Diseases of the Esophagus (2015) ••, ••–•• DOI: 10.1111/dote.12349
Original article
Hormonal and reproductive factors and risk of esophageal cancer in women: a meta-analysis B. J. Wang, B. Zhang, S. S. Yan, Z. C. Li, T. Jiang, C. J. Hua, L. Lu, X. Z. Liu, D. H. Zhang, R. S. Zhang, X. Wang Department of Surgical Oncology, The Eighty-first Hospital of People’s Liberation Army, Nanjing 210002, China
SUMMARY. Currently published studies on the relationship between hormonal and reproductive factors and esophageal cancer (EC) risk in women have yielded contradictory findings. For a better understanding of this relationship, we first performed this meta-analysis by pooling all available publications. Sixteen independent studies were retrieved after a comprehensive search in PubMed and Embase databases. The pooled relative risks (RRs) with 95% confidence intervals (95% CIs) were calculated. The pooled RRs implicated that hormone replacement therapy was negatively associated with the risk of EC (RR = 0.72, 95% CI 0.60–0.86, P < 0.001) and esophageal squamous cell carcinoma (RR = 0.68, 95% CI 0.48–0.97, P = 0.031). Menopausal women were at an increased risk of EC (RR = 1.47, 95% CI 1.07–2.03, P = 0.018), particularly esophageal squamous cell carcinoma (RR = 1.66, 95% CI 1.12–2.48, P = 0.012). Additionally, decreased risk of EC (RR = 0.79, 95% CI 0.68–0.92, P = 0.003) and esophageal adenocarcinoma (RR = 0.66, 95% CI 0.53–0.82, P < 0.001) was demonstrated among women with breast-feeding history. Moreover, such associations were more significant among Caucasians, but not Asians. Our study suggests that menopause is an independent risk factor for EC, while hormone replacement therapy and breast-feeding history play a protective role against EC, particularly among Caucasians. All results are consistent with the hypothesis that effects of estrogen may lower the risk of EC in women. KEY WORDS: esophageal cancer, hormonal factor, meta-analysis, reproductive factor.
INTRODUCTION Esophageal cancer (EC) is the eighth most common cancers worldwide.1,2 Diet, tobacco smoking, alcohol consumption, and genetic polymorphism are wellestablished risk factors for EC.3–6 However, little has been known about the etiology of EC in women, since there have been fewer studies with fewer female patients. Inverse association has been clearly documented between higher exposure to female hormonerelated factors and the risk of many kinds of cancer, such as gastric cancer, colorectal cancer, and lung cancer.7–9 A number of epidemiological studies have Address correspondence to: Dr Xuan Wang, MD, Department of Surgical Oncology, The Eighty-first Hospital of People’s Liberation Army, Nanjing, Jiangsu Province 210002, China. Email:
[email protected] Specific author contributions: Bingji Wang, Bin Zhang, Shushan Yan, Zengcai Li, and Tao Jiang equally contributed to this work. Financial support: This work has no funding to support it. Conflicts of interest: All authors declare no conflicts of interest. © 2015 International Society for Diseases of the Esophagus
investigated the association of female hormonal factors and EC risk in different populations, but existing data are inconsistent and inconclusive probably due to diverse ethnicity, methodology, and relative small sample size in some individual studies. Meta-analysis of all currently published studies is the most valuable method to weigh potential associations in carcinogenesis for its big statistical power. Importantly, a better understanding of the relationship between hormonal and reproductive factors and the risk of EC and how these factors contribute to the development of EC can provide more effective prevention strategies for this deadly disease. Thus, we performed this meta-analysis of 16 independent studies to provide a more precise estimate of the underlying effects of female hormone-related factors on esophageal carcinogenesis. In addition, a separate analysis according to ethnicity and histological types of EC (esophageal squamous cell carcinoma [ESCC] and esophageal adenocarcinoma [EAC]) was also conducted for further evaluation. 1
2
Diseases of the Esophagus
METHODS
RESULTS
Search strategy for eligible studies
Identification and characteristics of studies
Studies on the association between hormonal and reproductive factors and EC risk were searched in PubMed and Embase databases from their inception up to October 10, 2014 using the following items: esophageal cancer, esophageal squamous cell carcinoma, esophageal adenocarcinoma, EC, ESCC, or EAC; and hormonal factors, oral contraceptive (OC), hormone replacement therapy (HRT), reproductive factors, menstrual factors, age at menarche, age at first birth, menopausal status, age at menopause, parity, or breast feeding; and risk. References of retrieved studies were also screened for additional publications.
