Electronic databases MEDLINE, Web of Science, EMBASE and Cochrane were searched up until February 11, 2016. Search strategies were developed using Medical Subject Heading (MeSH) words such as: head and neck neoplasms, head and neck and oesophageal cancer, esophageal neoplasms, HRT, female hormone, early menopause and text words related to hormones and HNC or (o)esophageal cancer. Only search terms describing the association were used. Reference lists were also extensively searched and relevant papers obtained.

Randomized controlled trials (RCTs), controlled (non-randomised) clinical trials (CCTs) or cluster trials, prospective and retrospective comparative cohort studies, case-control or nested case-control studies, and cross-sectional studies, addressing the question of female hormones as a risk factor for HNC or oesophageal SCC, were considered. Studies were included when they: i) examined the general adult population (age >18 years), specifically patients with at least 50 cases of HNC/oesophageal SCC and healthy controls; ii) addressed the question of HRT or reproductive factors (menopause) and HNC/oesophageal SCC; iii) administered HRT as an intervention for prevention of cancer or being therapeutically taken due to symptoms of menopause; iv) collected data on age at menopause, smoking, alcohol, age and socioeconomic status or educational attainment; and v) reported odds ratios, risk ratios or incidence/prevalence of HNC or oesophageal SCC defined using the World Health Organisation (WHO) classification of diseases ICD-10 codes, C00-15 and C30-31.

Cohort studies were only eligible when follow-up time was at least 5 years; case series and case reports were excluded. Only English language articles published in peer-reviewed journals, from 1948 to 2016, were considered.

Titles and/or abstracts of studies retrieved using the search strategy and those from additional sources were independently screened by two review authors (C.M. and M.M.), to identify studies that potentially met the inclusion criteria outlined above. Studies combining HNC with oesophageal squamous cell cancers were considered, but data were extracted to separate HNC and oesophageal cancer where possible. Data were extracted in all forms (e.g. dichotomous, continuous) as reported in the included studies. The full texts of these potentially eligible studies were independently assessed for eligibility by C.M. and M.M. Any disagreement over the eligibility of particular studies was resolved through discussion with a third reviewer (J.K.F.).

The risk of bias was assessed using the Newcastle Ottawa Scale which includes the following areas of evaluation of the risk of bias: selection of patients, comparability of groups in the study, methods for assessing outcomes, proof of exposure and appropriate follow-up. Studies were categorised as low, medium, high or unclear risk of bias, using a star-based scoring system. A maximum of 9 stars are available; the higher the number of stars the lower the risk of bias. For ease of interpretation, a score of 7 or greater is considered ‘low-risk’ of bias, 4–6 is ‘medium-risk’ and 3 or below is considered ‘high-risk’ of bias.

Three papers (1 cohort study and 2 case-control studies) addressed the question of the use of HRT and incidence of HNC. Only the study by Freedman et al (8) was considered at ‘low-risk’ of bias, with a Newcastle Ottawa score of 7/9 stars. This was a cohort study conducted in the USA, using the NIH-AARP Diet and Health Cohort, of 125,887 women. Two hundred and ninety-seven cases of HNC were identified with mean follow-up of 7.5 years. The risk of HNC was 22% lower for ever-users of HRT (HR, 0.78; 95% CI, 0.61–0.99); 44.1% of cases (n=127) had ever-used HRT compared to 54.6% of controls (n=106,934). Further analysis by hysterectomy status revealed that the risk reduction was greatest for women with an intact uterus who were current users of HRT for >5 years (HR, 0.23; 95% CI, 0.09–0.57). Notably, use of HRT other than oestrogen-alone or oestrogen-progesterone therapy conferred a greater risk of HNC (HR, 2.31; 95% CI, 1.15–4.65); however, this analysis was based on only 9 cases who used an alternative HRT. Two case-control studies were considered at medium/high risk of bias; they reported a non-significant reduction in the risk of HNC for ever-users of HRT (4,9).

Four studies analysed HRT use and the risk of oesophageal SCC. Two were considered low-risk of bias (8,13) and two were medium/high risk of bias (10,11). Although, all studies reported an effect estimate of <1 for users of HRT, implying a protective effect, none of the results were statistically significant.

Two studies (8,9) assessed the link between age at menopause and risk of HNC. Bosetti et al (9) found a protective effect of later age at menopause (>50 years), with an OR of 0.46 (95% CI, 0.30–0.70). Freedman et al (8) found no significant effect on the risk of HNC with later age at menopause (>55 years).

Four out of 5 studies reported a significant effect of age at menopause and risk of oesophageal SCC. The method of reporting varied. Gallus et al (10) reported an OR of 0.43 (95% CI, 0.22–0.83) for age at menopause of >50 years vs. menopause at age <45 years. Yu et al (11) reported an increased risk of oesophageal SCC for women entering menopause at <45 years (OR, 2.27; 95% CI, 1.03–4.97) and for 45–49 years (OR, 2.16; 95% CI, 1.14–4.78) compared to menopause at age >50 years. Green et al (7) reported increased risk of oesophageal SCC for every 5 years younger a women was at the time of menopause (RR, 1.32; 95% CI, 1.11–1.56). Although, Freedman et al (8) found no significant effect for individual age categories, they did observe a significant trend (p=0.019) for a lower risk of oesophageal SCC with older age at menopause. Chen et al (12) observed no significant effect for age at menopause, although these authors classified older age at menopause as >48 years.

