stage; (e) M stage; (f) tumor stage; (g) recurrence. (B) The cancer prognostic factors which were not significantly related to sLeX overexpression. (a) Age; (b) sex; (c) tumor size; (d) differentiation. Weights are from random effects analysis.

Abbreviations: RR, relative risk; CI, confidence interval; sLeX, sialyl Lewis X.

Meantime, we found that sLeX overexpression was not significantly related to cancer prognosis and clinicopathology factors, including age (older versus younger) (pooled RR =1.08, 95% CI: 0.97–1.21, I2=0.0%), sex (male versus female) (pooled RR =0.97, 95% CI: 0.88–1.07, I2=47.0%), tumor size (larger versus smaller) (pooled RR =1.23, 95% CI: 0.94–1.62, I2=51.1%), tumor differentiation (lower differentiation versus higher differentiation) (pooled RR =0.94, 95% CI: 0.72–1.21, I2=75.1%) (Figure 2B).

sLeX overexpression on cancer survival

Eight studies analyzed the overall survival (OS) of human cancer with positive/negative sLeX overexpression, the HRs ranged from 2.42 to 9.10.18,30,32,34–36,38,39 The summarized HR of negative versus positive was 3.11 (95% CI: 2.25–4.32) with low heterogeneity (I2=0.0%) (Figure 3).

Figure 3

Meta-analysis with a random-effect model for the association of sLex overexpression with overall survival.

Note: Weights are from random effects analysis.

Abbreviations: HR, hazard ratio; CI, confidence interval; sLeX, sialyl Lewis X.

Subgroup analyses

We chose subgroup analyses in meta-analysis with relative high heterogeneity (I2>40%). In subgroup analyses, studies were stratified by cancer category (colorectal cancer, gastric cancer, lung cancer, breast cancer, head and neck squamous cell carcinoma, esophageal squamous cell carcinoma, oral squamous cell carcinoma, gallbladder cancer, pancreatic ductal adenocarcinoma, prostate cancer, and extrahepatic bile duct carcinoma) or ethnicity (Asia, America, and Europe). In addition, most of these analyses showed low heterogeneity after stratification (Tables 2 and 3).

Table 2

Subgroup analyses of country

	Number of studies	Summary RR (95% CIs)	I2 value	ph
Sex
Overall	12	0.97 (0.88, 1.07)	47.0%	0.036
Asia	7	0.92 (0.80, 1.06)	56.5%	0.032
Europe	3	0.99 (0.83, 1.18)	0.0%	0.593
Americas	2	1.13 (0.95, 1.34)	24.2%	0.251
Tumor size
Overall	5	1.23 (0.94, 1.62)	51.1%	0.085
Asia	4	1.43 (1.16, 1.77)	0.0%	0.853
Europe	1	0.85 (0.62, 1.16)	NA	NA
Differentiation
Overall	17	0.94 (0.72, 1.21)	75.1%	0.000
Asia	11	1.11 (0.80, 1.55)	82.3%	0.000
Europe	4	0.66 (0.46, 0.93)	0.0%	0.715
Americas	2	0.63 (0.25, 1.57)	67.8%	0.078
Venous invasion
Overall	13	1.41 (1.18, 1.67)	52.9%	0.013
Asia	12	1.49 (1.29, 1.72)	31.0%	0.143
Europe	1	0.69 (0.42, 1.11)	NA	NA
T stage
Overall	18	1.14 (1.04, 1.27)	59.6%	0.001
Asia	13	1.23 (1.03, 1.47)	67.5%	0.000
Europe	4	1.11 (1.05, 1.19)	0.0%	0.497
Americas	1	0.91 (0.71, 1.17)	NA	NA
N stage
Overall	23	1.46 (1.29, 1.66)	55.1%	0.001
Asia	17	1.53 (1.28, 1.82)	65.7%	0.000
Europe	5	1.40 (1.21, 1.61)	0.0%	0.724
Americas	1	1.23 (0.83, 1.83)	NA	NA
M stage
Overall	14	1.76 (1.34, 2.31)	42.1%	0.049
Asia	9	2.20 (1.47, 3.30)	38.3%	0.113
Europe	4	1.37 (1.09, 1.72)	0.0%	0.410
Americas	1	0.89 (0.39, 2.05)	NA	NA
Tumor stage
Overall	15	1.42 (1.19, 1.68)	69.9%	0.000
Asia	9	1.62 (1.24, 2.10)	69.4%	0.001
Europe	4	1.32 (1.10, 1.59)	22.3%	0.277
Americas	2	1.08 (0.79, 1.49)	58.7%	0.120