After a comprehensive search in databases of PubMed and Embase from their inception up to October 10, 2014, a total of 11 publications were retrieved.16–26 Characteristics of all included studies were summarized in Table 1. Among the 11 articles, five were respectively divided into two independent studies according to histological types of EC.17,18,20–22 Accordingly, there were 16 independent studies included into the present meta-analysis. Eight independent studies were related to ESCC risk, six were about EAC risk, and still two were regarding total EC risk. Among the 16 studies, five were in prospective cohort designs, six were in population-based case–control designs, four were in hospital-based case–control designs, and still one was in nested case– control design.
Inclusion criteria The inclusion criteria was as follows: (i) studies on the association of hormonal and reproductive factors with EC risk; (ii) cohort studies; (iii) case–control studies; and (iv) publications with enough information for odds ratio (ORs) or relative risks (RRs) or hazard ratios (HRs) with 95% confidence intervals (95% CIs). Non-related studies, reviews, animal research, case reports, and studies with overlapping data were all excluded. Data extraction Two investigators independently extracted data. The extracted data were as follows: first author of each study, year of publication, study design, country, sample size, study period, matching criteria, adjusted factors, RRs or HRs, or ORs with 95% CIs, and histological types of EC. Disagreements were solved by consensus. Statistical analysis Pooled RRs with 95% CIs were calculated to estimate the association between hormonal and reproductive factors and EC risk by use of STATA 12.0 software (StataCorp, College Station, TX, USA). P < 0.05 suggested statistical significance under corresponding comparisons. The between-study heterogeneity was estimated by Cochran’s Q and I2 tests.10,11 If the between-study heterogeneity was significant, the random-effects model by DerSimonian and Laird method was adopted for analysis;12 otherwise, the fixed-effects model by Mantel and Haenszel method was used when the between-study heterogeneity was insignificant.13 Separate analysis by histological types and ethnicity and sensitivity analysis by omission of each study were also conducted. Begg’s and Egger’s tests were applied to assess the potential publication bias risk in our study.14,15
Association between hormonal factors and EC risk Decreased risk of EC was related to HRT (RR = 0.72, 95% CI 0.60–0.86, P < 0.001) (Table 2, Fig. 1). This association was also demonstrated between HRT and ESCC risk (RR = 0.68, 95% CI 0.48–0.97, P = 0.031), but not EAC risk (RR = 0.92, 95% CI 0.59–1.43, P = 0.031) (Table 2, Fig. 1). There was no significant relationship between OC use and the risk of EC, ESCC, and EAC (Table 2). HRT exerted protective effects on EC development among Caucasians but not Asians (Table 2). Sensitivity analysis by sequential omission of single studies did not alter the pooled results (data not shown).
Association between reproductive factors and EC risk Menopause status was associated with increased risk of EC (RR = 1.47, 95% CI 1.07–2.03, P = 0.018), particularly ESCC (RR = 1.66, 95% CI 1.12–2.48, P = 0.012) (Table 2, Fig. 2). No appreciable association between menopause status and EAC risk was investigated in our study for lack of sufficient published studies. In addition, the pooled RRs suggested that decreased risk of EC (RR = 0.79, 95% CI 0.68– 0.92, P = 0.003) and EAC (RR = 0.66, 95% CI 0.53– 0.82, P < 0.001) was related to breast-feeding (Table 2, Fig. 3). As shown in separate analysis by ethnicity, such significant associations were also observed among Caucasians, but not Asians (Table 2). In addition, older age at first birth and older age at menopause were associated with increased risk of EC in Asians (Table 2). Other reproductive factors did not modify the risk of EC, ESCC, and EAC (Table 2). Sensitivity analysis further confirmed the findings (data not shown). © 2015 International Society for Diseases of the Esophagus
© 2015 International Society for Diseases of the Esophagus
2013 2012
2012
2012 2012
2012
2011
2011
2011
2011
2011
2011
2010
2010
2005
2001
Islami F et al. Lu Y and Lagergren J
Lu Y and Lagergren J
Green J et al. Green J et al.
Green J et al.
Chen ZH et al.
Chen ZH et al.
