Cancer risks for relatives of patients with serrated polyposis

Abstract

OBJECTIVES

Serrated polyposis (hyperplastic polyposis) is characterized by multiple polyps with serrated architecture in the colorectum. While patients with serrated polyposis are known to be at increased risk of colorectal cancer (CRC) and possibly extracolonic cancers, cancer risk for their relatives has not been widely explored. The aim of this study was to estimate the risks of CRC and extracolonic cancers for relatives of patients with serrated polyposis.

METHODS

A cohort of the 1,639 first- and second-degree relatives of 100 index patients with serrated polyposis recruited regardless of a family history of polyps or cancer from genetic clinics in Australia, New Zealand, Canada and the USA, were retrospectively analysed to estimate the country-, age- and sex-specific standardized incidence ratios (SIRs) for relatives compared with the general population.

RESULTS

A total of 102 CRCs were observed in first- and second-relatives (SIR 2.25, 95% confidence interval, CI 1.75-2.93; P<0.001), with 54 in first-degree relatives (SIR 5.16, 95% CI 3.70-7.30; P<0.001) and 48 in second-degree relatives (SIR 1.38, 95% CI 1.01-1.91; P=0.04). Six pancreatic cancers were observed in first-degree relatives (SIR 3.64, 95% CI 1.70-9.21; P=0.003). There was no statistical evidence of increased risk for cancer of the stomach, brain, breast or prostate.

CONCLUSIONS

Our finding that relatives of serrated polyposis patients are at significantly increased risk of colorectal and pancreatic cancer, adds to the accumulating evidence that serrated polyposis has an inherited component.

INTRODUCTION

Serrated polyposis, also known as hyperplastic polyposis, is a condition characterized by the presence of multiple epithelial polyps with serrated architecture in the colon and rectum (1). Patients with serrated polyposis are among the most difficult patient groups encountered in genetics clinics as they have no apparent causative germline mutation, and the phenotype is highly variable with a vast array of polyp numbers, sizes, and histological subtypes (2). Approximately 1 in 3000 individuals at age 55-64 years in the United Kingdom are thought to have serrated polyposis, and 50% of them have been identified to additionally have at least one conventional colorectal adenoma (3). These patients are at increased risk of early-onset colorectal cancer (CRC) (4-10) and possibly extracolonic cancers (11).

As reflected in the recently modified WHO clinical criteria (1), serrated polyposis likely encompasses a group of diseases, rather than a single disease, or a continuum, which is influenced by a variety of genetic and environmental modifiers. Although serrated polyposis has the hallmarks of a genetic disease (young-onset, multiplicity of polyps and cancers, and restricted ethnicity), an underlying genetic alteration has yet to be found. A family history of CRC has been reported for 33% to 59% of serrated polyposis patients (6, 12, 13), although other reports have suggested that this was a rare situation (14, 15). Boparai et al (16) estimated that the first-degree relatives of serrated polyposis patients had five times the incidence of CRC (standardized incidence ratio, SIR 5.4; 95% confidence interval, CI 3.7-7.8), which is a greater increased risk than for relatives of CRC cases, and 39 times the incidence of serrated polyposis (SIR 39; 95% CI 13-121) compared with the general population. Inherited CRC predisposition syndromes are, however, seldom confined to the colorectum. This can be seen in the autosomal dominantly inherited Lynch syndrome (17) and familial adenomatous polyposis (18) where increased risks of various extracolonic cancers are well reported.

The specific risk of extracolonic cancers for relatives of patients with serrated polyposis has not been reported, although extracolonic cancers have been noted in several previous publications (11, 19, 20). In this study, we have estimated the risks of CRC and extracolonic cancers for the first- and second-degree relatives of patients with serrated polyposis.

