Article Text

Download PDFPDF

Original article
Postmenopausal hormone therapy and colorectal cancer risk by molecularly defined subtypes among older women
  1. David Limsui1,
  2. Robert A Vierkant2,
  3. Lori S Tillmans3,
  4. Alice H Wang2,
  5. Daniel J Weisenberger4,
  6. Peter W Laird4,
  7. Charles F Lynch5,
  8. Kristin E Anderson6,
  9. Amy J French3,
  10. Robert W Haile7,
  11. Lisa J Harnack8,
  12. John D Potter9,
  13. Susan L Slager2,
  14. Thomas C Smyrk3,
  15. Stephen N Thibodeau3,
  16. James R Cerhan10,
  17. Paul J Limburg1
  1. 1Division of Gastroenterology & Hepatology, Mayo Clinic, Rochester, Minnesota, USA
  2. 2Division of Biomedical Statistics & Informatics, Mayo Clinic, Rochester, Minnesota, USA
  3. 3Department of Laboratory Medicine & Pathology, Mayo Clinic, Rochester, Minnesota, USA
  4. 4USC Epigenome Center, Norris Comprehensive Cancer Center, Los Angeles, California, USA
  5. 5Department of Epidemiology, University of Iowa, Iowa City, Iowa, USA
  6. 6Department of Epidemiology, University of Minnesota, Minneapolis, Minnesota, USA
  7. 7Department of Preventive Medicine, Keck School of Medicine of USC, Los Angeles, California, USA
  8. 8Masonic Cancer Center, University of Minnesota, Minneapolis, Minnesota, USA
  9. 9Fred Hutchinson Cancer Research Center, Seattle, Washington, USA
  10. 10Division of Epidemiology, Mayo Clinic, Rochester, Minnesota, USA
  1. Correspondence to Dr Paul J Limburg, 200 First Street SW, Rochester, MN 55905, USA; limburg.paul{at}mayo.edu

Abstract

Background Postmenopausal hormone (PMH) therapy may reduce colorectal cancer (CRC) risk, but existing data are inconclusive.

Objectives To evaluate associations between PMH therapy and incident CRC, overall and by molecularly defined subtypes, in the population-based Iowa Women's Health Study of older women.

Methods Exposure data were collected from Iowa Women's Health Study participants (55–69 years) at baseline (1986). Archived, paraffin-embedded tissue specimens for 553 CRC cases were collected and analysed to determine microsatellite instability (MSI-L/MSS or MSI-H), CpG island methylator phenotype (CIMP-negative or CIMP-positive) and BRAF mutation (BRAF-wildtype or BRAF-mutated) status. Multivariable Cox regression models were fit to estimate RRs and 95% CIs.

Results PMH therapy (ever vs never use) was inversely associated with incident CRC overall (RR=0.82; 95% CI 0.72 to 0.93), with a significantly lower risk for MSI-L/MSS tumours (RR=0.75; 95% CI 0.60 to 0.94), and borderline significantly lower risks for CIMP-negative (RR=0.79; 95% CI 0.63 to 1.01) and BRAF-wildtype (RR=0.83; 95% CI 0.66 to 1.04) tumours. For PMH therapy >5 years, the subtype-specific risk estimates for MSI-L/MSS, CIMP-negative and BRAF-wildtype tumours were: RR=0.60, 95% CI 0.40 to 0.91; RR=0.68, 95% CI 0.45 to 1.03; and RR=0.70, 95% CI 0.47 to 1.05, respectively. PMH therapy was not significantly associated with the MSI-H, CIMP-positive or BRAF-mutated CRC subtypes.

Conclusions In this prospective cohort study, PMH therapy was inversely associated with distinct molecularly defined CRC subtypes, which may be related to differential effects from oestrogen and/or progestin on heterogeneous pathways of colorectal carcinogenesis.

  • Postmenopausal hormone therapy
  • colorectal cancer
  • cohort study
  • molecular epidemiology
  • microsatellite instability
  • CpG island methylator phenotype
  • BRAF
  • KRAS
  • cancer epidemiology
  • cancer genetics
  • epidemiology
  • statistics
  • epidemiology
  • gastrointestinal neoplasia
  • pancreatic cancer
  • epidemiology
  • cancer prevention
  • cancer epidemiology
  • chemoprevention
  • Helicobacter pylori
  • acid-related diseases
  • non-ulcer dyspepsia
  • genetic polymorphisms
  • gastric neoplasia

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Significance of this study

What is already known on this subject?

