SEARCH

SEARCH BY CITATION

Keywords:

  • oestrogen;
  • hormonal;
  • exogenous oestrogen therapy;
  • oesophageal neoplasm

Abstract

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References

There is an unexplained strong male predominance in the aetiology of oesophageal adenocarcinoma (OAC). The hypothesis that oestrogens are protective, deserves attention. A potential protective influence of exogenous oestrogen exposure, that is, hormone replacement therapy (HRT) and oral contraceptives (OC) has been addressed only in studies of limited statistical power, and the individual studies have not provided conclusive results. We conducted a systematic literature search and meta-analysis on HRT and OC and the risk of OAC. We used the databases PubMed and the Web of Science. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated by the Mantel–Haenszel random-effect method. A total of five studies were included. Compared to never users, ever users of HRT had a statistically significantly decreased risk of OAC (pooled OR = 0.75; 95% CI: 0.58–0.98), and ever users of OC had a borderline significantly decreased risk of this cancer (pooled OR = 0.76; 95% CI: 0.57–1.00). In conclusion, HRT and OC use seems to be associated with a decreased risk of OAC. However, further research is warranted.

Oesophageal adenocarcinoma (OAC) is up to nine times more common in men than women, yet the reasons behind this pattern remain unclear.[1, 2] Gastro-oesophageal reflux and high body mass index (BMI), the two main risk factors for OAC, seem to influence the risk of developing OAC similarly in both genders, and might not explain the male predominance.[3-6] The typical male fat distribution, with mainly abdominal adiposity, is a stronger risk factor for OAC than BMI alone,[7] but it is uncertain if this exposure contributes the sex ratio of OAC.[8] It has also been hypothesised that female sex hormones, mainly oestrogen, are protective. This oestrogen hypothesis is supported by a 20-year delay in the onset of OAC in women compared to men,[9] and a particularly high male-to-female ratio during women's reproductive years.[10] Individual investigations of the potential role of exogenous hormonal therapy in women, including hormone replacement therapy (HRT) and oral contraceptives (OC), in the aetiology of OAC have failed to provide conclusive results.[11-15] The low incidence of OAC in women has hampered the statistical precision in the available studies. HRT is used to mitigate discomfort caused by decreased levels of circulating oestrogen and progesterone after menopause. Combined HRT includes both oestrogens and progesterone, and is recommended to women with an intact uterus, as it counteracts endometrial hyperplasia associated with oestrogen therapy. Oestrogen only HRT is recommended to women who have undergone hysterectomy.[16, 17] The most commonly used OCs contain both oestrogen and progesterone. The often called ‘mini pill' contains progesterone only,[18] and is recommended during breastfeeding as oestrogen reduces the amount of breast milk.[19]

With the purpose of improving the knowledge of the relationship between HRT and OC exposure and risk of OAC, we conducted a systematic review and meta-analysis.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References

A systematic literature review and meta-analysis were performed, which followed an a priori established study protocol. The results are reported in accordance with the PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses).[20]

Data sources and searches

The search aimed to identify all studies including data on the associations of exogenous oestrogen, that is, HRT or use of OC, with OAC incidence in women. No language restrictions were applied. The time period was from an unbounded start date to October 25, 2013. Two scientific search engines were used: PubMed and Web of Science. To identify relevant studies, the following terms were used: (o)esophagus, (o)esophageal, cardia, or gastro-(o)esophageal junction, and neoplasms, cancer, tumo(u)r, carcinoma, or adenocarcinoma, and (o)estrogens, HRT, (o)estrogen replacement therapy, gonadal hormones, gonadal steroid hormones, sex hormones, OC, and postmenopausal hormones. Of the identified studies, reference and citation lists in the Web of Science were screened to find other potentially relevant articles that may have been missed in the database search. The first selection of the search was performed by one investigator (K.L.) under supervision of the content expert (J.L.). Assessment of eligibility of the articles remaining after exclusion of irrelevant articles was performed by mutual consideration by all authors.