MATERIALS AND METHODS

Study Sample

We studied the first- and second-degree relatives of patients with multiple serrated polyps (more than 5) outside the rectum without a personal history of any cancer prior to diagnosis of serrated polyposis (index cases). Index cases were recruited between 2000 and 2010 from genetics clinics in Australia, New Zealand, Canada and the USA regardless of a family history of polyps or cancer, and represented the initial presentation in each family. All index cases diagnosed due to family screening, or carrying pathogenic mutations in mismatch repair genes or the MUTYH gene were excluded from the study. All index cases were referred to genetics clinics for hyperplastic polyposis (regardless of whether they had a personal or family history of CRC). Index cases were recruited by the Australasian Colorectal Cancer Family Registry (21), and the Genetics of Serrated Neoplasia study (http://www.qimr.edu.au/page/GSN/) to which Ohio State University Medical Center (USA), Familial Gastrointestinal Cancer Registry (Ontario, Canada), Memorial University of Newfoundland (Canada), the Genetics Clinics of Australia and the New Zealand Familial Gastrointestinal Cancer Registry have contributed (5, 12).

Attempts were made to contact (via the index case) and interview the relatives of index cases identified via the Australasian Colorectal Cancer Family Registry. Written informed consents were obtained from all participants to take part in research and the study protocols were approved by local institutional research ethics review boards, as well as the Human Research Ethics Committee of the Queensland Institute of Medical Research (Protocols P628 and P912).

Data Collection

Information on demographics, personal characteristics, personal and family history of polyps, cancer, colonoscopic surveillance and surgery were obtained from index cases (and all participating relatives in some clinics) at time of recruitment. Reported cancer diagnoses and age at diagnosis were confirmed, where possible, using pathology reports or medical records. Pathology review of polyps was undertaken by a specialist gastrointestinal pathologist (CR or NIW) and information regarding the number, size, distribution, gross morphology, and histology of polyps was derived from the colonoscopy and histopathology reports. The total numbers of each polyp type were estimated during colonoscopy or from the surgical specimen if a colectomy was performed. Permission to access tumor tissue was obtained from participants.

To reduce the possibility of unintentionally including cases of metastatic cancer; cancers of lung, liver, bone and brain were only included if no other cancer was reported at or before the age of diagnosis.

Definitions

Serrated polyps were defined as any polyp with serrated crypt architecture (22) which includes both non-dysplastic polyps (hyperplastic polyp and sessile serrated adenoma/polyp) and dysplastic polyps (traditional serrated adenoma and sessile serrated adenoma/polyp with cytological dysplasia). Where polyp count was known, polyposis was categorized into two groups: those fulfilling WHO criterion-3 (>20 polyps throughout the colon) and those fulfilling WHO criterion-1 (>5 serrated polyps beyond the rectosigmoid with two exceeding 10 mm in diameter) (1). Depending on polyp count, where known, polyposis was also defined as moderate (5-79 polyps) or dense (≥80 polyps).

Statistical Analysis

Cancer-specific SIRs were calculated by dividing the observed numbers of cancers for the first- and second-degree relatives of patients with serrated polyposis by the expected numbers. The expected numbers of cancers were calculated by multiplying the age-, sex- and country-specific incidence for the general population with the corresponding observation time (person-years from birth) of the relatives. Age- and sex-specific cancer incidences in 1988-1992 for each country (Victorian Cancer Registry, Australia; National Cancer Registry, New Zealand; Ontario Cancer Registry, Canada; and the Surveillance, Epidemiology and End Results (SEER) registry, USA) were obtained from the International Agency for Research on Cancer (IARC) publication on Cancer Incidence in Five Continents (23). The period of 1988-1992 was selected for analysis because it was the closest available dataset to the mean calendar year of cancer diagnoses in the sample. The Jackknife method was used to estimate 95% confidence intervals (CIs) by allowing for any correlation of risk between relatives from the same family (24). Observation time for each subject started at birth and ended at the age at first diagnosis of cancer or last contact or death, whichever occurred first. For CRC, we censored each subject at the age of polypectomy except when it occurred within a year of the diagnosis of CRC (n=1). All index cases were excluded from the analysis.