  • Postmenopausal hormone (PMH) therapy could be used to complement colorectal cancer (CRC) prevention strategies focused on diet, lifestyle and screening behaviours.

  • Existing data regarding the association between PMH and CRC risk are inconclusive.

  • Data referent to associations between PMH therapy and molecularly defined CRC subtypes are limited and inconsistent.

What are the new findings?

  • PMH therapy was inversely associated with incident CRC, with statistically significant trends by recency and duration of exposure, in this large prospective, population-based cohort study of older women.

  • PMH therapy was associated with a statistically significant lower risk for MSS/MSI-L tumours, as well as borderline statistically significant risk reductions for CIMP-negative and BRAF-wildtype tumours among women with prolonged exposure to PMH therapy (ie, >5 years duration); conversely, PMH therapy-related risk estimates for the MSI-H, CIMP-positive and BRAF-mutated CRC subtypes were not statistically significant.

  • These data suggest that PMH therapy may have more pronounced inhibitory effects on the ‘traditional’ pathway, as compared to the serrated or alternate pathways, of colorectal carcinogenesis.

How might it impact on clinical practice in the foreseeable future?

  • These findings may have direct implications for clinical practice, for example with respect to future application of PMH-based chemoprevention based on clinicopathological characteristics that predict the highest anticipated risk:benefit ratio, as well as for further translational research regarding the effects of PMH dose, duration and formulation on molecular mechanisms of colorectal carcinogenesis.

Introduction

Colorectal cancer (CRC) is the third most common malignancy among women in the USA, with 69 360 new cases projected in 2011.1 Encouragingly, emerging trends show that CRC incidence rates are declining for women (and men), with computer simulation models estimating essentially equal benefits from CRC screening and risk factor modification.2 Among older women, postmenopausal hormone (PMH) therapy could be used to complement CRC prevention strategies focused on diet, lifestyle and screening behaviours. Based on an early meta-analysis of 18 observational studies, Grodstein et al found that PMH therapy was associated with a 20% decrease in colon cancer risk, particularly among women who described themselves as current PMH users (RR=0.66; 95% CI 0.59 to 0.74).3 Subsequent epidemiological studies have observed generally similar, although not entirely consistent, inverse associations between PMH therapy and CRC risk (as recently reviewed),4 indicating that exogenous oestrogen and/or progestin compounds may inhibit the development of colorectal neoplasia.

To date, a limited number of randomised, controlled trials have examined PMHs as candidate CRC chemopreventive agents, with mixed results. In the Heart and Estrogen/Progestin Replacement Study (HERS), women with coronary disease were randomly assigned to receive oestrogen plus progestin versus placebo for an average follow-up period of 4.1 years. Incident CRC was analysed as a secondary outcome, with an observed HR of 0.69 (95% CI 0.32 to 1.49), suggesting a potentially protective effect.5 ,6 In the Women's Health Initiative clinical trial, intervention with oestrogen plus progestin yielded an even more striking 44% reduction in incident CRC (HR=0.56; 95% CI 0.38 to 0.81; mean follow-up period 5.6 years), while oestrogen alone did not appear to affect CRC risk (HR=1.12; 95% CI 0.77 to 1.63; mean follow-up period 7.1 years).7 ,8 However, further analyses of the Women's Health Initiative data revealed that women assigned to the oestrogen plus progestin arm were more likely to be diagnosed with advanced stage CRC, raising the possibility of differential effects from PMH therapy on heterogeneous pathways9–11 of colorectal carcinogenesis.

At present, data referent to associations between PMH therapy and molecularly defined CRC subtypes are limited and inconsistent.12–14 Thus, we sought to evaluate associations between PMH therapy and incident CRC, overall and by microsatellite instability (MSI), CpG island methylator phenotype (CIMP) and BRAF mutation status, among women enrolled in the prospective, population-based Iowa Women's Health Study (IWHS). The current report updates and extends previously described associations between PMH exposure and CRC risks based on IWHS data3 ,15 by including additional follow-up time and molecularly defined, subtype-specific analyses.

Materials and methods

This study was reviewed and approved by the institutional review boards for human research of the University of Iowa, University of Minnesota and Mayo Clinic.

