Article Text

Download PDFPDF

Original article
Modifiable lifestyle factors associated with risk of sessile serrated polyps, conventional adenomas and hyperplastic polyps
  1. James R Davenport1,2,
  2. Timothy Su1,
  3. Zhiguo Zhao1,3,
  4. Helen G Coleman4,
  5. Walter E Smalley2,5,
  6. Reid M Ness2,5,
  7. Wei Zheng1,5,
  8. Martha J Shrubsole1,5
  1. 1Division of Epidemiology, Department of Medicine, Vanderbilt Epidemiology Center, Vanderbilt-Ingram Cancer Center, Vanderbilt University School of Medicine, Nashville, Tennessee, USA
  2. 2Division of Gastroenterology, Hepatology, and Nutrition, Department of Medicine, Vanderbilt University School of Medicine, Nashville, Tennessee, USA
  3. 3Department of Biostatistics, Vanderbilt University School of Medicine, Nashville, Tennessee, USA
  4. 4Cancer Epidemiology and Health Services Research Group, Centre for Public Health, Queen's University Belfast, Belfast, UK
  5. 5Gastroenterology Section or Geriatric Research, Department of Veterans Affairs, Education and Clinical Center (GRECC), Tennessee Valley Healthcare System, Nashville, Tennessee, USA
  1. Correspondence to Dr Martha J Shrubsole, Division of Epidemiology, Department of Medicine, Vanderbilt University Medical Center, 2525 West End Avenue Room 837-A, Nashville, TN 37203, USA; martha.shrubsole{at}


Objective To identify modifiable factors associated with sessile serrated polyps (SSPs) and compare the association of these factors with conventional adenomas (ADs) and hyperplastic polyps (HPs).

Design We used data from the Tennessee Colorectal Polyp Study, a colonoscopy-based case–control study. Included were 214 SSP cases, 1779 AD cases, 560 HP cases and 3851 polyp-free controls.

Results Cigarette smoking was associated with increased risk for all polyps and was stronger for SSPs than for ADs (OR 1.74, 95% CI 1.16 to 2.62, for current vs never, ptrend=0.008). Current regular use of non-steroidal anti-inflammatory drugs was associated with a 40% reduction in SSP risk in comparison with never users (OR 0.68, 95% CI 0.48 to 0.96, ptrend=0.03), similar to the association with AD. Red meat intake was strongly associated with SSP risk (OR 2.59, 95% CI 1.41 to 4.74 for highest vs lowest intake, ptrend<0.001) and the association with SSP was stronger than with AD (ptrend=0.003). Obesity, folate intake, fibre intake and fat intake were not associated with SSP risk after adjustment for other factors. Exercise, alcohol use and calcium intake were not associated with risk for SSPs.

Conclusions SSPs share some modifiable risk factors for ADs, some of which are more strongly associated with SSPs than ADs. Thus, preventive efforts to reduce risk for ADs may also be applicable to SSPs. Additionally, SSPs have some distinctive risk factors. Future studies should evaluate the preventive strategies for these factors. The findings from this study also contribute to an understanding of the aetiology and biology of SSPs.

View Full Text

Statistics from

Significance of this study

What is already known about this subject?

  • Most colorectal cancers are derived from two separate precursor pathways: a conventional adenoma (AD)–carcinoma pathway and a serrated pathway.

  • Lifestyle factors such as cigarette smoking and increased red meat intake are known risk factors for conventional colorectal ADs.

  • Risk factors for sessile serrated polyps (SSPs) are less known, given their recent consensus regarding their defined pathology.

What are the new findings?

  • SSPs share some risk factors with conventional ADs, and other risk factors are unique to SSPs.

  • Regular use of non-steroidal anti-inflammatory drugs is associated with a reduction in risk of SSPs in addition to conventional ADs.

  • Red meat intake is strongly associated with increased risk of SSPs in addition to conventional ADs.

How might it impact clinical practice in the foreseeable future?

  • Given that SSPs are difficult to detect and may accelerate to a dysplastic state quicker than conventional ADs, primary prevention of SSPs through lifestyle modification may be an important strategy.

  • Preventive efforts to reduce risk factors in conventional ADs may also be applicable to SSPs.


