J Gastrointest Canc DOI 10.1007/s12029-015-9754-8
ORIGINAL RESEARCH
Associations of the Insertion/Deletion Polymorphism in the ACE Gene and Risk of Gastric Cancer: A Meta-Analysis Noel Pabalan 1 & Hamdi Jarjanazi 2
&
Hilmi Ozcelik 3
# Springer Science+Business Media New York 2015
Abstract Background Reported associations of ACE polymorphisms with gastric cancer have been inconsistent, prompting a meta-analysis of 12 published case-control studies where we estimated risk (odds ratio [OR]). Methods We searched MEDLINE through PubMed and EMBASE for suitable articles that had case-control design with gastric cancer as outcome. In this meta-analysis, our overall findings were subjected to modifier analyses (outlier and sensitivity treatments). We also performed subgroup analysis based on ethnicity (German and Japanese) and histological subtype (intestinal and diffuse). Results Significance of the protective effects among homozygote carriers of the II genotype (OR 0.54–0.63, P=0.01–0.02) disappeared with outlier analysis (OR 0.81–0.88, P=0.12– 0.14). Among DD homozygotes, this treatment altered the direction of association from weak protection (OR 0.95– 0.96, P=0.79–0.82) to increased risk (OR 1.13–1.19, P= 0.14–0.16). No significant associations were observed among ID genotype carriers (OR 0.91–0.94, P=0.69–0.72). Japanese
Dedication This paper is dedicated to the memory of Dr. Hilmi Ozcelik, a longtime friend, colleague, and mentor. * Hamdi Jarjanazi
[email protected] 1
Center for Research and Development, Angeles University Foundation, 2009 Angeles City, Philippines
2
Environmental Monitoring and Reporting Branch, Ontario Ministry of the Environment and Climate Change, 125 Resources Road, Toronto, ON M9P 3V6, Canada
3
Fred A. Litwin Centre for Cancer Genetics, Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 60 Murray St. Room L6-303, Box 29, Toronto, ON M5T 3L9, Canada
pooled effects varied across the genotype comparisons (OR 0.93–1.06, P=0.54–0.72). Sensitivity treatment demonstrated robustness of the II genotype, but not the other two, both in overall and subgroup analyses. Histological subtype analysis yielded protection from intestinal cancer across the comparisons (OR 0.38–0.71, P=0.15–0.50) but variable results for the diffuse type (OR 0.59–1.32, P=0.19–0.92). Conclusion In summary, carriers of the ACE II genotype appear to be protected from gastric cancer, regardless of ethnicity or tumor type. Keywords ACE . Gastric cancer . Meta-analysis . Polymorphism
Introduction Angiotensin I-converting enzyme (ACE) is expressed by many cell types of various organs [1] and generates angiotensin II, the key effector in the renin-angiotensin system (RAS) which regulates blood pressure. Evidence indicates a role of ACE in cancer given its differential expression in several nancies and upregulation of its receptors in many cancer tissues [1]. ACE may exert direct tumor promoting functions, such as promoting angiogenesis, cell proliferation, and tumor growth [2]. ACE regulates angiotensin II, and its upregulation overactivates RAS facilitates cancer progression [3–7]. However, it has been postulated that angiotensin II affects gastric carcinogenesis mainly through chymase rather than ACE [8]. Inhibition of increased ACE activity suppresses tumor growth in animal models [9, 10] and epidemiologic evidence shows that ACE inhibitors may decrease risk and mortality rate of cancer [11] and are being considered in cancer treatment [11, 12]. An insertion/deletion (I/D) polymorphism
J Gastrointest Canc
(rs4646994) in the human ACE gene has been reported to influence ACE activity, in which there is an insertion (I) or deletion (D) of a 287-base pair Alu-repetitive sequence in intron 16 [13]. Serum and tissue expression, hence ACE activity is influenced by the I/D polymorphism, with an increased amount of ACE messenger RNA (mRNA) from the D allele [14]. This increased activity of ACE in blood and tissues among individuals with the DD genotype has been associated with various cancers, including prostate and breast cancer [3, 5]. The highest level of circulating and tissue activity of ACE has been found in carriers of the DD genotype compared to the ID/II genotype [15]. In addition, it has been suggested that incidence of gastric cancer is much higher in subjects with the DD genotype than others [16] but published studies are somehow inconsistent [17–19]. To help clarify the findings, we conducted a meta-analysis of published studies on ACE I/D polymorphism and risk of gastric cancer.
