ADAM17 overexpression is associated with poorer clinical outcomes in cancer patients: a systematic review and meta- analysis

ADAM17 (a disintegrin and metalloproteinase 17) is overexpressed in many types of cancer, and its high expression in cancer patients is associated with increased risk of metastasis and a poorer prognosis. However, the clinical value of ADAM17 as a tumor biomarker remains largely unknown. In this meta-analysis, we investigated the relationships between ADAM17 levels, cancer metastasis, and patient prognosis. Twelve eligible studies involving a total of 1816 patients were identified in searches of the PubMed, Embase, Cochrane, and Chinese National Knowledge Infrastructure databases. Analysis of these studies revealed that ADAM17 overexpression is positively associated with lymph node metastasis in a fixed effect model and distant metastasis in a random effect model. Moreover, high ADAM17 expression was predictive of shorter overall and progression-free survivalof cancer patients in fixed effect models. These results suggest that ADAM17 maybe a useful biomarker for predicting metastasis and prognosis in cancer patients and may also aid in risk stratification and selection of individualized treatments.


INTRODUCTION
Cancer has become one of the most important diseases threatening human health and life [1]. Despite increasing knowledge of the complex mechanisms behind neoplastic disease, only a few biomarkers are routinely used to diagnose some cancers in the clinical setting (e.g., PSA [2] and AFP [3] in prostate and hepatic cancer, respectively). Most patients have either regional or distant metastatic disease, which requires more complicated therapies and is associated with poor prognosis, upon diagnosis [4]. It is therefore important to identify additional biomarkers that can predict prognosis and guide individualized treatments.

Meta-Analysis
Oncotarget s996 www.impactjournals.com/oncotarget (EGFR)/phosphoinositide 3-kinase/protein kinase B (AKT) pathway [8]. Elevated ADAM17 expression is associated with poor prognosis in a number of human cancers, such as clear cell renal cell carcinoma, non-small cell lung cancer, esophageal squamous cell carcinoma, and gastric cancer [9]. ADAM17 may act as both a novel prognostic biomarker and a potential therapeutic target. However, the prognostic value of ADAM17 has not been well-characterized. We therefore conducted a systemic review and meta-analysis to examine the prognostic value of ADAM17 with regard to tumor metastasis, progression, and survival.

Search results
A flow diagram of the study selection process is shown in Figure 1. A total of 276 primary research articles were identified in database searches. After removal of duplicates (176 records), the 100 remaining articles were screened for eligibility. Of these, 85 were excluded based on title, abstract, publication type, or full text, leaving 15 articles for further evaluation. Three additional articles did not meet the inclusion criteria (2 lacked crucial data, including HR and 95%CI; 1 used two cutoffs) and were therefore excluded. As a result, a total of 12 studies involving 1816 cases were ultimately included in this meta-analysis.

Characteristics of included studies
Basic information for the included studies is shown in Table 1. The 12 included articles were published between 2005 and 2015. Ten different tumor types were evaluated in this meta-analysis; 3 studies examined gastric cancer [11][12][13], and 1 study each examined clear cell renal cell carcinoma (cRCC) [10], non-small cell lung cancer (NSCLC) [9], esophageal squamous cell carcinoma (ESCC) [14], gallbladder carcinoma [15], uterine cervical carcinoma [16], breast cancer [17], extrahepatic cholangiocarcinoma (ECC) [18], glioma [19], and ovarian cancer [20]. Cutoff values differed among the included studies. Additionally, univariate analysis was used in three studies, while multivariate analysis was used in the remaining nine. OS was examined in ten studies and PFS was evaluated in five. Of the six studies that examined the relationship between ADAM17 expression and lymph node metastasis, positive correlations were identified in three. Original OS (HRs and 95% CIs) and PFS data from the included studies are shown in the forest plots in Figures 2 and 3; original lymph node and distant metastasis data are summarized in Table 2.

Quality assessment
The quality of the 12 studies included in our metaanalysis was evaluated according to the Newcastle-Ottawa scale. Quality scores ranged from 6 to 8, with a mean of 7; higher scores indicate better methodology. None of the studies were excluded from subsequent analyses based on these scores.

ADAM17 expression is associated with lymph node metastasis
Six articles examined ADAM17 overexpressionin cancer tissues from patients with and without lymph   (Figure 4).

ADAM17 expression is associated with distant metastasis
Three articles examined ADAM17 overexpression in cancer tissues from patients with and without distant metastasis. Meta-analysis of the random effect model revealed that ADAM17 overexpression was more common in patients with distant metastasis than in those without distant metastasis (I 2 =52.8%; P= 0.12, OR=3.50, 95% CI 1.79-6.87, P<0.001) ( Figure 5).