Missing ages were estimated by use of a defined protocol. For relatives with missing ages at cancer diagnoses (31% of all cases), we assumed the age of diagnosis to be one year prior to the last known age or, if last known age was not available, the median age at diagnosis of the specific cancer for the general population. The cancer-specific median age for each sex was obtained from the Queensland Cancer Registry for relatives from Australia and New Zealand (25), and SEER Cancer Statistics Review (1975-2000) for relatives from Canada and the USA (26). For unaffected relatives with missing ages at last contact (229 first- and 507 second-degree relatives), if an exact age was not known but an age range was provided, age was estimated as the midpoint of that range. In the absence of any age information, it was assumed that both parents of a common child were born in the same year, that a parent was aged 25 years at birth of a first child, and that there were two years between the births of children.

The SIRs of CRC for relatives were estimated stratified by the characteristics of polyps diagnosed in the index case: age at diagnosis (<50 and ≥50 years), nature of polyps (with or without adenocarcinoma), polyp distribution (proximal colon and pancolonic), and density of polyps (5-79 polyps and ≥80 polyps). All reported statistical tests were two-sided and P<0.05 was considered statistically significant. All these statistical analyses were done using Stata 11.0 (27).

Estimated cumulative risks (penetrance) of cancers to age 70 years and their 95% CIs for each sex were calculated by summing over sex-specific incidences incidence_i at age i multiplied by the estimated SIR, based on the population incidences of Australia, using the formula:

$$1-e^{-\sum _{i=0}^{age}SIR\cdot incidenc{e}_i}$$

RESULTS

A total of 120 cases were recruited regardless of a family history of colorectal polyps or cancer. Of these, 20 cases were excluded because 14 were diagnosed due to family screening, 5 carried mismatch repair gene mutations and 1 carried a biallelic MUTYH mutation. The remaining 100 cases (33 were recruited from Australia, 41 from New Zealand, 6 from Canada and 20 from the USA) had 609 first-degree relatives (316 females) and 1,030 second-degree relatives (488 females) contributing data for this analysis (Table 1).

Table 1.

Baseline characteristics of relatives of index cases with serrated polyposis

Country	Total index cases No (%)	First-degree relatives No (%)	Second-degree relatives No (%)	Combined No (%)
Australia	33 (33)	205 (33)	352 (34)	557 (34)
New Zealand	41 (41)	230 (38)	341 (33)	571 (35)
Canada	6 (6)	42 (7)	81 (8)	123 (7)
USA	20 (20)	132 (22)	256 (25)	388 (24)
Total	100	609	1,030	1,639

Of the 100 index cases (63% female), minimum polyp counts were available from 66 (66%) of whom all but 7 met WHO criterion-3 (>20 serrated polyps throughout the colon). Of the 84 index cases whose location of polyps was known, 81 (96%) had pancolonic polyposis and three (4%) had polyposis concentrated in the proximal colon. Thirty-one of the 100 index cases had CRC (31%) and 17 of these (55%) were located in the proximal colon. Three proximal CRCs were multiple synchronous cancers. Average age at diagnosis was 48 (standard deviation, SD 15) years and average minimal polyp number observed in index cases was 45 (SD 32) (Table 2). Sessile serrated adenomas were described in 32/64 females (50%) and 11/36 males (31%). Details of each case are available in Supplementary Table 1.

Table 2.

Baseline characteristics of index cases with serrated polyposis

		Number	%
Sex
	Male	37	37
	Female	63	63
Serrated polyposis
	Yes	64	64
	Suspected	36	36
WHO criterion
	1	7	11
	3	59	89
	unknown	34
Density of polyps
	Moderate (5–79 polyps)	54	82
	Dense (≥80 polyps)	12	18
	unknown	34
Distribution of polyps
	pancolonic	81	96
	concentrated in the proximal colon	3	4
	unknown	16
Adenocarcinoma
	absent	69	69
	present	31	31
Adenoma
	absent	15	22
	present	54	78
	unknown	31
		Mean	SD
Age at diagnosis		47.8	15.0
Minimal polyp count		45	32
Maximum dimension of polyp (mm)		13.3	7.7

