Techniques in Coloproctology

, Volume 14, Issue 2, pp 141–146

Risk factors predicting intra-abdominal desmoids in familial adenomatous polyposis: a single centre experience

Authors

  • A. Sinha
    • Polyposis RegistrySt. Mark’s Hospital and Imperial College
  • P. P. Tekkis
    • Royal Marsden Hospital and Imperial College
  • K. F. Neale
    • Polyposis RegistrySt. Mark’s Hospital and Imperial College
  • R. K. S. Phillips
    • Polyposis RegistrySt. Mark’s Hospital and Imperial College
    • Polyposis RegistrySt. Mark’s Hospital and Imperial College
Original Article

DOI: 10.1007/s10151-010-0573-4

Cite this article as:
Sinha, A., Tekkis, P.P., Neale, K.F. et al. Tech Coloproctol (2010) 14: 141. doi:10.1007/s10151-010-0573-4

Abstract

Background

Desmoids are myofibroblastic proliferations occurring in 15% of patients with familial adenomatous polyposis (FAP), 70% being intra-abdominal desmoids (IAD). Since the morbidity and mortality due to desmoids is almost entirely attributable to IAD, we aimed to identify specifically risk factors predicting IAD development in FAP.

Methods

We undertook a retrospective review of our institutional database. Multivariate analysis was performed, and hazard ratios (HR) calculated for variables including female gender, 3′ APC mutation, surgical intervention for FAP (colectomy with ileo-rectal anastomosis or restorative proctocolectomy), age at surgery and family history (FH) of desmoids.

Results

Of the 558 patients analysed, 49 (9%) developed IAD; 22 (4%) diagnosed intra-operatively and 27 (5%) developing over a median post-operative period of 34 (7–120) months. 75% of IAD had developed before age 40. A 3′ APC mutation (HR 5.2, 95% CI 2.1–13.3, P = 0.001), positive FH (HR 2.5, 95% CI 1.4–4.6, P = 0.003) and female gender (HR 1.9, 95% CI 1.0–3.5, P = 0.04) were found to be predictive of IAD development. No significant difference in IAD risk was detected between the type of surgical intervention (P = 0.37) or age at surgery (P = 0.29).

Conclusions

Our analysis confirms 3′ APC mutation to be the most significant risk factor for IAD development. The independent association between positive FH and IAD risk suggests the existence of modifier genes, independent of the APC genotype–phenotype correlation. Few of these risk factors can be meaningfully modified. Delaying prophylactic surgery may be appropriate in female patients with a 3′ APC mutation and attenuated polyposis.

Keywords

Desmoid tumourFamilial adenomatous polyposisRisk factors

Introduction

Familial adenomatous polyposis (FAP) is an autosomal dominant disease characterised by the development of hundreds to thousands of premalignant adenomatous polyps throughout the colorectum and duodenum [1]. Colorectal carcinoma develops if colectomy is not performed [2]. The usual choice of surgery lies between colectomy with an ileo-rectal anastomosis (IRA) or restorative proctocolectomy (RPC) [3, 4]. With improved endoscopic surveillance, the advent of genetic testing and prophylactic surgery, the extra-colonic manifestations of FAP are becoming more significant clinically [5, 6]. Duodenal polyposis and desmoid tumours (DT) are now the outstanding challenges in the management of patients with FAP.

DT are rare myofibroblastic proliferations arising from musculo-aponeurotic tissue or the small bowel mesentery [7]. They can occur sporadically but are approximately a thousand times more common amongst patients with FAP. The lifetime incidence in this group is 10–25%, although subclinical disease is present in a greater percentage [810]. DT are a leading cause of death in patients with FAP who have undergone prophylactic colectomy with a reported mortality rate of 10% [11, 12].

Seventy percent of desmoids in FAP are intra-abdominal desmoids (IAD) [11]. IAD are associated with significantly poorer survival in comparison with desmoids confined to the abdominal wall [13]. IAD whilst lacking metastatic potential can be aggressive in their infiltration and compression of surrounding structures leading to considerable morbidity or even mortality. They can cause ischaemic bowel perforation or renal insufficiency secondary to extrinsic ureteric compression. IAD can also jeopardise abdominal surgery, including completion proctectomy [14].

