|Home | About | Journals | Submit | Contact Us | Français|
Reproductive factors have been shown by epidemiology studies to alter colorectal cancer risk in women. Familial adenomatous polyposis (FAP) patients carry a germline adenomatous polyposis coli (APC) mutation predisposing to multiple adenoma formation in the intestine. The Min mouse provides a good model of FAP, and we recently reported a significant increase in intestinal tumour multiplicity in a recombinant line of mice following pregnancy.
We considered whether reproduction modulates intestinal tract disease in a large cohort of female patients with FAP (n=180).
Multiple regression analysis showed that the number of colonic polyps observed was not related to the person's pregnancy status nor the position of their APC germline mutation. The proportion of women attaining a high Spigelman stage (3 or 4) was unrelated to having a pregnancy prior to attaining the maximum Spigelman stage (p=0.6). On the other hand, having a pregnancy significantly increased the proportion of women that attained the highest Spigelman stage when their APC germline mutation occurred within the mutation cluster region or at or after codon 1020 (50%, 6/12, p=0.005 and 42%, 13/31, p=0.006, respectively; multivariable logistic regression).
Our data suggest that reproduction may influence disease severity in the upper gastrointestinal tract in patients with FAP.
Patients with familial adenomatous polyposis (FAP) develop hundreds to thousands of colorectal polyps and, consequently, have an enhanced risk of colorectal cancer (CRC). Therapy consists of prophylactic colectomy, usually performed in the late teens or early 20s.1,2 Adenomatous polyps also arise in the small intestinal tract, especially in the duodenum, and severity of disease can vary from no polyps to numerous lesions and carcinoma. Progressive disease is measured using a semi‐quantitative assessment, referred to as the Spigelman stage.3,4 patients with FAP carry a germline mutation in the APC gene, commonly within the β‐catenin binding domains (codons 1020–1170 and 1265–2035). Furthermore, the position of the germline mutation determines the nature and location of the second hit.5,6 If the mutation occurs within the mutation cluster region (MCR; codons 1194–1392), then there is strong selection for allelic loss of APC as the somatic mutation (“second hit”) in colorectal adenoma development.7
In the general population, men and women vary in CRC aetiology, and reproductive factors have been shown to alter CRC risk in women: parity has been inversely associated with colon cancer and use of hormone replacement therapy (HRT) decreases risk.8,9,10,11,12,13,14,15 An analysis of the influence of pregnancy on both duodenal and colonic disease severity in female FAP has yet to be undertaken. However, we have recently shown that pregnancy leads to an increase in adenoma numbers in the small intestine of a recombinant line of Min mice. These mice carry a truncating mutation at codon 850 of the APC gene and are regarded as a good model of human FAP.16 This observation provided impetus to considering pregnancy as a factor that might modulate intestinal disease severity in patients with FAP. Such a study is necessary because a significant proportion of female patients have completed a pregnancy before colectomy, and fecundity in patients with FAP is close to that of the general population. The extensive database of patients with FAP at St Mark's Hospital, which includes a unique collection of accurate counts of polyp numbers in the colon, provided an ideal opportunity to undertake a detailed analysis on the influence of pregnancy on both duodenal and colonic disease severity in female patients with FAP.
For this study, information was available from the polyposis registry at St Mark's Hospital for 180 female patients from 133 families, including 13 sibling pairs, 2×4 siblings, 4×3 siblings and 1×5 siblings. All had undergone a colectomy and side‐viewing gastroduodenoscopy to assess Spigelman stage.4 The number of pregnancies resulting in a live birth and the age at the time of recording the maximum Spigelman stage was established for each patient. Age at colectomy, along with accurate colonic polyp counts on colectomy specimens (performed as described previously17) was known for 72.2% (130/180) of patients; the number of pregnancies prior to colectomy was also recorded for these women. The usual practice at St Mark's is for the decision to perform colectomy to be tailored to an individual's wishes and needs; age at colectomy does not appear to influence polyp number.17 In addition, the patient's germline APC mutation was known for 71% (92/130) of patients with colonic polyp counts; the age distribution for this subset of cases did not differ significantly from the full dataset. Maximum Spigelman stage and germline data were available for 68% (123/180) patients. Again, the age distribution for this subset of cases did not differ significantly from the full dataset. Germline data were not available for cases that had not been screened or screened fully for APC germline mutations (32% 57/180). We also calculated patient survival by the time difference between the date of colectomy or date of attaining maximum Spigleman scores to the last examination date, and included this in the multiple regression analysis.