We observed a total of 102 CRCs in the first- and second-degree relatives combined (SIR 2.25, 95% CI 1.75-2.93; P<0.001), with 54 in first-degree relatives (SIR 5.16, 95% CI 3.70-7.30; P<0.001) and 48 in second-degree relatives (SIR 1.38, 95% CI 1.01-1.91; P=0.04) (Table 3). The risk of CRC for relatives was relatively greater when the index case was diagnosed under age 50 compared with when the index case was diagnosed at age 50 and above, but this did not reach statistical significance (P=0.07). There was no statistical evidence for difference in risks of CRC for relatives when numbers of polyps diagnosed in the index case were considered (see Table 4). The median age at which CRC developed in first-degree relatives was 55 years and in second-degree relatives was 66 years. The average age at diagnosis of CRC in either the first- or second-degree relatives of patients with serrated polyposis diagnosed under and over the age of 50 years was not significantly different (P=0.97).

Table 3.

Cancer-specific standardized incidence ratios (SIRs) for first- and second-degree relatives of patients with serrated polyposis compared with the general population

	Combined					First-degree relatives					Second-degree relatives
	Median age (min– max)	O	E	SIR (95% CI)	P	Median age (min– max)	O	E	SIR (95% CI)	P	Median age (min– max)	O	E	SIR (95% CI)	P
Both sexes
Colorectal cancer	60 (25–85)	102	45.25	2.25 (1.75–2.93)	<0.001	55 (25–83)	54	10.47	5.16 (3.70–7.30)	<0.001	66 (34–85)	48	34.78	1.38 (1.01–1.91)	0.04
Pancreatic cancer	75 (45–90)	11	6.93	1.59 (0.87–3.19)	0.16	73 (45–87)	6	1.65	3.64 (1.70–9.21)	0.003	75 (47–90)	5	5.28	0.95 (0.40–2.78)	0.92
Gastric cancer	71 (51–80)	10	7.81	1.28 (0.68–2.72)	0.49	67 (51–69)	3	1.95	1.54 (0.48–7.47)	0.53	75 (55–80)	7	5.87	1.19 (0.59–2.78)	0.66
Brain cancer	57 (7–64)	8	6.21	1.29 (0.62–3.17)	0.54	58 (56–59)	2	1.77	1.13 (0.25–10.98)	0.90	57 (7–64)	6	4.45	1.35 (0.64–3.40)	0.48
Lung cancer	68 (34–83)	24	43.37	0.55 (0.36–0.87)	0.01	59 (34–83)	9	10.36	0.87 (0.37–2.47)	0.77	68 (45–82)	15	33.00	0.45 (0.28–0.77)	0.002
Female
Breast cancer	59 (25–87)	41	41.71	0.98 (0.67–1.48)	0.92	55 (25–85)	15	10.46	1.43 (0.86–2.56)	0.20	59 (34–87)	26	31.25	0.83 (0.53–1.36)	0.43
Male
Prostate cancer	72 (50–86)	21	32.50	0.65 (0.42–1.03)	0.06	70 (62–75)	5	7.53	0.66 (0.29–1.87)	0.38	72 (50–86)	16	24.96	0.64 (0.38–1.15)	0.11

O = observed number of cancers, E = Expected number of cancers

Table 4.

Standardized incidence ratios (SIRs) of colorectal cancer (CRC) for first- and second-degree relatives stratified by the characteristics of polyps diagnosed in patients with serrated polyposis