Study population

The methods used for recruiting IWHS subjects have been previously described.16 In brief, a 16-page questionnaire was mailed in January 1986 to randomly selected women, age 55–69 years, who resided in Iowa and held a valid driver's licence at baseline. Of the 98 029 women who received questionnaires, 41 836 (43%) responded. CRC rates have been shown to be similar between the baseline questionnaire responders and non-responders.17 For the present study, exclusion criteria (not mutually exclusive) were history of malignancy other than skin cancer (n=3830), unable to be followed longitudinally for ≥1 day (n=10), not postmenopausal (n=569) or incomplete oestrogen data (n=200), leaving 37 285 women in the final analytic cohort.

Exposure assessment

Demographic, dietary, lifestyle, medication and other CRC risk factor data were collected by self-administered questionnaire during the IWHS baseline evaluation (1986). PMH therapy was determined by the following question: “Have you ever used pills other than birth control pills which contain oestrogen or other female hormones (eg, at the change of life or menopause, after surgery, or at any other time)?” with response levels of yes, currently; yes, but not currently; and never. Subjects who responded yes (currently or not currently) were then asked: “How long did you take oestrogens or other female hormone pills (other than birth control pills)?” with response levels of 1 month or less; 2–6 months; 7–12 months; 13 months–2 years; 3–5 years; and more than 5 years. Further information regarding specific type and dose of PMH therapy was not obtained.

Ascertainment of incident CRC cases

Follow-up questionnaires were mailed in 1987, 1989, 1992, 1997 and 2004 to update vital status. Participants who did not respond to the follow-up questionnaires were checked against the National Death Index to identify decedents. Incident CRC cases were identified through the Iowa Cancer Registry, which participates in the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) programme.18 Annual matching between a computer-generated list of all cohort members and the records of Iowans with incident cancer in the SEER programme registry was performed using combinations of first, last and maiden names; zip code; birth date; and social security number. CRC cases were identified using International Classification for Diseases in Oncology (ICD-O) codes of 18.0, 18.2–18.9, 19.9 and 20.9. Stage was categorised according to SEER definitions for localised, regional or distant disease. Cancers located in the caecum, ascending colon, hepatic flexure, transverse colon and splenic flexure were categorised as proximal CRC. Cancers located in the descending colon, sigmoid colon, rectosigmoid junction and rectum were categorised as distal CRC.

Tissue collection and processing

Beginning in 2006, archived, paraffin-embedded tissue specimens were requested from incident CRC cases diagnosed through 31 December 2002, as previously described.19 Tissue specimens were retrieved for 732/1255 (58%) subjects with incident CRC, with all cases confirmed by a single gastrointestinal pathologist (TCS). Comparisons of general demographics and tumour characteristics (size and stage) between incident CRC cases with retrieved versus non-retrieved tissue specimens were not statistically significant (p>0.05 for each comparison; data not shown). Paraffin blocks were serially cut into 5- or 10-μ thick sections. One slide was stained with H&E and areas of neoplastic (>50%) and juxtaposed normal tissue were identified. Tumour and normal tissues were scraped from unstained slides and placed into separate tubes for DNA extraction using the QIAamp Tissue Kit (QIAgen, Valencia, California, USA), according to the manufacturer's instructions. A total of 179 retrieved CRC cases were subsequently excluded from the present study due to inadequate/unusable tissue from the first primary CRC, multiple primary CRCs at initial diagnosis, or incomplete PMH therapy data, leaving 553 incident CRC cases for the defined molecular analyses.

Microsatellite instability

MSI testing was performed on paired tumour and normal DNA samples for each subject, using 10 established markers: four mononucleotide repeats (BAT25, BAT26, BAT40 and BAT34C4), five dinucleotide repeats (ACTC, D5S346, D18S55, D17S250 and D10197), and one complex marker (MYCL).20 Incident CRCs were classified as MSI-H if ≥30% of the markers demonstrated instability, and as MSI-L or microsatellite stable (MSS) if <30% of the markers demonstrated instability.21 ,22 MSI status could be determined for 539/553 incident CRC cases (98%).