Two distinct pathways to colorectal carcinogenesis have been identified. Well known is the conventional adenoma (AD)–carcinoma pathway, which involves the progression of non-advanced tubular ADs to larger or villous lesions with potential to develop into an invasive carcinoma.1 In contrast, the more recently recognised serrated pathway is thought to originate from hyperplastic polyps (HPs) and transition to distinct traditional serrated ADs or sessile serrated polyps (SSPs) prior to progression to dysplasia and carcinoma.2 SSPs, although comprising only 4–9% of all polyps discovered on endoscopy, may represent the origin for 20–35% of all colorectal cancers (CRCs), particularly those with microsatellite-instable (MSI-high) or CpG-island methylator phenotype (CIMP-high) features.3–5 Unlike ADs, which are diffusely distributed, SSPs are generally located in the proximal colon.6 For cancer screening, their importance is highlighted by new data, concluding that the decline of cancer incidence over 30 years has corresponded primarily to distal CRC lesions, while the comparative rate of decline of proximal CRC is four to seven times less.7 Furthermore, a sizeable proportion of interval CRCs, or cancers discovered between appropriate CRC screening intervals, are proximal and likely to have originated from SSPs which have either been missed, incompletely resected or have rapidly progressed to a carcinogenic state.4 ,8 ,9

Few studies have evaluated risk factors for SSPs due to challenges involved in their evolving histological definition and the relative rarity of these polyps. For multiple reasons, studies until now have often clustered HPs and SSPs into a collective ‘serrated polyp’ group, despite differences in malignant potential between the lesions. Similar to studies which evaluated risk for ADs, a few studies found risk for serrated polyps was associated with cigarette smoking,10–15 obesity,10–12 type II diabetes mellitus,11 a family history of CRC,12 age,11 ,13 higher education13 and non-steroidal anti-inflammatory drug (NSAID) use.10 ,14 Even fewer studies have evaluated the association of dietary factors with the risk of serrated polyps. These studies observed that red meat intake may be associated with increased risk in distal, but not proximal, serrated polyps.10 However, most of these serrated polyp studies are limited by the sample size and/or the likely grouping of HPs and SSPs. Given the possibility that endoscopy may not reduce mortality of proximal CRCs and that SSPs may be the primary precursor lesion for these tumours, there is a compelling need to assess modifiable lifestyle factors which may be associated with SSPs and to compare the associations with risk for ADs and HPs.

We sought to conduct a comprehensive analysis of modifiable lifestyle risk factors which may affect SSP risk, and subsequently compare the associations between ADs, HPs and SSPs. We used the Tennessee Colorectal Polyp Study (TCPS), a large case–control, colonoscopy-based study, which has standardised assessment of SSPs, ADs and HPs. Our goal was to understand the aetiology and develop a risk factor index to evaluate the joint contribution of risk factors to risk of SSPs and other polyps and to further compare risk factors between SSPs and ADs and HPs. This comparison may provide insight into the common and varied aetiology of colorectal polyps.


Study design and population

TCPS is a colonoscopy-based case–control study conducted from February 2003 to October 2010 in Nashville, Tennessee, USA. Further details regarding the methods used are previously described.16 For individuals 40–75 years of age, candidacy was discussed and consent obtained if the subject met eligibility standards. Ineligibility for the study was defined as any candidate having a history of IBD or if IBD was discovered on colonoscopy; any known family history of hereditary CRC syndromes; any history of cancer except for non-melanoma skin cancer; any previous colectomy and any diagnosis of ADs on previous colonoscopy or surgical resection. Colonoscopies were conducted as part of routine care by trained gastroenterologists. Institutional approval for human subjects research was granted through the Vanderbilt University Medical Center (VUMC) and Veterans Affairs (VA) institutional review boards and the VA Research and Development Committee.