Characteristics of polyps in index case	Combined				First-degree relatives				Second-degree relatives
	Median age (min–max)	O*	E	SIR (95% CI)	Median age (min–max)	O*	E	SIR (95% CI)	Median age (min–max)	O*	E	SIR (95% CI)
Age of onset
<50 years	60 (27–80)	44	14.53	3.03 (2.12–4.40)	51 (27–80)	18	2.22	8.11 (4.71–14.35)	65 (34–80)	26	12.31	2.11 (1.40–3.26)
≥50 years	60 (25–85)	58	30.72	1.89 (1.34–2.73)	56 (25–83)	36	8.25	4.31 (2.88–6.82)	70 (36–85)	22	22.47	0.98 (0.61–1.64)
P for difference				0.07				0.08				0.02
Nature of polyps
Without adenocarcinoma	59 (25–85)	79	28.94	2.73 (2.05–3.68)	55 (25–83)	43	6.63	6.48 (4.38–9.74)	66 (36–85)	36	22.31	1.61 (1.14–2.33)
With adenocarcinoma	60 (27–80)	23	16.30	1.23 (0.88–2.33)	60 (27–80)	11	3.83	2.87 (1.55–5.79)	64 (34–75)	12	12.47	0.96 (0.50–2.04)
P for difference				0.01				0.04				0.20
Location
Proximal colon	54 (43–82)	5	1.82	2.75 (1.71–4.43)	63 (43–82)	2	0.41	4.87 (2.11–13.11)	54 (43–80)	3	1.41	2.13 (1.10–3.76)
Pancolonic	61 (25–85)	76	36.95	2.06 (1.53–2.80)	55 (25–83)	38	8.87	4.29 (2.88–6.54)	66 (34–85)	38	28.08	1.35 (0.95–1.98)
P for difference				0.32				0.80				0.21
Density of polyps
Moderate (5–79 polyps)	60 (25–85)	56	25.21	2.22 (1.58–3.19)	55 (25–83)	29	5.32	5.45 (3.41–9.08)	66 (36–85)	27	19.89	1.36 (0.92–2.05)
Dense (≥80 polyps)	58 (27–62)	8	4.47	1.79 (0.87–4.02)	47 (27–62)	4	1.30	3.07 (1.69–6.44)	58 (34–60)	4	3.17	1.26 (0.37–6.29)
P for difference				0.62				0.18				0.92

O = observed number of CRC, E = Expected number of CRC

^*The summation of these numbers may not be the same with the total of CRC observed in first- and second-degree relatives because of missing data on polyp characteristics (age of onset, location, minimum polyp counts).

A total of 11 pancreatic cancers were observed in this study cohort in which six cases were in first-degree relatives (SIR 3.64, 95% CI 1.70-9.21; P=0.003). We observed no evidence of an increased risk for cancer of the stomach, brain, prostate or female breast cancer. Cancers of other organs were observed in the cohort as follows: kidney (n=3), urinary bladder (n=5), leukemia (n=5), lymphoma (n=5), myeloma (n=2), bone (n=3), thyroid (n=2), oesophagus (n=5), liver (n=5), biliary tract (n=1), endometrium (n=3), ovary (n=4), cervix (n=4) and testis (n=1).

Table 5 shows the estimated cancer-specific cumulative risk (penetrance) for first- and second-degree relatives of patients with serrated polyposis. The estimated cumulative risks to age 70 years for the first-degree relatives were: for CRC, 15% (95% CI 11-21%) for men and 12% (95% CI 8-16%) for women and for pancreatic cancer, 2% (95% CI 1-6%) for men and 1.5% (95% CI 0.7-4%) for women. When the index case was diagnosed under age 50, the cumulative risk of CRC for first-degree relatives was 24% (95% CI 15-39%) for men and 19% (95% CI 11-30%) for women. When the index case was diagnosed at age 50 and above, the cumulative risk of CRC for first-degree relatives was 13% (95% CI 9-20%) for men and 10% (95% CI 7-15%) for women.

Table 5.

Cumulative risk % of different cancers to age 70 years for first- and second-degree relatives of patients with serrated polyposis

		Cumulative risk % (95% confidence interval)
		Combined	First-degree relatives	Second-degree relatives
Colorectal cancer
	Male	7.01 (5.49–9.02)	15.34 (11.26–20.99)	4.36 (3.21–5.98)
	Female	5.23(4.10–6.76)	11.60 (8.46–16.01)	3.24 (2.38–4.46)
Pancreatic cancer
	Male	0.99 (0.54–1.97)	2.25 (1.06–5.59)	0.59 (0.25–1.72)
	Female	0.66 (0.36–1.32)	1.50 (0.70–3.75)	0.39 (0.17–1.15)