Study selection

We included studies providing original data on use of HRT or OC in relation to the risk of developing OAC in women. Eligible studies were cohort studies, case-control studies and intervention studies, providing that at least five cases of OAC were identified. Cross-sectional studies were excluded, as were animal studies and studies only addressing nutritional oestrogens, endogenous oestrogen exposure and oestrogen antagonists. We also excluded studies only addressing risk of oesophageal squamous cell carcinoma or studies where analyses of squamous cell carcinoma and OAC were not reported separately.

Data synthesis and statistical analysis

Pooled risk estimates were calculated for exposure variables reported in at least two studies. Exposure to HRT or OC were analysed as dichotomous variables, that is, ever versus never use. In addition, past users compared to current users, duration and type of use of HRT or OC were analysed whenever possible. The data were extracted and analysed in STATA 12.1 using the Mantel–Haenszel random-effect method, and associations were measured in odds ratios (ORs) or hazard ratios with 95% confidence intervals (CIs). The presence of small study effects and publication bias were evaluated by funnel plots and Egger's regression asymmetry analysis.[21] Statistical heterogeneity was assessed by means of Cochran's Q test and I-squared test. The I-squared represents the percentage of variation attributable to heterogeneity, which is categorised as low (25–50%), moderate (51–75%) or high (>75%).[22]

Results

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References

Included studies

The search identified a total of 478 articles for which the titles and abstracts were scanned to determine potential eligibility for inclusion. After a selection process (summarised in Fig. 1), five studies were finally included in the meta-analysis. These studies together identified 451 women with OAC and 367,033 female controls.[11-15] All selected studies addressed HRT exposure,[11-15] while three of these also had data on OC exposure.[11-13] These five studies were published in English during the period 2006 and 2011. Two studies were cohort studies[11, 13, 14] and three were case-control studies[12, 14, 15] (Table 1):

  1. A nested case-control study based on the General Practitioners Research Database in the UK included 1,619,563 person-years of follow-up and identified 58 women with OAC and 3,191 randomly selected controls.[14] The results were adjusted for confounding by age, calendar year, tobacco smoking, alcohol consumption, BMI, hysterectomy and various upper gastrointestinal disorders.
  2. A study pooling four population-based case-control studies, conducted in Ireland,[23] UK,[24] Australia[25] and US,[26] and including 218 OACs and 862 controls.[12] Confounding by age, tobacco smoking, alcohol use, education, BMI and reflux were adjusted for.
  3. A Chinese case-control study of 44 women with OAC and 132 hospital-based controls without any malignancy. Adjustments were made for confounding by age, tobacco smoking, alcohol consumption, education and reflux.
  4. A large US cohort study including 201,506 women in the National Institute for Healthcare-American Association of Retired Persons NIH-AARP Diet and Health cohort,[13] study with 65 cases of OAC. The study adjusted for influence of age, tobacco smoking, alcohol intake, education, BMI, physical activity and intakes of fruit, vegetables and total energy.
  5. A cohort study based on the Women's Health Initiative clinical trials and observational studies in the US (WHI), including 161,080 postmenopausal women and 23 cases of OAC.[11] The WHI included both a randomised controlled trial and an observational cohort study, both examining effects of HRT and OC, where data were analysed combined. The study adjusted the results for confounding by age, study type, ethnicity, BMI, reflux and hysterectomy.
image

Figure 1. PRISMA flow chart of the selection of relevant studies for this meta-analysis. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Table 1. Characteristics of the studies included in this meta-analysis of use of hormone replacement therapy (HRT) and oral contraceptives (OC) in relation to risk of developing oesophageal adenocarcinoma
First author and year of publicationStudy designStudy periodCountry in which the study was conductedStudy name or target populationNumber of cases/controlsExposures of interestComparison groupAge range in years (menopausal status)
  1. Abbreviations: RCT: Randomised Control Trial, HRT: Hormone Replacement Therapy, OC: Oral Contraceptives, BMI: Body Mass Index, WHI: Women's Health Initiative, NIH: National Institutions for Health, AARP: American Association of Retired Persons.