The precise aetiology of these intriguing tumours remains unclear, although various risk factors have been implicated. Hormonal influence on tumour progression has led to the widespread use of oestrogenic compounds like tamoxifen or toremifene in the management of both sporadic and FAP-associated DT [15]. Surgical trauma, particularly abdominal surgery like prophylactic colectomy, family history of DT and the position of the APC germline mutation have been reported in various small series to increase DT risk [9, 1519]. Caspari et al. [20] found that desmoid disease in FAP was more common in patients with a germline APC mutation 3′ to codon 1444. The number of retained 20-amino acid repeats (20-AAR) in the truncated APC protein has functional importance in providing an optimal level of β-catenin dysregulation for tumour growth [19, 20]. Sturt et al. [16] propose using codon 1399 as the cut-off because this mutation lies immediately after the second 20-AAR and ensures two retained β-catenin-binding repeats.

There has been some suggestion that young age at prophylactic surgery is associated with an increased risk of desmoids [21]. This, coupled with the finding that 3′ APC mutations not only predispose to DT formation but also result in mild colonic polyp phenotype, has led some clinicians to delay surgery. This is particularly applicable to patients with a known 3′ APC mutation or a strong family history of DT, whereby surgical intervention maybe deferred to reduce the risk of IAD post-operatively [22].

The average time lag between surgery and diagnosis of a DT is 2–3 years; DT do, however, occur spontaneously in FAP with a prevalence of 4% at the time of primary surgery [23]. There has been some suggestion that RPC, which can result in mesenteric tension during ileal pouch anal anastomosis, can lead to an increased rate of DT formation. However, this hypothesis has not been substantiated by others [9, 17, 18]. Treatment strategies for DT include non-steroidal anti-inflammatory drugs, anti-oestrogens, surgical excision, radiotherapy and cytotoxic chemotherapy, although true assessment of efficacy is compromised by the variable nature of FAP-associated DT and their rarity [24, 25]. Since the morbidity and mortality due to DT is almost entirely attributable to IAD, the aim of this study is to specifically identify risk factors predictive of IAD formation amongst patients with FAP. This is in contrast to previous studies, which have combined together desmoids arising at all sites, despite the fact that this is not as clinically meaningful.

Methods

The St. Mark’s Hospital Polyposis Registry database was searched for patients with a diagnosis of FAP, defined by either an identified APC mutation or the presence of over 100 colorectal adenomas with no evidence of MYH mutation. Only patients who underwent either an IRA or RPC at our institution between 1 January 1925 and 31 December 2008 were included. We analysed these prospectively collected data, the outcome of interest being the development of IAD. Patients developing abdominal wall or extra-abdominal desmoids only have been excluded. Data were also collected on patient demographics, type of surgery, age at surgery, family history of DT and APC mutation. APC mutation 3′ to codon 1399 was considered ‘desmoid-prone’.

Patients having one or more first-degree relative(s) with desmoids were classified as having a positive FH. To investigate the influence of the type of surgery on DT formation, patients were grouped into those undergoing an IRA or RPC. Durno et al. [21] suggest women undergoing surgery below eighteen years of age are the highest risk group for DT development. This was explored first by comparing DT risk in patients undergoing surgery at age above or below 18. We further stratified risk amongst women undergoing surgery above or below 18. Where APC mutation analysis was successful, patients were grouped into those having a mutation either 3′ or 5′ to codon 1399.

Statistical analysis

Statistical analysis was conducted using STATA™ SE 10.1 for Macintosh (STATA Corp, Texas, USA). Continuous data are presented as median (range). Fisher’s exact test was used to compare proportions and Mann–Whitney U test for continuous variables. Kaplan–Meier life table analysis was used to calculate the cumulative incidence of IAD. Univariate survival analysis based on the Cox proportional hazards regression methodology was undertaken to identify individual risk factors related to IAD risk. Risk factors with a univariate P value less than 0.25 were included in the multivariate analysis. All statistical tests performed were two-sided, with significance assumed at P less than 0.05. Hazard ratios (HR) were calculated for the six variables evaluated in our regression model including female gender, positive family history (FH) for desmoids, type of surgery (IRA or RPC), age at surgery (above or below 18 years), women undergoing surgery at age above or below 18 and APC mutation (3′ or 5′ to codon 1399). First-order interactions were tested between significant variables to assess independence or confounding effects.

