Impact of Menopausal Hormone Therapy on Colorectal Cancer Risk

A Systematic Review

Livia Nakhostin; Aurelia Stadler; Petra Stute


Clin Endocrinol. 2021;95(3):390-397. 

In This Article


Study Characteristics

Out of 1001 records, nine meta-analyses and 57 original studies were selected for this systematic review,[1–66] an additional 28 eligible references were identified by screening relevant reviews on the topic (see Figure 1). A comprehensive reference list can be found in Table S2. We only refer to key studies in the references listed in the main manuscript.

Most of the nonrandomized studies showed a low risk of bias for the selection and comparability domain, as well as for subunit 2 from the exposure domain. A substantial amount of studies provided no description for assessment of outcome and had a loss of follow-up >10% or no statement (Table S3). For included randomized studies, a low risk of bias was detected for selection bias (random sequence generation and allocation concealment), attrition bias and reporting bias. High or unclear risk of bias was evaluated for other sources of bias, performance bias and detection bias (Table S4). Most meta-analyses presented a moderate overall methodological quality for risk of bias. A high total score (8–9 of 9 items) was found for duplicate study selection and data extraction, comprehensive literature search and appropriate methods used to combine. A low total score (3–4 of 9 items) was found for publication status as an inclusion criterion, list of studies (included and excluded) provided and quality assessment (Table S5).

Apart from nine meta-analyses, there were 14 placebo-controlled (PC) randomized clinical trials (RCT), 24 cohort studies (prospective and retrospective) and 20 case-control studies included, respectively. Results of PC-RCTs only were separately highlighted in the following result sections as PC-RCTs play an important role for analysing a causal relationship between MHT and CRC risk (Table S6). Sample size ranged from 347[56] to 1,030,711.[65] However, some meta-analyses did not provide sample size information. Men participating in the studies were not considered.[19,28] Mean age ranged from 45[20] to 73.8 years,[62] and mean BMI from 23.3[25] to 33.9.[58] However, not all studies reported age and BMI of participants. Duration of intervention ranged from 1 month[12,20] to ≥20 years,[15,16] and follow-up from 6 months[1] to 18 years,[45] respectively.

Study participants were mostly postmenopausal (42 of all studies). 13 studies did not clearly differentiate between the reproductive stages pre-, peri- and postmenopausal. Peri- and postmenopausal participants were part of eight studies. Only three studies explicitly included pre- and postmenopausal women.[28,34,38]

MHT was either oestrogen only, oestrogen combined with progestogen, progestogen only, tibolone[39,60] or unclear.[10,18] However, some studies did not report on MHT components, dosage, application and regimen at all. The oestrogen components of MHT were (mostly) conjugated equine oestrogens, different kinds of oestrogens[29,39,60] or undefined. Oestrogen dosage was either (mostly) standard-dose (conjugated equine oestrogens at 0.625 mg/d) or unclear. The progestogen components were either a nor-testosterone derivate (norethisterone acetate[22] or other), a progesterone derivate (progesterone,[17] medroxyprogesterone acetate or other) or not defined. Combined MHT was either applied continuously, sequentially[13,17,22,39,60] or not defined. Routes of administration were either (mostly) oral, transdermal, by injection, intranasal[35] or not defined. Each study included one to multiple endpoints, the three most frequent being: risk/incidence for colorectal/colon and rectal cancer (52 studies), tumour/cancer location (28 studies) and colorectal/colon and rectal cancer mortality (16 studies) (Table S2).

MHT and Risk for Colorectal Adenoma

The impact of MHT on colorectal adenoma was assessed in six studies (Table S7). Three studies did not provide information on MHT.[31,56,65] The participants of the other two studies either received oestrogen only[6,25] or combined MHT.[6,25]

Overall, the impact of MHT on colorectal adenoma risk was either neutral[6,25,65] or protective.[31,51,56] Similarly, the impact of MHT on colorectal adenoma location[6,25,31,51] and size[6,51] was either neutral[31,51] or statistically not significant.[6,25,51] One study found no difference for recency of MHT use (past or current),[25] while another found that current MHT was associated with a lower prevalence of colorectal adenomas compared to former use.[51] Neither MHT duration,[6,25,51,56] time since last use[31] nor form of application[25] had an impact on colorectal adenoma risk. In respect to MHT type, only ever and past use of oestrogen therapy but neither current oestrogen therapy nor ever use of combined MHT was associated with a significantly higher risk for colorectal adenomas[25] with oestrogen dosage being irrelevant.[6,65]

MHT and Risk for Colorectal, Colon and Rectal Cancer

With the exception of nine studies,[1,4,5,23,27,31,45,51,56] all studies provided information on MHT and incidence of colorectal, colon or rectal cancer. MHT was either oestrogen only, combined MHT, progestogen only, tibolone[39,60] or undefined.[10,18,58] However, some studies did not report MHT details.