DISCUSSION

First- and second-degree relatives of patients with serrated polyposis had a significantly increased risk of developing CRC. In addition, first-degree relatives demonstrated a significantly increased risk of pancreatic cancer. Together these findings support an inherited component for serrated polyposis. Though increased risks for first-degree relatives could be due to shared environment, the observation that the CRC risk extended to second-degree relatives, and that relatives were also at increased risk for extracolonic cancers, increases the likelihood of a genetic aetiology for serrated polyposis (28). Our estimate of CRC risk for first-degree relatives did not differ significantly from the relative risk estimated by Boparai et al (16) (P=0.86).

Extracolonic cancers in relatives of a BRAF-mutated CRC proband (implying origin in a serrated polyp) were first described in a Swedish study of familial CRC in 2006 (20), and further extracolonic associations were seen in a Canadian study (29). A single anecdotal report (19), and a clinical description (11) reported occurrence of extracolonic cancers in serrated polyposis patients and their relatives. However, until our study, the magnitude of extracolonic cancer risk for relatives had not been estimated. A range of extracolonic cancers were observed in the study cohort but the cancer-specific numbers were too low to determine with any degree of certainty whether they occurred more often than expected. Common cancers such as those of the breast lung and prostate were not increased above what would be expected in the population. An apparent decreased risk for lung cancer was observed in second-degree relatives. We have no explanation for this association, however it may be a consequence of our analysis excluding any second cancers within one year of the first cancer diagnosis, which was done to minimise inclusion of metastatic disease.

We observed a significantly increased risk of pancreatic cancer for first-degree relatives of patients with serrated polyposis which is a novel finding. No cases of pancreatic cancer were seen in the serrated polyposis cases themselves, although given the young average age at diagnosis of polyposis (48 years), this is not unexpected. Pancreatic cancer occurs in familial cancer syndromes such as Peutz-Jeghers syndrome (30), and Familial Atypical Multiple Mole Melanoma syndrome (31) where its risk ranges from 9 to132-fold that of the population. It also occurs in familial pancreatitis, with a relative risk of 50-80 fold, and in families with mutations in BRCA2 and its binding partner PALB2, where the risk varies from 3.5-10 fold that of the population (32). Our estimate for pancreatic cancer risk in serrated polyposis at 3.64 is commensurate with the lower estimates for BRCA2 families. In population terms, the pancreas is less likely to develop a malignancy than any other organ in the gastrointestinal tract, with the exception of the small intestine (32). Pancreatic cancer is highly age-dependent, and occurs in the population around 70 yrs of age (33, 34). The median age at diagnosis for pancreatic cancer in our study was 73 years, not indicative of an early-age of onset. Rather, the increased frequency is suggestive of an enhanced response to risk factors such as smoking, obesity, and diabetes (34), factors which are also associated with the development of serrated polyps (35). Pancreatic cancer is difficult to prevent and there are currently no robust screening tests for early detection. Endoscopic ultrasound has been trialed in very high-risk families and has shown some promise in the detection of early precursor lesions in asymptomatic individuals (36).

Biological explanations for the association between serrated polyposis and extracolonic cancers are not known, however, our findings support the hypothesis that this condition represents an inherited cancer predisposition in which the phenotype is most strongly expressed in the colorectum. It is also possible that environmental triggers may be interacting with the genetic predisposition to produce both colonic and extracolonic cancers. Finally, more than one condition may be segregating in these families. Jarrar et al (19) reported that seven of 651 families who met the criteria of Amsterdam I, Amsterdam II, Amsterdam-like, or familial colorectal cancer, had a high prevalence of serrated polyps. One of the scenarios raised in that report was a co-occurrence of two cancer pathways; Lynch syndrome serrated polyposis. However our findings suggest that the extracolonic cancers and at least, observed in relatives of serrated polyposis patients, are unlikely to be due to co-occurring Lynch syndrome mutations as index cases with a mismatch repair gene mutation were excluded from the study.