CpG island methylation

Tumour DNA was treated with sodium bisulphite and subsequently analysed using automated real-time PCR-based MethyLight to amplify methylated CpG sites in the promoter regions of an established five-gene panel (CACNA1G, IGF2, NEUROG1, RUNX3 and SOCS1).23 CIMP status was reported as CIMP-positive or CIMP-negative, as defined by promoter hypermethylation in ≥3 or 0–2 genes in the five gene panel, respectively. CIMP status could be determined for 525/553 incident CRC cases (95%).

BRAF mutation

Tumour DNA was analysed using fluorescent allele specific PCR to detect the V600E point mutation in exon 15 of the BRAF gene. BRAF-mutation and BRAF-wildtype cases were defined by the presence or absence of the V600E point mutation, respectively. BRAF mutation status could be determined for 536/553 incident CRC cases (97%).

Statistical analyses

Data were descriptively summarised using frequencies and percentages for categorical variables, and means and SDs for continuous variables. Among cases, we assessed pairwise agreement between the various biomarker values using κ coefficients. Follow-up was calculated as the time from completion of the baseline questionnaire until either incident CRC diagnosis or censorship, as defined by a move from Iowa or death. Otherwise, study participants were assumed to be alive, cancer-free and living in Iowa through 31 December 2002. Cox proportional hazards regression analysis was used to estimate RRs and 95% CIs for associations between the PMH exposures of interest and incident CRC. All eligible IWHS subjects were included in these Cox regression analyses, regardless of eventual cancer status. Incidence was modelled as a function of age.24 PMH therapy at baseline was analysed with respect to exposure status (ever, never), with ever users further categorised with respect to recency of exposure (current, former) and duration of exposure (>5 years, ≤5 years). Never users were defined as the reference group for all risk associations. Tests for trend were carried out by ordering the categorised values from lowest to highest and including the resulting variable as a one degree-of-freedom linear term in a Cox proportional hazards model.

We first assessed associations of PMH therapy with any incident CRC. Subsequent analyses examined CRC risks defined by subtypes according to presenting stage (localised, regional or distant), anatomic subsite (proximal or distal), microsatellite instability phenotype (MSI-H or MSS/MSI-L), CIMP status (CIMP-positive or CIMP-negative) and BRAF mutation status (BRAF-mutated or BRAF-wildtype). For the subtype-specific analyses, the outcome variable was incident CRC with the molecular marker of interest; all other incident CRCs (including those with missing or unknown values for the molecular marker of interest) were considered censored observations at the date of diagnosis. We determined if risk ratios for the oestrogen-related exposures differed according to these cancer subtypes using a competing risk form of Cox proportional hazards analysis.25 This approach allowed us to model and test the interaction between oestrogen exposure (modelled as a covariate) and molecular/tumour subtype (modelled as a stratum variable). We also examined associations of PMH therapy with CRC risk based on subsets defined by tissue availability (available vs not available) to determine if incomplete tissue access introduced any biases. Two sets of Cox regression models were fit: one accounting for age only and one additionally adjusting for other potential CRC risk factors, including body mass index (quartiles), waist to hip ratio (quartiles), smoking status (current, former, never), age at menopause (≤44, 45–49, 50–54, ≥55 years), age at menarche (8–12, 13–14, ≥15 years), oral contraceptive use (ever, never), physical activity level (low, moderate, high), alcohol consumption (0, 0–3.4, >3.4 g/day based on cohort median split), history of diabetes (ever, never), and daily intake (quartiles) of total energy (kcal/day), total fat (g/day), sucrose (g/day), red meat (g/day), calcium (mg/day), folate (mg/day), methionine (g/day) and vitamin E (mg/day). Family history of CRC, non-steroidal anti-inflammatory drug use and exposure to CRC screening were not systematically recorded at baseline and were not included in the current analyses. All statistical tests were two-sided, and all analyses were carried out using the SAS (SAS version 9.2.) and R software systems.