Search terms: “angiotensin-converting enzyme”, “polymorphism”, “gastric cancer”, “stomach”, “tumor” and “malignancy” as of 12 May 2014 10 citations identified 3 screened: 2 review; 1 functional 7 abstracts retrieved for detailed evaluation
7 full-text articles retrieved for detailed evaluation 1 excluded case only 6 articles potentially included in the meta-analysis 1 article with data error Goto 2005
Materials and Methods Selection of Studies and Data Extraction We searched MEDLINE using PubMed and EMBASE for association studies as of May 12, 2014 using the following keywords: Bangiotensin-converting enzyme,^ BACE,^ and Bpolymorphism^ or Bgenotype^, or Binsertion/ deletion^, or BI/D^, or BD/I^, or BD/D^ or BI/I^ and Bgastric,^ or Bstomach cancer,^ as well as Bstomach,^ Btumor,^ and Bmalignancy.^ Studies were eligible if they had genotypic data with a case-control design. Figure 1 summarizes the literature search, which yielded ten citations. After a series of exclusions, six articles were included in the meta-analysis. Two investigators independently extracted data and reached consensus on all the items. Of the seven included articles, three and four examined two distinct populations, respectively, German [16, 17, 20] and Japanese [8, 18, 19, 21]. Four articles [16–18, 20] presented data in multiple categories. Separate data overall and for tumor type (intestinal and diffuse) were considered for three studies each in Ebert et al. [17], Kupcinskas et al. [20], and Röcken et al. [16]. Histological subtype treatment of Sugimoto et al. [8] did not include genotypic data; thus, our analysis for tumor type was confined to the German population only. The Goto et al. article [18] was followed by an erratum [21] where we considered only the latter. These ten studies plus the two single-study articles [8, 19] resulted in a total of 12 studies [8, 16, 17, 19–21]. The following information was obtained from each publication: first author’s name, published year, race of the study populations, and genotype data. Hardy-Weinberg equilibrium (HWE) was assessed using the χ2 test.
1 erratum [Goto 2006] furnished by author after contacting Goto 6 articles included in the meta-analysis Fig. 1 Summary flowchart of literature search
Quality Assessment of the Studies The Newcastle-Ottawa Score (NOS) quality assessment scale [22] was used to evaluate the methodological quality of included studies. These studies were judged based on three broad perspectives: selection, comparability, and exposure (case-control studies) or outcome (cohort studies) by a Bstar^ rating system with a score ranging from zero star (worst) to nine stars (best). A score of seven stars or greater was an indication of high quality. Meta-Analysis Risks (odd ratios [ORs]) of gastric cancer with the ACE I/D polymorphism were estimated for each study. The minor allele differed between the two-race subgroups, Ι in the German population (x =0.48±0.01) and D among the Japanese (x = 0.36±0.03). Adjusting for this difference between the two populations, we carried out independent statistical analyses using the low activity II and high activity DD genotypes as references. Following the method of Sun et al. [23], we thus compared the following: (i) DD homozygotes compared to
J Gastrointest Canc
ID/II genotype; (b) II homozygotes compared to ID/DD genotype, and (iii) ID genotype carriers with homozygotes of the II and DD genotypes. To compare effects on the same baseline, we used raw data to calculate pooled OR which were obtained using either fixed [24] (in the absence of heterogeneity) or random [25] (in its presence) effects models. Heterogeneity between studies was addressed in the following manner: (i) estimated using the χ2-based Q test [26], (ii) explored using subgroup analysis [26] with race and tumor type as variables, (iii) its sources detected using the Galbraith plot [27, 28] which seeks to identify outlying studies, and (iv) quantified with the I2 statistic which measures the degree of inconsistency among studies [29]. Sensitivity analysis which involves sequentially excluding each study and recalculating the pooled OR was used to test for robustness of the findings. Data were analyzed using Review Manager 4.3 (Cochrane Collaboration, Oxford, England) and SigmaStat 2.03 (Systat Software, San Jose, CA). Two-sided P values of ≤0.05 were considered significant except in estimation of heterogeneity which was set at ≤0.10, given the low power of the chisquare-based Q test [30]. Publication bias [31] was assessed with two tests; Egger’s regression [32] and the BeggMazumdar correlation [33].