Results

Five hundred and fifty-eight patients were analysed, 321 (58%) being men. Four hundred and fifty-five (82%) underwent IRA and 103 (18%) underwent RPC. In total, 49 (9%) were diagnosed with IAD, the median age at diagnosis being 27 (range 15–61) years. The median age at diagnosis of IAD was younger in women compared to men—24 (15–58) versus 32 (15–61) years, respectively (P = 0.03). Figure 1a and b display the cumulative and age-specific distribution of IAD, respectively. This figure demonstrates the peak prevalence of IAD to be during the second and third decades of life.
https://static-content.springer.com/image/art%3A10.1007%2Fs10151-010-0573-4/MediaObjects/10151_2010_573_Fig1_HTML.gif
Fig. 1

Intra-abdominal desmoid occurrence by age

In 22 (4%) patients, the IAD was identified at primary surgery. Further, in 27 (5%) patients, IAD developed over a median post-operative period of 34 (7–120) months. Fifteen (33%) IADs were treated conservatively, 11 (24%) with non-steroidal anti-inflammatory drugs and 19 (42%) treated surgically. Further, two patients were treated with irradiation, and data were missing on two patients. Mutation analysis was successful in 456 (82%) patients.

Characteristics of patients with and without IAD are compared in Table 1. Tables 2 and 3 summarise the results of our risk factor analysis using a Cox regression model. A 3′ APC mutation (HR 5.2, 95% CI 2.1–13.3, P = 0.001), a family history of DT (HR 2.5, 95% CI 1.4–4.6, P = 0.003) and female gender (HR 1.9, 95% CI 1.0–3.5, P = 0.04) were found to be predictive of IAD development in FAP. First-order interaction was found to be non-significant between APC mutation and family history (P = 0.20), thus suggesting complete independence between the two variables. Interestingly, no significant difference in IAD risk was detected between the type of surgical intervention (P = 0.37) or age at surgery (P = 0.29). Table 4 compares the characteristics of patients with IAD diagnosed intra-operatively with those that developed post-operatively. No significant differences were detected between the two groups.
Table 1

Characteristics of patients with and without intra-abdominal desmoids

 

Patients without IAD

n = 509 (91%)

Patients with IAD

n = 49 (9%)

Gender

 Males

297 (93%)

24 (7%)

 Females

212 (90%)

25 (10%)

Type of surgery

 IRA

413 (91%)

42 (9%)

 RPC

96 (93%)

7 (7%)

APC mutationa

 5′ to 1399

386 (91%)

40 (9%)

 3′ to 1399

23 (77%)

7 (23%)

Age at surgery

 Over 18

336 (90%)

37 (10%)

 18 or less

173 (94%)

12 (6%)

Family history of DT

 Negative

364 (95%)

19 (5%)

 Positive

145 (83%)

30 (17%)

aData based on 456 patients

Table 2

Univariate Cox proportional hazards regression analysis of risk factors for intra-abdominal DT

Univariate analysis

Number (%)

Hazard ratio

95% CI

P value

Female gendera

 Male

321 (57.5)

1

 Female

237 (42.5)

1.48

0.84–2.63

0.178

Family historya,b

 Negative

383 (68.6)

1

 Positive

175 (31.4)

2.90

1.62–5.20

<0.0001

Type of surgery

 IRA

455 (81.5)

1

 RPC

103 (18.5)

1.46

0.64–3.33

0.368

Age at surgery

 Over 18

373 (66.8)

1

 18 or less

185 (33.2)

0.70

0.36–1.35

0.289

Females by age at surgery

 Over 18

154 (65.0)

1

 18 or less

83 (35.0)

0.76

0.30–1.92

0.555

APC mutationa,b,c

 5′ to 1399

426 (93.4)

1

 3′ to 1399

30 (6.6)

5.61

2.30–13.7

<0.0001

aUnivariate P value less than 0.25, evaluated in multi-variate model

bStatistically significant result

cData based on 456 patients

Table 3

Multivariate Cox proportional hazards regression analysis of risk factors for intra-abdominal DT

Multivariate analysis

Hazard ratio

95% CI

P value

Female gender

 Male

1

 Female

1.91

1.04–3.52

0.038

Family history

 Negative

1

 Positive

2.51

1.36–4.62

0.003

APC mutation

 5′ to 1399

1

 3′ to 1399

5.21

2.05–13.3

0.001

Table 4

Characteristics of patients diagnosed with intra-abdominal desmoid either intra- or post-operatively

 

Intra-operative IAD

n = 22 (45%)

Post-operative IAD

n = 27 (55%)

P value*

Gender

 Males

14 (58%)

10 (42%)

0.058

 Females

8 (32%)

17 (68%)

Family history

 Negative

12 (60%)

8 (40%)

0.070

 Positive

10 (35%)

19 (65%)