Data were analysed using Stata V.8.2 (Stata statistical software release 8.0; Stata Corp., College Station, Texas, USA). The data were analysed using multiple logistic regression for the outcome measure of proportion of cases attaining a Spigelman stage of 3 or 4 and multiple linear regression for polyp counts at colectomy. Regressions investigated whether the outcomes varied with pregnancy, number of pregnancies or germline mutation, and were adjusted for age at maximum Spigelman stage/colectomy of patient. Age was grouped as <20, 20–29, 30–39, 40–49 and 50 years. As the dataset included women from the same family, all regressions were clustered on family and this was not a significant confounding factor in our regression analysis.
In patients for whom the number of colonic polyps at colectomy was known, a wide range of values was recorded and the distribution was very skewed. The data were normalised by taking the fourth root transformation of polyp numbers. Multiple regression analysis was performed by two methods (1) with the fourth root of polyp number for an individual as the outcome and (2) with the number of polyps ranked from the smallest to the largest over the series. There was no difference in results from the two regressions, and so only results from the fourth‐root analyses are presented.
Germline mutations were analysed using two methods of multiple testing: (1) the germline mutations were grouped as MCR (codons 1250–1400) and non‐MCR (all other codons) and (2) they were ranked from 5′ to 3′ of codon 163–1925 and two mutation groups were created: codon <1020 and 1020. Bonferroni correction was applied to the germline mutation variable to allow for multiple testing: MCR and <1020 reached statistical significance at p=0.03. The codon 1020 mutation group included all beta‐catenin binding sites within the APC gene.
We first considered whether pregnancy influenced the number of polyps detected at colectomy. In total, 130 patients with FAP had accurate colonic polyp counts, and 63% (82/130) were identified as having completed at least one pregnancy before surgery. After adjusting for age at colectomy, multiple regressions showed that polyp number did not vary between parous and non‐parous women. Similarly, the number of pregnancies before colectomy did not influence polyp numbers. In addition, after adjusting for age, survival of the patients after colectomy was not significantly affected by pregnancy. Analysis of archival data on the severity of upper gastrointestinal disease in women with FAP (n=180; 61%, 109/180, completed at least one pregnancy) revealed, as expected, that the proportion of cases attaining a high Spigelman stage (3 or 4) increased significantly with age (age <20, 1/5, 20%; >50, 17/37, 46%) and was unrelated to having a pregnancy prior to attaining the maximum stage (p=0.6, logistic regression). After adjusting for patient age, the proportion of cases reaching a high stage was not influenced by a single pregnancy or multiple pregnancies (p=1.0; logistic regression). In addition, after adjusting for patent age, survival of a patient after the maximum Spigelman score is not affected by pregnancy.
A consistent genotype–phenotype correlation exists in FAP, based on the position of the germline mutation, the non‐random nature of the subsequent somatic mutations and the linked selection for a particular number of β‐catenin degradation repeats.4,5,6 Consequently, we analysed the subset of patients for whom the germline mutation and colonic polyp counts were known (71%, 92/130). Patients were grouped as having a germline mutation within the MCR or a mutation outside this region (non‐MCR). We then nested, within each mutation group, women who had completed a pregnancy and others who had not. Multiple logistic regressions adjusted for age and cluster on family showed that pregnancy before colectomy did not affect polyp counts in the non‐MCR group (p=0.8). In contrast, there was a borderline significant (p=0.04) difference within the MCR germline mutation group between the pregnancy and non‐pregnancy subgroups (table 11).). However, when patients were divided into those with germline mutation before codon 1020 (removing all β‐catenin degradation repeats in the mutant germline allele) and those with a mutation at or after codon 1020, then the number of colonic polyps was not significantly affected by pregnancy before colectomy in either group (p=0.6; table 11).