Results Characteristics of the Studies Table 1 summarizes the features of the included articles, majority (66.7 %) of which were hospital-based. The NOS results show a mean (SD) score of x =6.67±1.21 indicating that the Table 1
overall methodological quality of the component articles was moderate. Stratified by race, the methodological quality of the German studies was higher (x =7.67±0.58) than those of the Japanese (x =5.67±0.58). Table 2 lists the genotype data from 1459 cases and 2581 controls obtained from six articles that included 13 study populations of which 9 were German (579 cases/ 572 controls) and four of Japanese origin (880 cases/ 2009 controls). We also subgrouped gastric cancer by tumor type: intestinal (161 cases/572 controls) and diffuse (103 cases/572 controls), according to the Lauren classification [34]. Control frequencies in all the studies conformed to the HWE. Overall and Subgroup Findings Table 3 summarizes the ACE I/D findings both overall and in the subgroups. Figure 2 shows a statistically significant protective effect (OR 0.63, P=0.01) among low activity II homozygotes when compared with ID/DD genotype carriers. Ethnic analysis showed that this mainly comes more from the German subgroup (OR 0.54, P = 0.02) than the Japanese effect (OR 0.97, P= 0.74). The high activity DD homozygotes were shown to have overall non-significant association with gastric cancer (OR 0.96, P=0.79) when compared with ID/II genotype carriers (Fig. 3). This did not vary among the Germans (OR 0.95, P=0.82) and Japanese (OR 1.06, P=0.64). Risk of gastric cancer among the ID genotype carriers showed no significant association (OR 0.94, P=0.69) which also did not vary among the Germans (OR 0.91, P = 0.72) and Japanese (OR 0.97, P = 0.72). Pooled effects in the
Characteristics of the studies included in the analyses of the association of the ACE I/D polymorphism with gastric cancer
First author
Year
[R]
GC type
SD
Cases
Controls
NOS
Ebert
2005
[17]
LC
PB
Samples from patients with noncardia cancer were obtained by gastric resection
8
Rocken
2005
[16]
LC
PB
Kupcinskas
2011
[20]
LC
HB
Sera or tissue samples were obtained from patients who had undergone either complete or partial gastrectomies Gastric cancer was determined by histology
Goto
2006
[21]
nm
HB
Sugimoto
2006
[8]
LC
HB
Blood samples were obtained from healthy controls without gastric cancer Sera or tissue samples obtained from control patients without gastric cancer Healthy volunteers with no history of malignancy, gastrointestinal disease or surgery Health checkup examinees with no history of cancer who attended a health checkup program Subjects without H. pylori infection using serologic testing, rapid urease test and/or culture
Hibi
2011
[19]
nm
HB
Patients with pathologically confirmed diagnosis of gastric adenocarcinoma who underwent tumor resection Patients with H. pylori infection based on serologic testing, rapid urease test, and/or culture and proven histopathologically endoscopically Patients with gastric cancer diagnosed from 2001 to 2005
Sex and age frequency-matched cancer-free controls.
8
7
5
6
6
R reference, LC Lauren classification, nm not mentioned, GC gastric cancer, SD study design, PB population-based, HB hospital-based, NOS Newcastle-Ottawa Score
J Gastrointest Canc Table 2 Genotype frequencies of the included studies
Genotypes (n) Case First author [R]
Race
II
Control ID
DD
II
ID
DD
maf*
HWE
0.96
1
Gastric cancer no subtype classification Ebert [17] German 6
46
36
33
72
40
0.52
2
Rocken [16]
German
24
57
32
41
95
53
0.53
0.90
3 4
Kupcinskas [20] Goto [21]
German Japanese
27 64
59 89
28 26
62 60
110 62
66 15
0.51 0.34
0.24 0.69
5
Sugimoto [8]
Japanese
54
53
12
50
60
22
0.39
0.58
6
Japanese
252
255
75
745
791
204
0.34
0.79
7 8 9
Hibi [19] Intestinal tumor type EbertI [17] KupcinskasI [20] RockenI [16]
German German German
5 8 14
22 21 38
19 18 16
33 62 41
72 110 95
40 66 53
0.52 0.51 0.53
0.96 0.24 0.90
10 11 12
Diffuse tumor type EbertD [17] KupcinskasD [20] RockenD [16]
German German German
0 14 4
13 27 17
10 6 12
33 62 41
72 110 95
40 66 53
0.52 0.51 0.53
0.96 0.24 0.90
R reference; maf* minor allele frequency, for Germans (I allele) for Japanese: (D allele); HWE Hardy-Weinberg Equilibrium
Japanese samples were homogeneous (Pheterogeneity =0.10– 0.61, I2 =0–56 %) marked by no significant associations (OR 0.93–1.06, P = 0.54–0.72) across all comparisons. Stratified by cancer site, both alleles and ID genotype elicited protective effects (OR 0.38–0.71, P=0.15–0.50) from intestinal cancer, while effects were variable in diffuse cancer where it spanned from protection (OR 0.59, P=0.38) to increased risk (OR 1.32, P=0.19). Of the tumor type findings, this increased risk effect was the only homogeneous pooled OR (Pheterogeneity = 0.72, I2 = 0 %) while all others were obtained under heterogeneous conditions (Pheterogeneity =0.0001–0.07, I2 =63–95 %). Outlier Treatment Because the overall and German subgroup effects were heterogeneous (Pheterogeneity