Results

According to data from the baseline questionnaire, PMH therapy was categorised as ever use for 14 299 (38%) women and never use for 22 986 (62%) women. Age at baseline (mean±SD) was similar between ever users (62.3±4.1 years) and never users (62.2±4.3 years). Other baseline characteristics are shown in table 1, by PMH therapy status. Among women who reported using PMH therapy, 4062 (11%) and 10 237 (28%) were further categorised as current users and former users, respectively. With respect to duration of use, 10 017 (27%) women reported ≤5 years and 4017 (11%) women reported >5 years of PMH therapy at baseline. For the 553 incident CRC cases with complete molecular marker data, the mean (±SD) age at diagnosis was 74.0 (±5.9) years. By SEER stage, cases were distributed as 181 (33%) localised, 236 (43%) regional, and 71 (13%) distant disease (stage data were not available for the remaining 65 cases). The molecular subtype distributions were: 394 (73%) MSI-L/MSS and 145 (27%) MSI-H; 361 (69%) CIMP-negative and 164 (31%) CIMP-positive; and 383 (72%) BRAF-wildtype and 153 (28%) BRAF-mutated. Relatively strong agreement was observed between the paired biomarker values (MSI and CIMP, κ=0.70; MSI and BRAF, κ=0.66; CIMP and BRAF, κ=0.82).

Table 1

Baseline characteristics of study subjects, by postmenopausal hormone (PMH) therapy status

In age-adjusted risk models, PMH ever users were at lower risk for any incident CRC compared to PMH never users (RR=0.78, 95% CI 0.69 to 0.88). Multivariate adjustment resulted in a similar, though slightly attenuated, risk association (RR=0.82; 95% CI 0.72 to 0.93) (table 2). Therefore, further analyses are reported based on the multivariate risk model. Risk estimates for PMH ever versus never use by presenting SEER stage were RR=0.91 (95% CI 0.66 to 1.27) for localised, RR=0.78 (95% CI 0.58 to 1.04) for regional, and RR=0.72 (95% CI 0.42 to 1.23) for distant disease (p=0.13 for comparison across presenting stages). Risk estimates for PMH ever users and CRC overall were not statistically different between women 55–59 and ≥60 years of age at baseline (RR=0.85; 95% CI 0.65 to 1.10 and RR=0.80; 95% CI 0.69 to 0.93, respectively; test for interaction p=0.99), or for women who reported previous hysterectomy (RR=0.85; 95% CI 0.68 to 1.07) or no previous hysterectomy (RR 0.82; 95% CI 0.69 to 0.98, test for interaction p=0.65) at baseline. CRC risk estimates based on recency of exposure were RR=0.85 (95% CI 0.74 to 0.98) for former PMH users and RR=0.73 (95% CI 0.58 to 0.92) for current PMH users, with a statistically significant trend across categories (p=0.001). Duration of exposure was also inversely associated with CRC risk, with estimates of RR=0.86 (95% CI 0.75 to 0.99) and RR=0.72 (95 CI 0.57 to 0.90) for ever users reporting ≤5 years and >5 years of PMH therapy at baseline, respectively (p trend <0.001). Similar associations were observed when PMH duration at baseline was evaluated within the current user and former user subgroups.

Table 2

Associations between postmenopausal hormone (PMH) therapy at baseline and incident colorectal cancer (CRC), overall and by anatomic subsite

Analyses based on anatomic subsite demonstrated that the inverse association between PMH use and incident CRC was more pronounced for distal than for proximal tumours (table 2). Further separation of distal colon and rectal cancer cases did not appreciably alter the subsite-specific risk estimates (data not shown). When the CRC outcome was defined as only those cases for which adequate tissue samples were obtained for molecular testing, the multivariate risk estimate for PMH therapy ever users versus never users (RR=0.80; 95% CI 0.67 to 0.97) was not statistically different from the estimate based on all incident CRC cases, supporting a low likelihood of selection bias introduced by tissue availability status.

With respect to the molecularly defined CRC subtypes, baseline PMH therapy status was associated with a statistically significant reduction in risk for MSI-L/MSS tumours, but not MSI-H tumours (RR=0.75; 95% CI 0.60 to 0.94 and RR=0.98; 95% CI 0.69 to 1.39 for comparison of ever vs never users, respectively) (table 3). Further, PMH therapy of >5 years duration was associated with a 40% lower risk for MSI-L/MSS tumours with no appreciable influence on MSI-H tumour risk. Longer duration of PMH therapy was also associated with borderline statistically significant risk reductions for CIMP-negative (RR=0.68; 95% CI 0.45 to 1.03) or BRAF-wildtype (RR=0.70; 95% CI 0.47 to 1.05) tumours, while associations with the CIMP-positive or BRAF-mutated subtypes were less apparent. Due to sparse case numbers in some cells, we were not able to meaningfully estimate molecularly defined CRC risks stratified by proximal versus distal subsites.