Overall, the impact of MHT on CRC risk showed a heterogeneity in findings. The majority of studies displayed an either neutral or protective effect. Nevertheless, six studies showed a non-significant increase in risk for CRC.[18,37,42,43,47,50] CRC risk was divided into MHT ever use, current use and past use.

When analysed by nine PC-RCTs only, the findings remained heterogeneous for CRC current[41–43,47,49,50,53] and ever use,[44,49,52] while past use showed a non-significant risk reduction,[53] respectively. MHT duration did not have an impact on CRC risk in most studies.[25,37] Similarly, age at baseline of MHT use did not have an impact on risk for CRC.[10,17] However, when differentiating between MHT types, combined MHT use was neutral for age,[32,50] whereas oestrogen only use at higher age was associated with a significantly increased CRC risk.[32,43,50] MHT regimen,[13,17,39] route of application,[15,17] oestrogen dosage,[3,6,21] progestogen dosage[3] and type of progestogen[17] did not have a major impact on CRC risk.

Some studies further differentiated between different sites of CRC (colon, rectum and distal large bowel). Similar to CRC, most studies reported heterogeneous results concerning colon cancer risk by ever,[40,46] current[47,52] and past MHT use,[58,61] respectively. Nevertheless, seven studies reported a non-significant increase in risk for colon cancer.[24,28,29,34,36,37,47] Results remained heterogeneous for colon cancer by current use[47] when analysed by four PC-RCTs only, while risk by ever use was non-significantly decreased.[48]

In terms of rectal cancer risk, most studies also reported a reduced risk by ever,[3,24] current,[47,52] recent[22,37] and past[38] MHT use, respectively. An increased risk of rectal cancer was not reported by the analysed studies. PC-RCTs only analysed risk for current use, which was non-significantly reduced by the three studies.[42,47,52] MHT duration either showed no influence[15,22,33,37] or an inverse trend in rectal cancer risk with longer duration of use (significant[8,38]). The risk of distal large bowel cancer (defined as rectal, recto-sigmoid or sigmoid cancer) was significantly reduced by MHT.[26] Longer duration significantly further decreased the risk (lowest risk ≥ 15 years).[26] For further information regarding MHT and risk for colorectal, colon and rectal cancer, please refer to Table S8.

Impact of MHT on CRC Stage and Invasiveness

The impact of MHT on CRC stage was reported by 13 studies (Table S9). MHT was either oestrogen only, combined MHT or undefined.[10]

In respect to cancer stage at diagnosis, no concluding impact of MHT was found. When analysed by PC-RCTs only, the findings remained heterogeneous.[47,50,52] Some studies found no difference by tumour stage.[47,59] Two trials stated a more advanced stage under MHT (significant overall,[52] respectively, for subgroup combined MHT[50]), whereas in other studies stage III was more likely than IV,[4] and less stage II-IV than early stage cancer appeared under MHT versus controls (non-significant[33]). For colon and rectal cancer individually analysed, MHT showed a significant reduction of stage IV cancer, although for rectal cancer results were non-significant after more than 10 years.[22] Regarding invasive CRC and MHT, one study showed a significantly decreased risk,[52] whereas another study showed a non-significant increase.[42] The risk for invasive colon cancer was significantly decreased.[9] Regarding lymph node involvement, an increased risk with MHT (significant[52]) and a greater number of positive lymph nodes (significant[52,53]) were discovered. MHT mostly decreased (significant overall,[52] respectively, for subgroup oestrogen only;[36] non-significant overall,[10] respectively, for subgroup combined MHT[36]) or increased (non-significant[3]) the risk of local disease (defined as limited to starting place). For regional disease (defined as spreading to nearby lymph nodes, tissues or organs), the risk was also decreased in all studies (significant overall,[3] respectively, for subgroup oestrogen only;[36] non-significant overall,[10,52] respectively, for subgroup combined MHT[36]). Risk for distant disease (defined as spreading to distant parts of the body) was also mostly decreased (significant;[36] non-significant[3,10,52]), rarely increased (significant[53]). In colon cancer separately analysed, the risk for regional stage was either decreased (significant[9]) or increased (non-significant[59]) and for distant stage decreased (significant;[9] non-significant[59]).