There were 12 585 candidates initially identified for participation in TCPS, with 7621 participants (60.6%) providing an informed written consent and participating in at least one component of the study. A majority of participants (90.5%) were recruited prior to colonoscopy, and the remaining were recruited post-colonoscopy. Among the participants, 7396 were diagnosed with ADs, HPs, SSPs or no polyps and were thus eligible for this analysis. The current analysis is based on a total of 6404 eligible participants (86.6%) who completed a telephone interview (the median time to interview was 13 days). For dietary analyses, analyses were further limited to participants who also completed a 108-item food frequency questionnaire (FFQ; the median time to FFQ return was 23 days) and reported daily consumption of at least 600 kcal/day (5398 individuals; 84.3%).17

Data collection

Following the colonoscopy, interviewers used a standard telephone interview to obtain information relating to the participant's demographics, medication use, family history and other lifestyle factors. Detailed questions regarding status, intensity, duration, age of cessation and age of initiation of tobacco use were asked, with current smokers defined as one cigarette consumed daily for each of the past 6 months and over 100 cigarettes within their lifetime. Former smokers must have quit more than 1 year prior to their procedure. Any reported smoking in the last 12 months placed them in the current smoking group. Current alcohol use was defined as five or more alcoholic beverages per week over the past year. Former users did not meet this criteria for 12 months or greater prior to their procedure. Body mass index (BMI) was calculated from self-reported height and weight. Regular exercise was defined as non-occupational exercise for at least 2 hours/week over a 6-month period within the past decade, with further breakdown using metabolic equivalent of task (MET) hours per week. For defining current or former NSAID use, current users took NSAIDs at least three times weekly for the past 12 months, while former users took NSAIDs three times weekly for 12 months over the past 15 years, but without use within the last 12 months. Dietary information was self-administered using the FFQ, except in the case of red meat intake, which was obtained during the telephone interview in methods previously described.16 Dietary components in the FFQ which were examined include daily intake of total energy (kcal), fibre (g/day), dietary folate equivalents (DFE, µg/day), calcium (mg/day) and fat (g/day) as previously described.17

Classification of case groups

All participants were recruited between 2002 and 2010, during which SSPs were not uniformly recognised in clinical practice, nor was a standard pathology definition developed. As a result, the potential for misclassification of SSPs as another type of polyp (eg, hyperplastic) was substantial. To overcome this limitation of the original clinical diagnosis, we newly reviewed all polyps from all study participants, regardless of the initial clinical diagnosis to standardise all diagnoses. The study pathologist and a senior GI clinical and research pathologist established a consensus on application of the diagnostic criteria from expert panel standards (at least one distorted, dilated or horizontally branched crypt within the polyp) by joint review of cases.3 In addition, the study pathologist identified about 10% of cases in which there was a potential for disagreement and both pathologists reviewed those cases to reach consensus. Based on the pathology diagnosis, we excluded individuals who were found to have evidence of CRC (n=26) or traditional serrated ADs (n=12), due to limited statistical power. Control participants underwent a full colonoscopy, with evidence of reaching the caecum and complete colon visualisation without a notation of polyps. Visualisation of the ileocaecal valve and/or appendiceal orifice was achieved for 98.8% of polyp cases. The HP cases had one or more HPs without any synchronous AD or SSP. The AD cases had one or more tubular, tubulovillous or villous AD with or without dysplasia and with or without synchronous HPs. The SSP cases had one or more SSPs, with or without synchronous HPs and ADs.

Statistical analysis

Descriptive comparisons between case and control groups were calculated using general linear models (for continuous variables) or Mantel-Haenszel χ2 testing (for categorical variables), with adjustments in most comparisons for age (based on categories grouping individuals into 5-year age categories from 40 to 75) and sex. Dietary intake quartiles were derived from intake levels among controls. Initial assessment of risk for case–control and case–case comparisons was completed using multinomial logistic regression modelling which included each case and control group in each model to allow direct comparison of each case group. Models were adjusted for sex, age of the participant (based on the categories listed above), year of the colonoscopy, educational attainment, study site, cigarette smoking and NSAID use. Additional models, which included dietary factors, were adjusted for total energy intake. In order to test for trends, we treated categorical variables as continuous factors in the model. To assess whether factors had an independent association with polyp risk, we conducted further analysis in which factors which were statistically significantly associated in initial models were included in a subsequent multinomial logistic regression model in which they were mutually adjusted for each other. All statistical analyses were completed using R V.3. p Values of ≤0.05 (two-sided probability) were considered statistically significant in all analyses.