Bodelon et al., 201111Prospective cohort and RCT1993–1998USAWomen's Health Initiative23/161,057HRT and OCParticipants in the WHI study50–79 (all postmenopausal)
Cronin-Fenton et al., 201012Case control2002–2004Ireland, UK, Australia, USAFINBAR, The UK women's study, The Australian Cancer Study, The US Kasier Permanent study218/862HRT and OCPopulation-based controls and Hospital-based controls(both premenopausal and postmenopausal)
Freedman et al., 201013Prospective cohort1995–1996USANIH-AARP Diet and Health cohort25/201,481HRT and OCParticipants in the NIH-AARP study50–71 (both premenopausal and postmenopausal)
Lindblad et al., 200614Case control1994–2001UKGeneral Practice Research Database (GPRD)299/3191HRTRandomly selected controlsMean age 74 for controls and cases (both premenopausal and postmenopausal)
Yu et al., 201115Case control2008–2010ChinaShandong province44/132HRTHospital-based controlsMean age 57.2 for esophageal cancer and 56.8 for controls (both premenopausal and postmenopausal)

A further quality assessment of the included studies is presented in Figure 2. All included studies had representative populations, well-defined criteria for inclusion and exclusion, reported ever or never use of HRT and OC, and adjusted the results for BMI. Four of the five studies also adjusted the results for reflux. None of the studies exclusively assessed risk of OAC, but all studies exclusively investigated upper gastrointestinal cancer as the aoutcome.

image

Figure 2. Quality assessment of the five included studies in this meta-analysis. The quality variables are listed on the Y-axis, whereas the X-axis represents the percentages of the five studies that fulfilled the variables asked for. A 0% on the X-axis means that none of the five studies fulfilled this criteria, whereas a 100% bar means that all studies fulfilled this criteria. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Hormone replacement therapy

For the analysis of HRT use, data from all five studies were included.[11-15] Study-specific point ORs of ever HRT use versus never HRT use ranged from 0.62 to 0.96 but each of the 95% CIs included OR = 1. The pooled OR showed a statistically significant 25% decreased risk of OAC among ever users of HRT compared to never users (OR 0.75, 95% CI: 0.58–0.98). The between-study heterogeneity was low (I2 = 0%; Fig. 3). There was limited evidence of publication bias or small study effects bias (p = 0.054; funnel plot not shown).

image

Figure 3. Forest plot-ever versus never use of Hormone Replacement Therapy. Abbreviations: HRT: Hormone Replacement Therapy, RCT: Randomised Control Trials, OR: Odds Ratio. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Information on the duration of HRT, expressed as less or more than 10 years was available only in two studies.[11, 13] Women with less than 10 years of HRT use compared to never users had statistically nonsignificantly reduced point risk estimates of OAC in both studies.[11, 13] Among women with more than 10 years use results were contradictory.

Among the two studies that reported data on current and past use of HRT and analysed the two types of HRT separately, no statistically significant associations were found.[11, 13]

Oral contraceptives

In the three included studies that examined the association between use of OC and risk of OAC,[11-13] the study-specific ORs for OC users compared to non-OC users ranged from 0.70 to 0.88. The pooled OR was 24% decreased and of borderline statistical significance (OR 0.76, 95% CI: 0.57–1.00). There was a low heterogeneity across studies (I2 = 0%; Fig. 4). There was no evidence of publication bias or small study effects bias (p = 0.746; funnel plot not shown).

image

Figure 4. Forest plot-ever versus never use of Oral Contraceptives. Abbreviations: OC: Oral Contraceptives, OR: Odds Ratio. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Only one study analysed duration of OC use[13] and no meta-analysis could be conducted.

No study reported data comparing past and current use of OC or different types of OC and the risk of OAC.

Discussion

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References

This meta-analysis, based on five relevant studies, together including over 300 patients with OAC from five countries, found a 25 and 24% reduced risk of OAC among women using HRT and OC.

This is, to the best of our knowledge, the first study to show a statistically significantly reduced risk of OAC among HRT users. The association between OC and OAC was less robust in terms of the number of studies and the borderline statistical significance, but the available data indicate an association of similar strength to that found among HRT users.