APC Mutationa

 5′ to 1399

17 (44%)

22 (56%)

0.399

 3′ to 1399

4 (57%)

3 (43%)

Age at surgery

 Over 18

17 (46%)

20 (54%)

0.532

 18 or less

5 (42%)

7 (58%)

Type of surgery

 IRA

19 (45%)

23 (55%)

0.618

 RPC

3 (43%)

4 (57%)

* Fisher’s exact test

aData based on 46 patients

Discussion

This unique study concentrating on risk factors for pure IAD disease has found a prevalence of 9%, similar to other published series [9, 1618]. It confirms that FAP-related IAD is a disease of young patients, with three-quarters of the 49 IAD in our series presenting before age 40. Whereas the majority of previous studies have considered risk factors predicting overall desmoid development in patients with FAP, in this study we have specifically identified variables predicting the development of clinically most important IAD. To avoid a referral bias, we have only included those patients in our analysis who underwent primary surgery for FAP at St. Mark’s Hospital. A 3′ APC mutation is the most significant risk factor for the development of IAD. No first-order interaction was seen between APC mutation and family history, suggesting having an affected first-degree relative is independently and significantly associated with an elevated risk. This association with a positive family history lends credence to the existence of modifier genes, independent of the APC genotype–phenotype correlation, that predispose to desmoid risk.

Previous studies have documented increased DT risk amongst women, consistent with a possible hormonal influence on tumour growth [9, 17]. This finding was substantiated in our analysis with an increased IAD risk amongst women. Our analysis failed to demonstrate age at colectomy as being predictive of IAD. Durno et al. [21] have suggested that women undergoing surgery below the age of 18 have the highest risk for developing desmoids, but we were unable to validate this, there being not even a subtle trend.

All patients with FAP face the prospect of undergoing abdominal surgery, which makes it difficult in a retrospective series to quantify the DT risk associated with a history of abdominal surgery, since most patients have undergone prophylactic colectomy. It is notable, however, that 45% of the IAD diagnosed in our series were identified at the primary operation. In other words, 4% of our patients with FAP were found to have an incidental IAD at primary surgery, similar to the Cleveland Clinic series experience where Hartley et al. [26] report a 3% rate of incidental IAD at primary surgery. However, we were unable to demonstrate any differences in risk factors between patients diagnosed with IAD intra- or post-operatively (Table 4), which may be a function of small sample size in the two groups.

Similar to other series, another limitation of our study is its inability to evaluate the role of any previous abdominal surgery and IAD risk. All patients analysed here have undergone IRA or RPC. Consistent with previous studies [9, 17, 18], we found no significant difference between the type of surgery performed and the development of IAD. However, we do have longer follow-up data on patients undergoing IRA compared to those undergoing an RPC, as the operation has a longer pedigree. RPC became commonplace for FAP prophylaxis over the last two decades, and it may be that with increasing numbers and longer surveillance, the outcome for the two surgeries becomes more comparable. With the advent of minimally invasive surgery, the effect of laparoscopic surgery on the development of IAD remains to be seen.

Clark et al. [13] and Sturt et al. [16] have previously published results analysing desmoids in FAP from our institutional dataset in 1999 and 2004, respectively. This study differs in that we are now analysing greater numbers and are performing risk factor analysis specifically for IAD only, abdominal wall DT having been excluded. Clark et al. [13] evaluated 88 patients with FAP whilst Sturt et al. [16] specifically explored the association between APC genotype and desmoids. This study explores the effect of more variables besides genotype, including type of surgery performed, family history of desmoids, age at surgery, and interactions between gender and age at surgery. Interestingly, of these risk factors, surgical intervention and timing of surgery are the only clinically modifiable variables. In this regard, it seems sensible where the cancer risk is low and the desmoid risk high, to opt for surveillance rather than immediate prophylactic surgery. For example, compliant patients with a family history of desmoids, a 3′ APC mutation, and an attenuated phenotype may be better served with endoscopic surveillance than surgery. Improved endoscopic prowess leads to better local control of polyp burden, and there is always the added potential of introducing chemoprevention with celecoxib. Surgery may be prudently delayed in selected compliant patients to minimise IAD risk.

Acknowledgments

The authors gratefully acknowledge Cancer Research UK for funding the St. Mark’s Hospital Polyposis Registry database and thank Muditha Samarasinghe, Jo Rawlings, Tina Isherwood and Pam Nye for maintaining the data.

Copyright information

© Springer-Verlag 2010