Demographic characteristics of each of the four groups examined (no-polyp controls, ADs, HPs and SSPs) are shown in table 1. No significant differences were found between control and case groups in comparing the procedure site, race, indication for the colonoscopy or family history of CRC. Age (pheterogeneity<0.001), sex (pheterogeneity<0.001), educational attainment (pheterogeneity<0.001), household income (pheterogeneity=0.002) and energy intake (pheterogeneity=0.009) were significantly different between groups.

Table 1

Characteristics of the study participants; the Tennessee Colorectal Polyp Study

Evaluation of modifiable non-dietary factors and polyp risk

Cigarette smoking status, duration and intensity were associated with increased polyp risk for all case types (table 2). In case–case comparisons, smoking was more strongly associated with SSPs than ADs for all measures of smoking (eg, OR 1.74, 95% CI 1.16 to 2.62 for current versus never smokers, ptrend=0.008). Obesity (BMI ≥30 kg/m2) was associated with a 30–50% increased risk of polyps and risk did not significantly differ between polyp types. In comparison with those who never regularly used NSAIDs, current regular use of NSAIDs was associated with a decreased risk of SSPs (OR 0.62, 95% CI 0.62 to 0.85 for SSP cases vs controls, ptrend=0.003) and ADs (ptrend<0.001) but not HPs, and risk reduction was dose dependent for years of use for both ADs (ptrend=0.02) and SSPs (ptrend<0.001). In addition, NSAID use of more than 10 years was more strongly associated with reduced risk of SSPs than ADs (OR 0.53, 95% CI 0.31 to 0.92 for >10 years vs never regular use). The use of NSAIDs more than seven times a week was also associated with reduced risks of SSP and AD. Alcohol use and exercise were not associated with risk of any polyp type.

Table 2

Associations between modifiable non-dietary factors and polyp risk; the Tennessee Colorectal Polyp Study

Evaluation of modifiable dietary factors and polyp risk

Higher daily dietary intake of fibre was associated with a reduced risk of SSPs (OR 0.36, 95% CI 0.19 to 0.68 for highest vs lowest intake quartile, ptrend=0.006) but was not statistically significantly different between SSPs and ADs (table 3). Folate intake (DFE) was associated with an approximate 50% reduction in risk for all polyp types. Calcium intake was only associated with statistically significantly reduced risks of ADs and HPs and was not associated with a statistically significantly reduced risk of SSPs, although the associations were in the same direction and of similar magnitude. However, the risk was not statistically significantly different between any of the case groups. Fat intake was associated with a strong dose-dependent threefold increased risk of SSPs in comparison with controls (OR 3.09, 95% CI 1.24 to 7.72 for highest vs lowest intake quartile, ptrend=0.01) and AD cases (OR 3.20, 95% CI 1.26 to 8.12 for highest vs lowest intake quartile, ptrend=0.02). Higher red meat intake was associated with all types of polyp risk, but displayed a particularly strong association with SSP risk (OR 3.38, 95% CI 1.90 to 6.00 for highest vs lowest intake quartile, ptrend<0.001). In case–case comparisons, SSP risk was approximately twofold greater than risks of either ADs or HPs for individuals consuming higher red meat intakes.

Table 3

Associations between modifiable dietary factors and polyp risk; the Tennessee Colorectal Polyp Study

Evaluation of independent associations

To evaluate which factors in tables 2 and 3 were independently associated with polyp risk after mutual adjustment, we conducted an analysis in which factors which were statistically significantly associated with risk of any polyp type were included in a single multinomial logistic regression model (table 4). After adjustment for other factors, SSP risk was no longer statistically significant for obesity and fibre, folate and fat intakes, although fibre intake was associated with a borderline statistically significant reduced SSP. Conversely, several associations persisted after adjustment. Smoking remained strongly associated with risk of all polyps. NSAID use and red meat intake were associated with SSP risk.