Serrated polyposis as currently defined is likely to comprise a heterogeneous group of conditions. The pattern of inheritance in at least some serrated polyposis patients may be consistent with a recessive mode (37), analagous to MUTYH-associated polyposis (MAP) caused by germline mutations in the MUTYH gene. Increased risks of cancer for relatives of patients with MAP have been found for CRC, duodenal, ovary, bladder and skin cancer for biallelic mutation carriers (38). Moreover, an increased risk of CRC, gastric and endometrial cancer has been reported for monoallelic mutation carriers (39). Until a genetic cause is identified for serrated polyposis, studies will continue to be based on a clinical definition.

The strengths of this study are 1) that it was based, to our knowledge, on the largest sample to date of relatives of index patients with serrated polyposis; 2) it is the first study to quantify the risks of extracolonic cancers for relatives of patients with serrated polyposis; 3) index cases were not ascertained because of a previous family history of polyps or cancer and therefore the estimates from this study are less likely to be inflated due to ascertainment bias (40); and 4) we accounted for familial correlation in the risk of cancer to derive appropriate measure of estimate imprecision.

Our study has several limitations including the presence of unverified cancers, unaccounted for time and geographic variation. These might increase the imprecision of estimates more than indicated by the reported confidence intervals. Estimates from this study should be generalizable to relatives of symptomatic patients with serrated polyposis identified in a clinical setting. As all index cases were recruited from genetics clinics, this raises the possibility that relatives of symptomatic patients may represent a more aggressive phenotype associated with a higher ‘familial risk profile’ compared with other asymptomatic patients. However, symptoms such as abdominal pain or change in bowel habit which first brought the index patients to their primary care clinician are also very common outside the setting of serrated polyposis, and may have had nothing to do with their serrated polyposis condition (41, 42). Therefore it is not possible to state at this time that an aggressive phenotype should be inferred. A further limitation of the study is that it is not possible to know whether an endoscopist referred the patient to a genetics clinic because of polyp burden or because of family history. As the family history data used in this study were determined using genetics clinic pedigree records, we do not know the extent of family history recorded by the endoscopists. However, it has been observed that even in high-risk patients diagnosed under age 50, family history was not noted in 49-83% of cases (43, 44).

Our findings suggest that relatives of patients with serrated polyposis could benefit from appropriate colonoscopic surveillance. Current practice for first-degree relatives is usually five-yearly colonoscopy from the age of 40-50 years or from an age 5 years younger than the age at which serrated polyposis was diagnosed in the family. The median age at which CRC was diagnosed in first-degree relatives in our study (55 years old) supports these guidelines for surveillance. Further independent studies are required to confirm the risks of extracolonic cancers for relatives of patients with serrated polyposis. Also, given the arbitrary nature of the current criteria for the identification of serrated polyposis (1) and the initial observation of extracolonic cancers in relatives of index cases whose CRC arose from a serrated polyp (20), the implications of this finding may be applicable beyond the stringent criteria for diagnosis of serrated polyposis, identifying some families with both CRC and pancreatic cancer (in which Lynch syndrome is excluded) as having a serrated neoplasia predisposition.

This large, international study has observed that relatives of serrated polyposis patients are at significantly increased risks of colorectal and pancreatic cancer. These findings add to the accumulating evidence that serrated polyposis has an inherited component, whilst also highlighting the need for surveillance in relatives of patients with serrated polyposis.

WHAT IS CURRENT KNOWLEDGE?

• Patients with serrated polyposis are at increased risk of early-onset colorectal cancer (CRC) and possibly extracolonic cancers.

• Risk of colorectal and extracolonic cancers for relatives of patients with serrated polyposis has not been explored extensively.

WHAT IS NEW HERE?

• Both first- and second-degree relatives of index serrated polyposis patients are at increased risk of CRC.

• First-degree relatives of patients with serrated polyposis are at significantly increased risk of pancreatic cancer.

• This finding adds to the accumulating evidence that serrated polyposis has an inherited component.

• Relatives of patients with serrated polyposis could benefit from appropriate colonoscopic surveillance.