Elevated ADAM17 expression is associated with poor prognosis
Ten studies reported data on OS for eight different tumor types ( Figure 2). Because no significant heterogeneity was identified (I 2 =14.7%, P=0.308), the fixed-effects model was used to estimate pooled HRs  and corresponding 95% CIs for low and high ADAM17 expression groups. Compared with low ADAM17 expression group, OS times were shorter in the high ADAM17 expression group than in the low ADAM17 expression group (HR=2.04, 95% CI 1.66-2.52, P<0.001). Five studies reported data on progressionfree survival (PFS) for five different tumor types (Figure 3). Because no significant heterogeneity was  Oncotarget s1000 www.impactjournals.com/oncotarget identified (I 2 =33.9%, P=0.195), the fixed-effects model was used to pool PFS data. The pooled HR of 1.81 (95% CI = 130-2.53, P < 0.001) indicated that ADAM17 overexpression was associated with poorer OS and PFS. Subgroup analysis was also performed to explore the influence of analysis type on survival findings. As shown in Table 3, the results of subgroup analysis confirmed the above findings.

Publication bias and sensitivity analysis
Publication bias was evaluated for LNM, DM, OS, and PFS using funnel plots. The funnel plots were almost symmetrical ( Figure 6). Begg's test indicated that there was no publication bias in our meta-analysis. Analyses were then repeated after each study was removed one at a time to determine whether any single study altered the overall results (Figure 7). The results indicated that no single study had a significant effect on the final results, suggesting that the findings are robust.

DISCUSSION
The adisintegrin and metalloprotease (ADAM) family proteins are Zn 2+ -dependent proteases that cleave  Oncotarget s1001 www.impactjournals.com/oncotarget membrane-bound proteins and/or degrade the extracellular matrix [21]. ADAMs participate in cellular adhesion and in proteolytic cleavage of various cell surface molecules. They are therefore important mediators of cell signaling events which regulate cellular fate, proliferation, and growth [22,23]. Recent studies have demonstrated that ADAMs are involved in various diseases, such as inflammatory bowel disease [24], systemic lupus erythematosus [25], neurodegenerative diseases [26]. Meanwhile, abnormal expression of ADAMs has been detected in tumor tissues and is a crucial contributor to oncogenic processes and aggressive progression in malignant tumors [27]. These findings indicate that the functions of ADAMs should be studied further.
ADAM17 was simultaneously identifiedin 1997 by two research groups as the enzyme responsible for cleavage of the transmembrane protein Tumor Necrosis Factor (TNF)-α; it was therefore originally named TNFα-converting enzyme (TACE) [28,29]. It is composed of 824 amino acids, and its gene is located on chromosome 2q25 [8]. ADAM17 is expressed in the kidneys, skeletal muscle, brain, and other normal tissues, and its expression is up-regulated in various types of tumor tissue [28,30]. Downregulation of ADAM17 expression can suppress proliferation and migration in tumor cell lines, suggesting that ADAM17 might be a potential antitumor therapeutic target [31]. In clear cell renal cell carcinoma, Li et al. [10] found that higher ADAM17 expression was associated with more advanced clinical characteristics and poorer prognosis. In 2016, Shenetal. reported that a marked increase in ADAM17 expression on the cell surface was associated with increased tumorigenesis, invasiveness, and drug resistance in breast cancer [30]. Lv et al. [32] demonstrated that downregulation of ADAM17 using an RNA silencing approach suppressed cell proliferation and invasion in vitro and tumor growth in non-small cell lung carcinoma. Accumulating evidence demonstrates that ADAM17 modulates many signaling pathways that control physiological and pathophysiological processes such Oncotarget s1002 www.impactjournals.com/oncotarget as development, regeneration, immunity, inflammation, and tumorigenesis [33]. ADAM17 has such wideranging effects mainly because it interacts with many different substrates. So far, more than 70 substrates have been identified, including epidermal growth factor receptor (EGFR/ErB1) ligands, amphiregulin (AREG), epiregulin, heparin-binding epidermal growth factor (HB-EGF), c-kit, CD40, interleukin-6 receptor, vascular cell adhesion molecule-1, and Notch [34,35]. However, the exact molecular mechanisms underlying associations between ADAM17 expression and cancer progression need to be further elucidated.
To our knowledge, this is the first metaanalysis to examine correlations between ADAM17 expression and clinical outcomes in cancer patients. The pooled results strongly suggest that elevated ADAM17 expression decreases OS and PFS in cancer patients, indicating that ADAM17 may be a promising biomarker for predicting prognosis in cancer patients. In addition, we found that higher ADAM17 expressionwas associated with increased LNM and DM, indicating that increased ADAM17 expression may be related to cancer progression. Furthermore, a sensitivity analysis revealed that no single study had a significant effect on the observed effect size (pooled HR), indicating that the findings were robust.
Several limitations to the present study should be considered when interpreting the results. First, the cut-off values used to define high ADAM17 expression differed among the studies examined. Second, all included studies were retrospective analyses, and some had small sample sizes. Third, most of the included studies reported significant findings, and it is likely that any relevant studies reporting negative results have not been published. Therefore, multi-center, prospective studies with larger sample sizes and uniform criteria for categorizing ADAM17 expression are needed to confirm the findings of this study.
In conclusion, our systematic review and metaanalysis indicates that elevated ADAM17 expression is significantly associated with increases in LNM and DM and poorer survival outcomes in cancer patients. Moreover, ADAM17 might serve as a novel prognostic biomarker and a promising therapeutic target.