Table 3

Associations between postmenopausal hormone (PMH) therapy at baseline and molecularly defined colorectal cancer (CRC) subtypes

Discussion

In this large cohort study of older women, we found that baseline-reported PMH therapy was inversely associated with incident CRC, with statistically significant trends by recency and duration of exposure, as well as by CRC anatomic subsite. Further analyses based on molecularly defined CRC subtypes demonstrated that PMH therapy was associated with a statistically significantly lower risk for MSS/MSI-L tumours, as well as borderline statistically significant risk reductions for CIMP-negative and BRAF-wildtype tumours among women with prolonged exposure to PMH therapy (ie, >5 years duration). Conversely, PMH therapy-related risk estimates for the MSI-H, CIMP-positive and BRAF-mutated CRC subtypes were not statistically significant. To our knowledge, we report the first prospective evaluation of PMH therapy and CRC risks based on MSI, CIMP and BRAF mutation status. Based on emerging data, several molecular models have been described to account for the clinicopathological heterogeneity in colorectal carcinogenesis.11 ,26–29 Using the framework proposed by Issa29 and later expanded by Leggett and Whitehall,11 our data suggest that PMH therapy may have more pronounced inhibitory effects on the ‘traditional’ pathway, as compared to the serrated or alternate pathways, of colorectal carcinogenesis. Further molecular marker analyses (ie, CIMP-low marker panel, KRAS mutation status), as well as confirmation by other studies, would strengthen this impression.

Our overall observations are similar to a previous case–control study of older women (ages 50–74 years) reported by Newcomb et al.13 Among 1004 cases and 1062 controls recruited from the Seattle, Washington area, exposure to PMH therapy was non-significantly associated with decreased risk for any CRC risk (OR=0.9; 95 CI 0.7 to 1.1 for comparison of ever use vs never use). Stronger associations were observed when recency and duration of exposure were factored into the analyses, with current PMH use for ≥5 years resulting in a 30% CRC risk reduction (OR=0.7; 95% CI 0.6 to 0.9). Further consideration of CRC subtypes defined by a panel of nine MSI markers demonstrated a statistically significant association between oestrogen plus progestin use and MSI-L/MSS tumours (OR=0.6; 95% CI 0.4 to 0.9), but not MSI-H tumours (OR=0.7; 95% CI 0.4 to 1.4). Of note, PMH therapy with oestrogen alone did not appear to favourably affect CRC risk, overall or by MSI status. Since IWHS participants were not asked to provide detailed information regarding the type of PMH therapy used, we were unable to estimate CRC risks stratified by exposure to oestrogen plus progestin versus oestrogen alone in our study. However, based on data from pharmaceutical marketing surveys conducted near the time of IWHS cohort inception, it has been estimated that only ∼20% of study participants were using combination PMH therapy.30 ,31 Further, based on results from the Women's Health Initiative observational study32 and other recently reported cohort studies,33–35 the degree to which specific hormone regimens modify the association between PMH therapy and CRC risk remains unresolved.

Additional data referent to MSI-defined CRC risks associated with PMH therapy are limited to a multicentre case–control study from Slattery et al.12 Using a previously published panel of 12 MSI markers, CRC cases among postmenopausal women were categorised as MSI-positive (n=129) or MSI-negative (n=487). Compared to controls (n=982), neither MSI-positive (OR=1.0; 95% CI 0.7 to 1.5) nor MSI-negative (OR=0.9; 95% CI 0.7 to 1.1) tumours were significantly associated with ever versus never use of PMH therapy. Nonetheless, based on risk patterns observed for other hormonal exposures and reproductive factors, the authors proposed that oestrogen may interrupt colorectal carcinogenesis by preventing DNA methylation-induced silencing of oestrogen receptor expression, with additional speculation that one or more DNA mismatch repair genes may be oestrogen responsive as well (resulting in oestrogen-mediated protection for MSI-positive, rather than MSI-negative, tumours). Subsequently, Miyamoto et al showed that oestradiol can induce MLH1 and MLH2 protein expression in an endometrial cell culture system.36 However, data from a more recent case–control study14 suggest that PMH therapy may actually increase the methylation of ESR and related genes, emphasising the need for further evaluation of the mechanism(s) through which oestrogen and/or progestin might protect against colorectal neoplasia formation.