Note: ph: P-value for heterogeneity within each subgroup.

Abbreviations: RR, relative risk; CI, confidence interval; NA, not available.

Table 3

Subgroup analyses of cancer types

Subgroup	Number of studies	Summary RR (95% CIs)	I2 value	ph
Sex
Overall	12	0.97 (0.88, 1.07)	47.0%	0.036
Colorectal cancer	4	0.92 (0.80, 1.06)	0.0%	0.978
Gastric cancer	3	1.12 (0.97, 1.29)	0.0%	0.981
HNSCC	2	1.13 (0.95, 1.34)	24.2%	0.251
EBDC	1	0.79 (0.59, 1.07)	NA	NA
Lung cancer	1	0.61 (0.44, 0.83)	NA	NA
ESCC	1	0.96 (0.82, 1.11)	NA	NA
Tumor size
Overall	5	1.23 (0.94, 1.62)	51.1%	0.085
Colorectal cancer	2	0.99 (0.68, 1.44)	46.7%	0.171
Breast cancer	1	1.38 (0.98, 1.93)	NA	NA
Lung cancer	1	1.42 (0.42, 4.85)	NA	NA
Gastric cancer	1	1.60 (1.13, 2.27)	NA	NA
Differentiation
Overall	17	0.94 (0.72, 1.21)	75.1%	0.000
Colorectal cancer	8	1.06 (0.74, 1.52)	69.6%	0.002
Gastric cancer	3	0.63 (0.53, 0.75)	0.0%	0.978
Breast cancer	2	1.07 (0.72, 1.60)	0.0%	0.548
Prostate cancer	1	0.87 (0.53, 1.41)	NA	NA
EBDC	1	2.70 (0.84, 8.63)	NA	NA
ESCC	1	1.46 (0.81, 2.64)	NA	NA
HNSCC	1	0.39 (0.15, 1.01)	NA	NA
Lymphatic invasion
Overall	10	1.36 (1.15, 1.61)	62.3%	0.005
Colorectal cancer	4	1.36 (1.09, 1.68)	56.7%	0.074
Gastric cancer	2	1.23 (0.55, 2.73)	85.4%	0.009
EBDC	1	1.31 (0.97, 1.78)	NA	NA
Lung cancer	1	2.53 (0.39, 16.31)	NA	NA
Gallbladder cancer	1	1.39 (0.92, 2.11)	NA	NA
ESCC	1	1.71 (1.40, 2.08)	NA	NA
Venous invasion
Overall	13	1.41 (1.18, 1.67)	52.9%	0.013
Colorectal cancer	5	1.57 (1.33, 1.84)	0.0%	0.746
Gastric cancer	3	1.48 (1.04, 2.12)	35.6%	0.212
Breast cancer	1	0.69 (0.42, 1.11)	NA	NA
EBDC	1	0.95 (0.61, 1.49)	NA	NA
Lung cancer	1	3.16 (0.50, 19.87)	NA	NA
Gallbladder cancer	1	1.05 (0.68, 1.64)	NA	NA
ESCC	1	2.05 (1.48, 2.83)	NA	NA
T stage
Overall	18	1.14 (1.04, 1.27)	59.6%	0.001
Colorectal cancer	7	1.22 (1.08, 1.38)	65.6%	0.008
Gastric cancer	4	1.04 (0.85, 1.28)	29.7%	0.234
Breast cancer	2	0.66 (0.31, 1.40)	0.0%	0.895
EBDC	1	1.13 (0.79, 1.62)	NA	NA
Lung cancer	1	0.83 (0.66, 1.04)	NA	NA
Gallbladder cancer	1	1.00 (0.47, 2.14)	NA	NA
ESCC	1	2.09 (1.43, 3.06)	NA	NA
HNSCC	1	0.91 (0.71, 1.17)	NA	NA
N stage
Overall	23	1.46 (1.29, 1.66)	55.1%	0.001
Colorectal cancer	9	1.54 (1.34, 1.75)	24.5%	0.226
Gastric cancer	4	1.28 (1.11, 1.47)	0.0%	0.393
Breast cancer	3	1.46 (1.04, 2.04)	41.6%	0.180
Lung cancer	3	2.00 (0.44, 8.97)	80.2%	0.006
EBDC	1	1.06 (0.57, 1.97)	NA	NA
Gallbladder cancer	1	1.13 (0.56, 2.29)	NA	NA
ESCC	1	2.70 (1.98, 3.68)	NA	NA
HNSCC	1	1.23 (0.83, 1.83)	NA	NA
M stage
Overall	14	1.76 (1.34, 2.31)	42.1%	0.049
Colorectal cancer	5	1.47 (1.15, 1.87)	9.2%	0.354
Gastric cancer	2	3.23 (1.67, 6.22)	0.0%	0.678
Lung cancer	2	3.21 (1.07, 9.69)	0.0%	0.871
Breast cancer	1	1.35 (0.20, 9.16)	NA	NA
EBDC	1	1.19 (0.60, 2.37)	NA	NA
ESCC	1	5.25 (2.18, 12.67)	NA	NA
HNSCC	1	0.89 (0.39, 2.05)	NA	NA
OSCC	1	1.24 (0.70, 2.21)	NA	NA
Tumor stage
Overall	15	1.42 (1.19, 1.68)	69.9%	0.000
Colorectal cancer	8	1.58 (1.36, 1.82)	13.0%	0.328
Gastric cancer	2	1.11 (0.88, 1.39)	19.5%	0.265
HNSCC	1	1.08 (0.79, 1.49)	58.7%	0.120
Breast cancer	1	0.90 (0.27, 2.97)	NA	NA
EBDC	1	1.12 (0.74, 1.70)	NA	NA
ESCC	1	3.04 (1.95, 4.73)	NA	NA