For Spigelman stage, we found, using logistic regression adjusted for age and clustered on families, that within the MCR group there was a significantly higher proportion of parous women with a high stage (50%, 6/12) than among women who had not experienced a pregnancy (0%, 0/13; p=0.005, logistic regression; table 22).). This significant difference was retained when patients were grouped according to position relative to codon 1020; significantly, more women carrying a 1020 germline mutation who had experienced a pregnancy attained a high Spigelman stage (42%, 13/31 p=0.006) than women who had not reproduced (12%, 4/34). Furthermore, logistic regression analysis, adjusted for age, indicated that the proportion of women with a high Spigelman stage (3/4) increased with respect to the number of pregnancies, but only in the MCR mutation group (OR =3.7, 95% CI 0.9 to 15.2, p=0.07).
As standard therapy in FAP usually consists of restorative proctocolectomy or colectomy with an ileorectal anastomosis, it is essential that the clinician is aware of risk factors, genetic or environmental, associated with subsequent adenomatous polyp formation in the remaining small intestine. Fecundity in female patients with FAP is close to that of the general population and in our study 39% (71/180) women had completed a pregnancy before colectomy. Thus, advice on reproductive matters needs to be given to women with FAP, including contraception and pregnancy (which is known to influence desmoid disease) and any associated risk of polyp formation in the remaining small intestine and CRC. Analogous to extensive investigations in breast cancer, it is very likely that local steroid metabolism in the colon plays a key role in modulating the effects of hormones such as oestrogen. There may be variation between men and women in CRC aetiology by subsite.18,19 For example, the incidence of cancer of the caecum and ascending colon is higher in women than men; such sex‐specific differences in colon cancer have been attributed to oestrogen,20 and/or the effects on bile acid metabolism secondary to hormonal changes.21,22,23,24
The Min mouse is a good model of FAP and we have reported recently on two recombinant inbred lines (I and V) and the location of a modifier (Mom3) close to Apc that alters adenoma numbers by modifying the frequency of wild‐type allele loss at Apc; line V had more severe disease and higher rates of loss in tumours.24 Subsequently, we have shown that only line I is susceptible to pregnancy‐associated adenoma formation and that this is under genetic control. The major environmental factor, namely diet, was the same for both lines V and I. We observed a statistically significant increase in the frequency of loss of wild‐type allele in parous line 1 mice relative to virgin controls.16 These observations raised the distinct possibility of pregnancy causing more severe disease in female patients with FAP. Our analysis of colonic polyp number and Spigelman stage at colectomy in women with FAP identified several intriguing possibilities with regard to the influence of pregnancy. We observed that women with a germline mutation in the MCR region of APC who had had a pregnancy tended to have a significantly higher Spigelman stage than mutation‐equivalent women that had not been pregnant. Furthermore, mothers with a mutation at or after codon 1020 also had a higher Spigelman stage compared with those who were not pregnant. Although not significant, there is also a suggestion that the number of pregnancies influences the proportion of women with a higher Spigelman stage; this needs to be tested in a larger independent set of samples. This would suggest a potential deleterious effect of pregnancy in patients with FAP with this type of germline mutation (MCR mutations or codon 1020). The codon 1020 region contains all the β‐catenin binding domains and the MCR region.
Clearly, the findings presented here will need to be confirmed in an independent dataset. It should also be acknowledged that this study does not account for several other factors that have the potential to influence colonic polyposis or Spigelman stage, including use of hormonal oral contraceptives or HRT, duration of breastfeeding, miscarriages and termination of pregnancy. Nonetheless, the results on patients with FAP and on Min mice do implicate hormonal effects as a potential important aspect of intestinal tumorigenesis in women. The precise mechanisms by which hormones such as oestrogens influence susceptibility to CRC remain to be elucidated.
We thank Jackie Wright, Sandra Burke and Kay Neale of the St Mark's Polyposis registry.
APC - adenomatous polyposis coli
CRC - colorectal cancer
FAP - familial adenomatous polyposis
HRT - hormone replacement therapy
MCR - mutation cluster region
Competing interests: None declared.