Table 4

Evaluation of independent associations between modifiable factors and polyp risk; the Tennessee Colorectal Polyp Study


This analysis assesses modifiable lifestyle risk factors in a screening colonoscopy-age population to evaluate risk factors for SSPs and to compare them with other common colorectal polyps. Given the recent identification within the past 1–2 decades of SSPs as a CRC precursor, we are still in the infancy of understanding the aetiology of these lesions and which risk factors may be associated with these polyps. With their importance in the pathways' underlying progression to cancer and the relative difficulty in identification on colonoscopy, finding ways to assess risk in a population are of utmost importance. This is the first study to evaluate dietary intake with risk for SSPs and one of the largest epidemiological studies until now of SSPs. In initial models, we newly found that red meat, fat and fibre intakes were associated with SSP risk, and we also confirmed previous findings of associations with cigarette smoking and with NSAID use and a lack of association with alcohol use. After mutual adjustment, these associations remained for red meat intake, cigarette smoking and NSAID use.

Unlike a consistent association with polyp risk,14 ,18 cigarette smoking has been modestly and inconsistently associated with CRC risk.19 One possible reason for the inconsistency in past studies is a mixing of the types of CRC tumours which have different associations. Recent studies have more consistently identified smoking as a risk factor of MSI-high or CIMP-high CRC tumours which are part of the serrated pathway20 ,21. Indeed, cigarette smoking is strongly and consistently associated with risk of sporadic serrated polyps, including in this study,3 ,10–14 ,22 and risk of serrated polyposis syndrome.15 ,23 Smoking cessation has many benefits for health and we found cessation as short as 10 years was associated with decreased risk of all polyps compared with current smokers. Further, after cessation for more than 20 years, risk was similar to never smokers. This relationship was particularly strong for SSPs versus ADs.

NSAIDs and aspirin use may be an approach for colorectal neoplasia prevention; however, very little is known regarding NSAIDs and their association with SSP risk.10 ,13 ,24 ,25 A previous study of serrated polyps found an inverse association between aspirin use and serrated polyp risk, which was particularly strong for proximal lesions.10 In the only previous study to evaluate SSP risk, regular NSAID use was associated with reduced risk. We also observed this.13 We also found the reduction in risk associated with more than 10 years of use was stronger for SSPs than for ADs. The absence of an association with HP risk and the presence of an association with SSP risk may provide insight into the aetiology of SSPs and may be a distinguishing factor in inhibiting transition from HP to SSP. Thus, NSAID use may hold promise as a chemopreventive strategy for SSPs and should be evaluated in future studies.

Body composition and exercise are well-studied modifiable factors evaluated in AD and CRC risk.26 ,27 An association between SSP risk and obesity is currently equivocal.10–13 Although we observed a statistically significant association between BMI-defined obesity and colorectal polyp risk in all case groups, after adjustment for other factors, a statistically significant association was no longer observed for SSP risk. Interestingly, no association was observed for physical activity measures, including a measure of intensity and duration (MET hours). Both of these findings are consistent with a previous study which found no association between either BMI or hours of exercise and SSP risk.13

Dietary fibre has been speculated to protect against polyp formation by bulking the stool and increasing transit time, which may decrease the surface area of the colon exposed to carcinogenic toxins and bile acids within faecal matter.28 Although we initially observed decreased risks of ADs with fibre intake, these associations did not persist in subsequent models after adjustment for other risk factors. However, a suggestive borderline significant inverse association was observed with highest fibre intake and SSP risk. Future studies with a larger sample size are needed to confirm this finding. Likewise, both calcium and folate intakes initially appeared to be associated with decreased risk of SSPs. However, the associations disappeared after adjustment for other factors. Thus, this study does not support a strong relationship between calcium or folate intakes and SSP risk (although these factors should be evaluated in future larger studies). This result is also consistent with the findings from recent randomised trials in which supplementation of calcium or folic acid has not successfully decreased risk of AD recurrence.29–32

Red meat intake is consistently reported as a risk factor for CRC and colorectal ADs,33 ,34 although it has not previously been known whether an association exists between red meat intake and SSP risk. The risk of MSI-high CRC, for which SSPs are the presumed precursor lesion, is increased with well-done red meat intake, suggesting a possible role of red meat intake in SSP risk.35 Consistent with this finding, we found, for the first time, that high consumption of red meat was strongly associated with SSP risk. Interestingly, we also found that higher dietary fat intake was associated with risk of SSPs but not ADs or HPs; however, this relationship did not maintain statistical significance after adjustment for other risk factors. These included red meat intake, which may have been due to our sample size, over adjustment or may suggest that the fat intake association is a potential measure of red meat intake. Previous studies of serrated polyps have observed an association between high fat diet and serrated polyp risk, although this was not specific to SSPs.10 The potential mechanism behind an association is unclear, and further studies are needed.