Note: ph: P-value for heterogeneity within each subgroup.

Abbreviations: RR, relative risk; CI, confidence interval; HNSCC, head and neck squamous cell carcinoma; OSCC, oral squamous cell carcinoma; EBDC, extrahepatic bile duct carcinoma; ESCC, esophageal squamous cell carcinoma; NA, not available.

Publication bias

Begg’s test was created for assessment of possible publication bias. It suggested that publication bias had little influence on these meta-analysis results (P>0.05) (Figure 4).

Figure 4

Begg’s test results of sLex overexpression and prognostic factors.

Notes: (A) Age; (B) sex; (C) tumor size; (D) differentiation; (E) lymphatic invasion; (F) venous invasion; (G) T stage; (H) N stage; (I) M stage; (J) tumor stage; (K) recurrence; (L) overall survival.

Abbreviations: sLex, sialyl Lewis X; SE, standard error.

Discussion

The cancer statistics of the USA, in 2013,41 clearly indicated that the methods of treatment for cancer need to be improved. Exploring new molecular biological prognostic and predictive markers is a hot topic in modern medicine. Nakagoe et al first reported that sLeX was expressed in serum of patients with gastric and colorectal cancer as a tumor-associated carbohydrate antigen, which was also proven by clinicopathological and immunohistochemical studies.42 The relationship between sLeX expression and cancer prognosis was identified by a number of studies, which did not show conformable results. To our knowledge, this is the first meta-analysis that systematically evaluates the relationship between sLeX expression and cancer prognosis and clinicopathology.