Yu H et al.
Yu H et al.
Bodelon C et al.
Bodelon C et al.
Cronin-Fenton DP et al.
Freedman ND et al.
Lagergren J and Jansson C
Gallus S et al.
Italy and Switzerland
Sweden
USA
USA and Europe
USA
USA
China
China
China
China
UK
UK UK
Sweden
Iran Sweden
Origins
Caucasians
Caucasians
Caucasians
Caucasians
Caucasians
Caucasians
Asians
Asians
Asians
Asians
Caucasians
Caucasians Caucasians
Caucasians
Caucasians Caucasians
Ethnicity
PCC
PCC
Prospective cohort study
PCC
Prospective cohort study
Prospective cohort study
HCC
HCC
HCC
HCC
Prospective cohort study
NCC Prospective cohort study
PCC
PCC PCC
Study design
ESCC
EAC
ESCC
EAC
EAC
ESCC
EAC
ESCC
ESCC
EC
EAC
EC ESCC
EAC
ESCC ESCC
Histological types
114/425
63/141
201 506
218/862
161 080
161 080
132/132
132/132
68/157
73/157
1 319 409
1 054/5 245 1 319 409
115/1 150
149/290 363/3 630
No. (cases/controls or total n)
1984–1999
1995–1997
1995–2003
1964–2005
1993–2009
1993–2009
2008–2010
2008–2010
2004–2010
2004–2010
1996–2001
1995–2005 1996–2001
1932–2008
2003–2007 1932–2008
Study period
Age, sex, and neighborhood of residence Age, sex, education, diabetes, obesity, smoking and alcohol abuse Age, sex, education, diabetes, obesity, smoking and alcohol abuse Smoking, alcohol, and body mass index Body mass index, smoking, alcohol, strenuous exercise, use of oral contraceptives Body mass index, smoking, alcohol, strenuous exercise, use of oral contraceptives Age, reflux, smoking, alcohol, education, employment, body mass index, and intake of fresh vegetables and fruits Age, reflux, smoking, alcohol, education, employment, body mass index, and intake of fresh vegetables and fruits Age, sex, history of reflux, body mass index, smoking, and drinking status Age, sex, history of reflux, body mass index, smoking, and drinking status Age, heartburn, body mass index, smoking, hysterectomy status, study type, and treatment Age, heartburn, body mass index, smoking, hysterectomy status, study type, and treatment Age, reflux symptoms, education, smoking, alcohol, body mass index and study Age, body mass index, fruit and vegetable consumption, smoking, alcohol intake, physical activity, and total energy intake Age, reflux symptoms, body mass index, smoking, alcohol intake of fruit and vegetables, living with a partner, and socioeconomic status Age, study center, education, body mass index, nonalcohol energy intake, tobacco, and alcohol consumption
Adjusted or matching factors
EAC, esophageal adenocarcinoma; EC, esophageal cancer; ESCC, esophageal squamous cell carcinoma; HCC, hospital-based case–control studies; NCC, nested case–control studies; PCC, populationbased case–control studies.
Year
Characteristics of all studies included into the meta-analysis
Study
Table 1
Esophageal cancer risk 3
4
Diseases of the Esophagus
Table 2 Summary results of meta-analysis Pooled analyses† Contrasts EC Oral contraceptive Hormone replacement therapy Age at menarche Parity Age at first birth Menopausal status Age at menopause Breast-feeding ESCC Oral contraceptive Hormone replacement therapy Age at menarche Parity Age at first birth Menopausal status Age at menopause Breast-feeding EAC Oral contraceptive Hormone replacement therapy Age at menarche Parity Age at first birth Age at menopause Breast-feeding Caucasians Oral contraceptive Hormone replacement therapy Age at menarche Parity Age at first birth Menopausal status Age at menopause Breast-feeding Asians Hormone replacement therapy Age at menarche Parity Age at first birth Age at menopause Breast-feeding
Tests for heterogeneity‡ 2
No. of studies
RR [95%CI]
P
I (%)
P
5 8 10 9 12 3 9 7
0.84 [0.60–1.16] 0.72 [0.60–0.86] 0.96 [0.82–1.12] 0.95 [0.89–1.02] 1.10 [0.91–1.32] 1.47 [1.07–2.03] 1.24 [0.82–1.87] 0.79 [0.68–0.92]
0.290