Impact of MHT on CRC Tumour Grade

The impact of MHT on CRC tumour grade was investigated by three PC-RCT[42,47,53] (Table S10). Women were treated with either oral conjugated equine oestrogens at 0.625 mg/d or oral conjugated equine oestrogens at 0.625 mg/d plus medroxyprogesterone acetate continuously.[53] Overall, MHT did not have an impact on tumour grade when compared with controls or placebo-treated women, respectively.[42,47,53]

Impact of MHT on Molecular and Histologic Characteristics of CRC

Currently, the signalling mechanism of how MHT modifies CRC risk is not identified. Nevertheless, potential signalling mechanisms of MHT on three distinct molecular pathways causing CRC have been studied: (1) the classic adenoma-carcinoma sequence involves the mutation of the tumour suppressor gene APC (a key negative regulator of B-catenin and component of the WNT signalling pathway). This is followed by additional mutations, including proto-oncogene mutations in KRAS. (2) The microsatellite instability (MSI) pathway involves mutations of mismatch repair genes (MLH1, MSH2, etc), resulting in sessile serrated adenomas. If these mutations affect genes involved in cell survival and proliferation (TGFβR2, BAX, BRAF, TCF-4, IGF2R, etc), cancer may develop. MSI-H implicates a high degree of instability, whereas MSI-L stands for little degree of instability.[67] (3) The global genome hypermethylation pathway results in a switch of tumour suppressor genes, indicated as CpG island methylator phenotype (CIMP).[68]

Overall, seven studies evaluated the impact of MHT on molecular and histologic CRC characteristics (Table S11). MHT was either oestrogen only, combined MHT or undefined.[10] One study did not provide information on MHT details.[11]

KRAS mutation-negative and KRAS mutation-positive tumours were not significantly inversely associated with MHT, and no statistical difference between these two types of CRC was confirmed.[11] The case-control study showed no difference in results regarding MSI and MSS CRC under MHT,[55] whereas the prospective cohort study associated MHT with a significant risk reduction for MSI-L and MSS tumours (especially >5 years duration), but not for MSI-H.[10] This study also associated longer duration of MHT with a marginal significant risk reduction for CIMP-negative and BRAF wild-type tumours.[10]

The majority of studies (all studies being PC-RCTs) found no difference in histologic CRC characteristics between MHT and placebo group.[42,47,52,53]

Impact of MHT on Colon Cancer Site

Overall, the impact of MHT on colon cancer subsite (proximal or distal colon) was investigated by 26 studies (Table S12). MHT was either oestrogen only, combined MHT, progestogen only[13,29] or undefined.[10,58] However, some studies did not provide any details on MHT.

The majority of studies found no significant difference in colon cancer subsite (proximal or distal colon) with MHT use.[3,37] Analysed by two PC-RCTs separately, no significant difference was found either.[47,53] The few studies, which showed a stronger association for distal colon cancer with MHT, observed heterogeneous results (either a more pronounced decrease[10,19] or an increase[25] of risk). The studies, which showed a stronger association for proximal than distal colon cancer, evaluated a decrease of risk.[6,8,15] Regarding the proximal colon, no association between MHT and KRAS-defined tumour subtypes was identified, whereas the distal colon was associated with a significant protection from mutation-negative but not mutation-positive tumours.[11]

Impact of MHT on CRC Mortality

Overall, 19 studies investigated the impact of MHT on CRC mortality (Table S13). MHT was either oestrogen only or combined MHT. Three studies did not provide information on MHT details.[1,27,40]

The majority of studies found no association between MHT and CRC mortality.[1,32] Separately analysed by ten PC-RCTs, no association between MHT and CRC mortality was found either.[52,53] Only two studies discovered a decreased risk of CRC mortality (significant <5 years use[4] and for EPT only[66]). Research that separated colon and rectal cancer mortality showed a decreased colon cancer mortality (significant[5,27,40]) and rectal cancer mortality, although the latter was not significant.[5] Concerning the duration, results were heterogeneous.[23,41] Increasing time since last use of MHT was associated with a modest reduction in risk (significant[23]). Regarding the recency, no effect for CRC[23] and rectal cancer[5] was seen, whereas a higher protection for current than past MHT users in colon cancer[5] was found.