We found that earlier age at menopause is associated with a higher risk of oesophageal cancer, based on 4 studies with a total of 836 cases of oesophageal SCC.

Most women experience menopause between the ages of 45 and 55 years; the median age at menopause is 47.2 years, according to a prospective cohort study of over 5,000 women enrolled at the Royal College of GP's Oral Contraception study (15). Menopause is considered to be early in women aged 40–45 years (~5% of women) and premature in women <40 years (~1% of women) (16).

Early menopause is more frequent in women with certain genetic or autoimmune disorders, infections or a history of chemotherapy/radiotherapy or surgery to remove the ovaries (16).

Mean age of menopause for smokers is significantly lower than non-smokers (45.6 vs. 46.9 years) (15). Women with early natural menopause are more likely to be smokers, ever-users of oral contraceptive medication, undergone tubal ligation, have at least one episode of endometriosis and are less likely to use HRT. No association with alcohol, BMI, physical activity or parity was reported (15). Diabetic women have also been found to be at risk of early menopause (OR, 2.76; 95% CI, 1.32–5.66) (17). In a pooled analysis of case-control studies, diabetes diagnosed at age <50 years conferred a greater risk of HNC (OR, 1.37; 95% CI, 1.07–1.74) when analysing 6,448 cases of HNC and 13,747 controls (18), but no link with age at menopause was considered in the present study.

A recent systematic review and meta-analysis of oesophageal SCC and reproductive factors also found a protective effect for older age at menopause (SRR=0.70; 95% CI, 0.51–0.95). The authors concluded that ‘properly extending the time of menstruation for pre-menopausal women is a possible way to reduce the risk of oesophageal SCC’ (19). A separate meta-analysis found that menopausal status was associated with higher risk of oesophageal SCC (RR, 1.66; 95% CI, 1.12–2.48), but age at menopause was not significant for Caucasians. In Asians, older age at menopause was found to confer greater risk of oesophageal SCC (RR, 1.68; 95% CI, 1.05–2.70) (6). This implies that ethnicity and genetic background may influence the role of female hormones in oesophageal carcinogenesis.

The effect of age at menopause has been explored in several other cancer types. A meta-analysis of 8 studies of gastric cancer and reproductive factors found a protective effect of longer years of fertility (RR=0.74; 95% CI, 0.63–0.86) (20). Early menopause has also been found to be a risk factor for bladder cancer (meta-analysis including 7 studies: RR=1.59; 95% CI, 1.31–1.92) (21). Two studies from Hungary, from the same unit, found that the mean age at menopause was significantly lower in females with oral cancer compared to controls (43.5 vs. 50.9 and 45.4 years vs. 51.3 years, respectively) (22,23). These studies were excluded from this review due to the failure to provide accurate effect estimates with confidence intervals. Other recent meta-analyses found no significant effect of age at menopause in relation to risk of brain tumours (24), pancreatic (25) or thyroid cancer (26). A large cohort study found no association with age at menopause and risk of colorectal cancer (27).

HRT is long-established in the management of the symptoms of the menopause and has also been shown to reduce the risk of osteoporotic fractures, cardiovascular disease, Alzheimers, depression, stroke and colon cancer. Approximately 30% of UK women used HRT in 2001–2002 (28). Following this, a large US-based trial (Women's Health Initiative) was prematurely halted due to concerns over the evidence of increased risk of breast cancer, coronary heart disease, stroke and pulmonary embolism among users of HRT (29). The UK-based Million Women Study (MWS) also reported increased risk of breast cancer with HRT in 2003 (30). Following media coverage of the results of these trials, use of HRT declined steadily in the UK for the next 3–4 years. In 2005, only 10–11% of menopausal women were using HRT (28). Further reports of a slight increased risk of ovarian cancer for ever-users of HRT compounds enhanced concerns over the safety of HRT (31).

However, concerns have been raised by some authors surrounding the reporting of the WHI and MWS trial results, and the fact that little coverage was given to the evidence of the reduced incidence of osteoporotic fractures and colon cancer (32). Both trials recruited women aged over 50 years; therefore, the results cannot be applied to women who undergo premature menopause (33). HRT for women with premature menopause (primary ovarian insufficiency), prescribed up to the age of natural menopause (~51 years), is endorsed by the British Menopause Society and NICE guidelines. The NICE guidelines also recommend the development of a collaborative ‘primary ovarian insufficiency’ registry to allow data collection to clarify, among other factors, the long-term risk of cancers in this group (34).