In the present study, a combined analysis of 29 articles (3,253 cancer patients) which showed the detection of high sLeX expression in tumor tissues with poor prognosis outcome in cancer patients was conducted. Our results indicated that sLeX expression was significantly correlated with lymphatic invasion, venous invasion, deep invasion (T stage), lymph node metastasis (N stage), distant metastasis (M stage), tumor stage, tumor recurrence, and OS. On the other hand, although a high level of sLeX expression was found in patients like the elderly, females, or patients with large size tumor and high differentiation, these results did not show any significance.

What makes sLeX overexpression account for the poor prognosis in cancer? By chemical analyses, it was shown that sLeX oligosaccharide was the minimal structure binding to E-, L-, and P-selectin,43 which was closely involved in the interaction between the endothelium and cancer cells. sLeX is most commonly found in malignant tumors and plays a key role in cancer stem cell metastasis, hypoxia, and TNF-α, and promotes tumor adhesion, invasion, and metastasis by upregulating the sLeX expression in the tumor microenvironment.44–46 In the present meta-analysis study, we also found that sLeX expression was correlated with tumor recurrence. On the other hand, it is widely accepted that expression of cell surface carbohydrates is altered during malignant transformation and tumor progression, and may influence determination of metastatic behavior of tumor cells.21,47 It has been identified that sLeX was a terminal tetrasaccharide moiety present on numerous membrane glycoproteins and glycolipids of epithelial and lymphatic cells.28 With such characters, a high level of sLeX contributes to cell adhesion, metastasis, and invasion because the cell surface antigens can combine with other cells directly. sLeX in conjunction with mucins, promotes cellular motility, thus contributing to tumor cell spreading and metastasis.11,48 Furthermore, sLeX is expressed on granulocytes and monocytes which mediates inflammatory extravasation.49,50 However, the molecular biological mechanisms of how sLeX overexpression affects the cancer prognosis are complicated and still need further exploration. For the first time, our meta-analysis study revealed that sLeX could be a potential biomarker for poor cancer prognosis.

Due to the differences in nationality and cancer types which could cause heterogeneity among the studies, we conducted a subgroup analysis. In the subgroup analysis, the sLeX overexpression may play different roles caused by differentiation, venous invasion, T stage, M stage, tumor stage, and sex factors among different types of cancers. These factors contribute to the possible presence of heterogeneity between the studies. The difference might be owing to the molecular biological mechanisms of interactions between sLeX overexpression, and the occurrence and development of different types of cancers. Otherwise, ethnicity may be another factor that contributes to heterogeneity in sex, tumor size, differentiation, venous invasion, T stage, and M stage. It might be owing to the differences in genetic backgrounds and the environment among different races. We also found high heterogeneity in some subgroups, because biological behavior of cancer might be affected by many possible factors during the complicated process of tumor development.

Some limitations of this meta-analysis need to be acknowledged. First, all published studies and papers were written in English, some related published or unpublished studies that met the inclusion criteria were missed. Most of the studies reported positive results, while studies of negative results were all rejected. Second, some cancers such as oral squamous cell carcinoma, gallbladder cancer, pancreatic ductal adenocarcinoma, prostate cancer, and extrahepatic bile duct carcinoma were included in only one article respectively, so we could not evaluate pooled data in subgroup analyses. Third, all of the included studies had data of the sLeX expression which was detected by IHC methods. It might have some bias because of different antibodies and different standards of positive/negative sLeX expression. However, it was not available for us to do a subgroup analysis to analyze the underlying bias of IHC on the pooled odds ratios or HRs. Finally, multivariate analyses were not performed on OS data in most included studies, we calculated the pooled HR only from available HRs.

In conclusion, our meta-analysis showed that a high level of sLeX expression was significantly associated with lymphatic invasion, venous invasion, deep invasion, lymph node metastasis, distant metastasis, tumor stage, tumor recurrence, and OS in cancer. sLeX might be a new prognostic biomarker, and it might become a new diagnostic and therapeutic target for cancer. Further studies are required to explore the molecular biological mechanisms of sLeX and factors that caused significant heterogeneity in the present meta-analysis study.