This meta-analysis was limited by the low number of studies, and there were insufficient data to allow subgroup analysis of duration, type or dosage of HRT and OC use or adjust for potential confounding. The rough categorisation of HRT or OC use into ever or never use should, however, dilute true effects rather than enhance them, as the misclassification following such categorisation was likely to be similar among cases and controls. As with any meta-analysis, we cannot exclude the possibility that other studies may have been missed during our literature search, or that studies that observed null effects were absent from the literature altogether. Nevertheless, the funnel plot and Egger's regression asymmetry analysis found only little evidence of presence of small study effects and publication bias. Although the included studies differ in several aspects, including study design, the studies showed statistical homogeneity. The clinical heterogeneity, that is, the comparability in terms of study participants, exposure and outcome measures and other variables, is an inherent problem of meta-analyses, yet this should be limited by virtue of the strict inclusion criteria. In addition, potential confounding factors, that is, BMI and reflux, could not be taken into account due to the limited availability of data. However, it is unlikely that these factors would be strongly associated with use of HRT or OC, and thus, should not act as important confounders. Even though the included studies had similar exclusion and inclusion criteria as well as recruitment procedure, there was variability in the exposure and patient characteristics within and between the included studies. Due to the limited number of studies, we could not analyse the impact of the study design or design-specific biases, that is, recall bias and other information biases.

There is, to the best of our knowledge, no previous meta-analysis of exogenous oestrogen therapy and OAC. However, some meta-analyses indicate that HRT might decrease the risk of gastric adenocarcinoma,[27, 28] colorectal adenocarcinoma[28] and oesophageal cancer of any histological type.[27] However, regarding oesophageal cancer it is, crucial to separate adenocarcinoma and squamous cell carcinoma of the oesophagus in any aetiological research, as these histological types have very different patterns of incidence and aetiology, including a greater level of male predominance in OAC.[28]

There are several mechanisms that might explain a possibly preventive effect of exogenous oestrogens in OAC development.[29-31] The serum oestradiol concentration is much higher in premenopausal women (produced mainly in the ovaries) than in men,[32] but decreases substantially after menopause, and ultimately becomes lower than in elderly men.[33] This pattern could explain the 20-year delay in OAC incidence among women.[9] The presence of oestrogen receptors has repeatedly been shown in OAC[34-37] and in oesophageal tissue.[34, 35, 37, 38] Involvement of oestrogen signalling in regulation of adipose tissue metabolism indicates a possible link between the effects of oestrogen and abdominal adiposity (male obesity), one of the main risk factors for OAC.[32, 39, 40] Moreover, a recent study suggests that OAC and Barrett's oesophagus cells respond to treatment with selective oestrogen receptor ligands, resulting in decreased cell growth and apoptosis.[40] Fat cells are a major source of endogenous oestrogen among postmenopausal women. If HRT protects against OAC, it might be expected that the association between body mass and OAC is weaker for women than men. The adverse effects of high fat levels in women might, to some extent, be offset by their higher oestrogen levels.

We restricted our meta-analysis to assess risk of invasive OAC. However, given the recognised multistep progression from gastro-oesophageal reflux, erosive oesophagitis, Barrett's oesophagus with and without dysplasia, to OAC, it would be valuable also to investigate if oestrogens influence these earlier steps in the carcinogenic pathway. Paradoxically, the risk of gastro-oesophageal reflux disease has been found to be increased among women using HRT,[41-43] despite the fact that such reflux is a main risk factor for OAC.[5] However, it is possible that the increased risk of reflux and the decreased risk of OAC among HRT users are explained by different timing of HRT exposure. oestrogen might, for example, prevent OAC only at a later stage in the tumour progression.

There is a need for more research before any role of exogenous oestrogen exposure in the aetiology of OAC is established. However, if such research shows that HRT substantially reduces the risk of OAC, there is a potential for future research also addressing oestrogen therapy in highly selected risk groups of OAC, for example, patients with dysplastic Barrett's oesophagus, or as adjuvant therapy in patients with OAC, although side effects must be considered before such research is initiated. Our study showed a reduced risk of developing OAC of about 25% by HRT use in postmenopausal women, and the true risk estimates might well be lower considering the dilution of risk estimates typically followed by rough categorisation of the exposure.[44]

In conclusion, this systematic review and meta-analysis suggest that use of HRT, and possibly also OC, decreases the risk of OAC, thus providing support for the hypothesis that exogenous oestrogen exposure counteracts OAC in women. These findings require more research before any association can be established.

References

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. References