Despite reports of an increased risk of breast and ovarian cancers in certain groups using HRT, there are many observational studies reporting a reduced risk of other cancers in HRT users. There are several reports of a reduced risk of colon cancer with HRT (27,35,36). Green et al reported a relative risk reduction of 17% (RR, 0.83; 95% CI, 0.79–0.88) in their meta-analysis of 30 studies of colorectal cancer and HRT (36). In a recent meta-analysis of oesophageal cancer and reproductive factors, the authors reported a 33% relative risk reduction with HRT use (SRR, 0.67; 95% CI, 0.56–0.81) (19). Similar results were reported for reduced risk of gastric cancers (RR, 0.77; 95% CI, 0.64–0.92) (20). Combined HRT (oestrogen and progesterone) has also been shown to reduce the risk of lung cancer, in a meta-analysis of 25 studies (RR, 0.83; 95% CI, 0.76–0.91) (37). HRT has not been found to significantly reduce the risk of melanoma or pancreatic cancer (25,38).

This systematic review did not find evidence of a significant risk reduction for oesophageal cancer among users of HRT, in contrast to the findings of the meta-analysis by Zhu et al (19). However, all effect estimates were <1 and the studies included contained only modest numbers of oesophageal cancer cases. The Freedman et al (8) cohort study included 297 cases of HNC and was the only study to report a significant relative risk reduction for HNC in users of HRT (HR, 0.78; 95% CI, 0.61–0.99). Further large studies are needed to confirm the finding that HRT is protective against HNC.

Confounding factors must be considered. Users of HRT tend to be of higher socioeconomic status (SES) and have higher levels of education. Both of these factors may reduce the risk of HNC. To address this, only studies controlling for a measure of SES or education were eligible for inclusion in this review. Freedman et al (8) controlled for education, alcohol, BMI, tobacco smoking, physical activity and diet (fruit and vegetable intake), although residual confounding factors could still be relevant.

Females have been found to have survival advantage for HNC, oesophageal, gastric and pancreatic cancer, as well as cancers at 11 other sites. For all cancers combined, women have a 5% lower risk of death than men; for HNC, a 12% increased survival is reported (39). This finding is consistent across all European regions in the EUROCARE-4 cohort of 1.6 million population-based cancer cases (39). The advantage is most pronounced in younger women and declines with age, with a marked decline beyond the age of menopause. It is possible that female hormones play a part in this 12% improvement in survival.

Oestrogen is known to promote cancer in oestrogen-responsive tissues, such as breast, endometrium and cervix; however, evidence from mouse models suggests that oestrogen has an inhibitory role in oesophageal SCC growth (40). Oestrogen receptors have been found in oesophageal SCC tissue samples and HNCs (40–42). Oesophageal SCC cells with oestrogen receptors have been shown to be inhibited by oestrogen exposure and this may initiate apoptosis (43). If oestrogen is responsible for inhibiting the growth of some cancer cells, oestrogen deficiency could be considered a risk factor for certain cancers. A women who undergoes premature menopause has less oestrogen exposure over her lifetime and this may increase her risk of cancers such as oesophageal SCC. However, oestrogen appears to have both tumorigenic and antitumour properties, depending on the tissue and presence of oestrogen receptors. HNC cell lines, from males and females, have been found to contain oestrogen receptors, and laboratory studies appear to show that oestrogen promotes the growth of HNC cells (44,45). Further high quality basic science studies are required to confirm the role of oestrogen in HNC and oesophageal SCC.

The studies included in this review were all assessed for risk of bias, using the Newcastle Ottawa Scale. Only 3 of the 8 studies included were low risk for bias. This does limit the significance of our findings and is an indication of the need for further, high quality, studies addressing the issue of female hormones and squamous cancers.

We only included studies that controlled for significant potential confounding factors, such as smoking and socioeconomic status, however, the risk of residual confounding factors remains. Smokers undergo menopause at a younger age than non-smokers. We recommend that future studies should control for this important confounder.

The rationale for combining HNC and oesophageal SCC is based on the strong similarities in their epidemiology and aetiology as mentioned earlier. We deliberately excluded oesophageal adenocarcinomas as these cancers have quite different aetiology. One of the studies included in our review by Bosetti et al (9) failed to clarify the histology of the oral/pharyngeal cancer cases included, which introduces a potential source of bias. However, over 90% of oral cancers and >80% of pharyngeal cancers are of squamous histology (46), therefore the bias is unlikely to be significant.

We found that earlier age at menopause is a risk factor for oesophageal SCC, with women entering menopause at <45 years having double the risk of those entering menopause at age >50 years. Similar, but less striking, results were observed for HNC. HRT was found to reduce the risk of HNC/oesophageal SCC, but the evidence is inconclusive. We used strict eligibility criteria and only considered studies that controlled for other risk factors, however, there is still the risk of residual bias. Consideration should be given to collecting data on reproductive factors and exposure to HRT, as routine practice, in future epidemiological and clinical studies of these cancers. The concept of oestrogen deficiency as a risk for HNC/oesophageal SCC warrants further investigation in future laboratory and clinical studies.