Following publication of the Women's Health Initiative (WHI) clinical trial results in 2002, the prevalence of PMH therapy declined by approximately 80%.37 More recently, usage rates appear to be rebounding, based in part on updated information about the potential health benefits of PMH therapy in the early postmenopausal period.37 According to one recent review,38 the composite risk:benefit ratio may be more favourable for women who are 50–59 years of age or <10 years postmenopause. Among IWHS participants, the association between PMH therapy and incident CRC was nearly identical for women aged 55–59 or >60 years at baseline. Observations from our study are also in keeping with secondary analyses of CRC endpoints from the WHI clinical trial and observational study,32 ,39 although one recent report from the WHI oestrogen-alone trial40 suggested greater potential harm associated with conjugated equine oestrogen therapy among women aged 70–79 versus 50–59 or 60–69 years. Thus, based on limited existing data, the possibility of differing effects on CRC risk from PMH therapy when used in the earlier versus later postmenopausal periods remains inconclusive.

General strengths of our study include the population-based, prospective design; ability to control for multiple potential confounding factors in multivariate risk models; prolonged follow-up time with comprehensive ascertainment of incident CRC cases; and large sample size, which permitted relevant subgroup analyses. One acknowledged limitation of our study is that exposure to PMH therapy was assessed at a single time point (baseline), without enquiring about the dose, formulation, start date or stop date (for former users), which could have provided deeper insights for defining potential CRC chemoprevention strategies. Also, data regarding family history of CRC and exposure to CRC screening were not routinely available from the baseline or follow-up questionnaires. Because PMH therapy users may be more willing to undergo CRC screening than non-users, at least based on utilisation rates for faecal occult blood testing,41 it is conceivable that differential screening might have influenced our risk estimates. However, as noted by others,42 this effect is difficult to predict, since CRC incidence rates may be higher (due to increased detection) or lower (due to preventive polypectomies) among screen-adherent women. As with any observational study, our findings may have been influenced by the demographics of our subject population (ie, older and primarily Caucasian women) and should only be extrapolated to younger, more racially/ethnically diverse groups with appropriate caution. In addition, our molecularly defined risk estimates were based on a representative, but not complete, subset of CRC cases. Nonetheless, as noted above, biases related to tissue collection did not appear to influence the observed associations between PMH therapy and incident CRC.

In summary, data from this prospective, population-based cohort study provide support for an inverse association between PMH therapy and specific, molecularly defined CRC subtypes. Our results are consistent with the possibility of differential chemopreventive effects from PMH therapy on CRCs that arise through the traditional pathway of carcinogenesis.11 ,29 Analyses of additional molecular markers using data and tissue resources from the IWHS cohort are planned, to reinforce the currently reported observations. If confirmed by other studies, our findings may have direct implications for clinical practice, for example with respect to future application of PMH-based chemoprevention based on clinicopathological characteristics that predict the highest anticipated risk:benefit ratio, as well as for further translational research regarding the effects of PMH dose, duration and formulation on molecular mechanisms of colorectal carcinogenesis.

References

Footnotes

  • Funding This work was supported by grant funding from the National Institutes of Health (USA), including R01 CA107333 (Limburg, PI). Dr Weisenberger's participation was also partially supported by seed funding from the American Cancer Society.

  • Competing interests Dr Weisenberger holds patents and receives royalties from the University of Southern California (outside of the submitted work). Dr Laird serves on the Scientific Advisory Board and as a Consultant to Epigenomics, AG., and also receives financial support from patents and royalties related to methylation technology from this group (outside of the submitted work). Dr Limburg also receives support for his research from Olympus America, BENEO-Orafti Group, Bayer Healthcare, Fujinon, Boston Scientific, and Astra Zeneca (outside of the submitted work). He is listed as a co-inventor on US patent 5891651, served as a consultant for Genomic Health, Inc. from 8/12/08-4/19/10, and Mayo Clinic has licensed Dr Limburg's intellectual property to Exact Sciences, for which he and Mayo Clinic have contractual rights to receive royalties(outside of the submitted work).

  • Ethics approval This study was reviewed and approved by the institutional review boards for human research of the University of Iowa, University of Minnesota and Mayo Clinic.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Data from this study will be made available upon request to the Corresponding Author (Dr Limburg), in accordance with relevant guidelines from the funding agency.

Linked Articles

  • Digest
    Emad El-Omar William Grady Alexander Gerbes