There are several strengths within this study. Until now, it is one of only two large studies evaluating modifiable lifestyle risk factors for SSPs and is the first study to evaluate dietary risk factors for SSPs.13 We were able to rigorously standardise the diagnosis of all polyps regardless of initial clinical diagnosis using recently developed standards for HPs, ADs and SSPs and our observed prevalence of SSPs was consistent with recent prevalence studies.3 ,36 We were able to comprehensively evaluate several different modifiable factors.

There are also weaknesses in this study, which we attempted to limit. As with all case–control studies, we cannot exclude the possibility of recall bias, although it may have been minimised because colorectal polyps are a benign diagnosis and the data collection period was short. Recent studies have indicated that several factors may contribute to detection rates of polyps, including the quality of the bowel cleansing and withdrawal time.37 We did not collect data on these factors and so cannot exclude the possibility of missed polyps, which may have resulted in case misclassification. Given that SSPs are relatively rare, our SSP case group also included individuals with synchronous ADs (43%), which could potentially have affected the results if the risk factor was associated with ADs and not SSAs. However, we also observed associations that were only present for SSP risk, suggesting that we were able to evaluate risk factors for SSPs. We did perform sensitivity analysis by examining individuals with SSPs who did not have any ADs (see online supplementary tables). Although this diminished statistical power, we observed very similar results for all factors analysed as we observed when including individuals with ADs in the SSP case group, thus, indicating that the presence of an AD was not likely driving the observed associations. We may have failed to detect an association because statistical power in some of the subgroup analyses could have been limited. Thus, future larger studies are needed. Although this study included individuals with a wide range of characteristics and behaviours and we observed associations which both increased and decreased risk, we cannot exclude the possibility that individuals who receive colonoscopies are different in ways from individuals who do not receive colonoscopies, which may affect the observed associations in an unknown manner.

In summary, this study provides an extensive evaluation of lifestyle risk factors for SSPs and a comparison of risk for SSPs with risks for ADs and HPs. Given that SSPs are difficult to detect and fully remove on endoscopic screening and may accelerate to a dysplastic state quicker than ADs,4 ,8 ,9 ,38–40 primary prevention of SSPs through lifestyle modification may be an important strategy. The study found that many of the same risk factors are shared between ADs, HPs and SSPs. Thus, preventive efforts to reduce risk factors in ADs may also be applicable to SSPs. The study also found some differences in risk factors between the polyp types. Larger studies of SSPs will be needed to confirm these findings and future studies should also evaluate potential interactions of these risk factors with genetic or molecular risk factors, as well as preventive strategies that may be unique to SSPs.


The authors wish to thank Dr Kay Washington for her assistance with SSP diagnosis. The authors also wish to thank the many research staff and investigators who have contributed to the TCPS. Finally, the authors thank the study participants who contributed their time and biospecimens for research.


View Abstract


  • Contributors RMN, WZ and MJS contributed to study conception, design and supervision. TS, WES, RMN and MJS contributed to acquisition of data. TS and MJS provided administrative, technical or material support. JRD, ZZ, HGC, WZ and MJS contributed to analysis and interpretation of data. All authors contributed to writing, review and/or revision of the manuscript and approved the final manuscript. MJS is the guarantor of the submitted manuscript.

  • Funding This study was supported by grants P50CA950103, R01CA97386 and K07CA122451. JRD was supported by the Molecular and Genetic Epidemiology of Cancer fellowship (R25CA160056). Surveys and sample collection and processing for this study were conducted by the Survey and Biospecimen Shared Resource, which is supported in part by P30CA68485. The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health. A portion of the participants were studied as the result of resources and the use of facilities at the VA Tennessee Valley Healthcare System.

  • Competing interests None declared.

  • Ethics approval Written informed consent was obtained from all study participants, and the study protocol was approved by the institutional review board at each study site.

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

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.