Search strategy and selection of studies
This meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [36]. We searched the PubMed, Embase, Cochrane, and Chinese National Knowledge Infrastructure databases for studies published through June 2017. Both MHSH (medical subject heading terms) and free-words were used in the search strategy to increase its sensitivity. The following search terms used were used: "cancer or tumor or neoplasm" and "ADAM17 or Disintegrin and Metalloproteinase Domain Containing Protein 17 or Tumor Necrosis Factor Alpha Convertase or Tumor Necrosis Factor alpha Converting Enzyme or TNFalpha Converting Enzyme" and "survival or prognostic or prognosis or recurrence or clinical outcome". There were no language restrictions. We also screened the references of all papersidentified using this search strategy to identify additional potential eligible studies.

Inclusion and exclusion criteria
Studies were eligible for inclusion in this metaanalysis if they met the following criteria: 1) evaluated associations between ADAM17 protein expression and prognosis incancer patients; 2) cancer diagnosis was pathologically confirmed; 3) cohort design; 4) studies supplied sufficient information for calculating hazard ratio (HR) and 95% confidence interval (CI). If the articles only provided survival curves, relevant data were extracted from the survival curves usingEngauge Digitizer 4.1 software, and logHR and selogHRwere calculated according to Parmar et al. [37] and Tierney et al. [38]. The exclusion criteria were as follows: 1) duplicated studies using the same population or overlapping patient database; 2) reviews, comments, letters, conference abstracts, and case reports; 3) insufficient data for calculating HR and 95% CI; 4) evaluated associations between ADAM17 mRNA expression and clinical characteristics; 5) evaluated associations between ADAM17activity and clinical characteristics.

Quality assessment
The quality of the included studies was assessed according to the Newcastle-Ottawa Scale (NOS) [39]. This included an assessment of subject selection, comparability of groups, and clinical outcome. Quality scores ranged from 0 (lowest) to 9 (highest); studies with a NOS score ≥7 were considered high-quality.

Data extraction
All data were extracted from the studies by two independent investigators. Any disagreements were resolved by a third reviewer. Study characteristics included first author's name, year of publication, country, case number, time of follow-up, type of tumor, cut-off value, treatment, survival analysis methods (univariate, multivariate); outcome measures included HRs and 95% CIs for OS or PFS.

Statistical analysis
Statistical analyses were conducted using STATA version 12.0 (Stata Corporation, College Station, TX, USA). HRs and 95% CI were used to evaluate the relationship between ADAM17 and cancer prognosis (overall survival and progression-free survival). Odds ratios (ORs) and 95% CIs were used to assess the relationship between ADAM17 and clinical characteristics, which included distant metastasis (DM) and lymph node metastasis (LNM). Homogeneity of HRs and ORs across the studies was evaluated using the Q and I 2 statistics;statistically significant heterogeneity was defined by a p-value <0.10 for the Q-test or an I 2 value >50%. In the absence of significant heterogeneity, we pooled results using a fixed effect model; otherwise, a random effect model was used. Publication bias was evaluated by examining asymmetry of an inverted funnel plot. We also performed Begg's test to obtain quantitative evidence regarding publication bias. Sensitivity analyses were carried out to evaluate the stability of our results when individual studies were removed one at a time. All statistical tests were two-sided, and a P < 0.05 was considered statistically significant.