Self-reported Metabolic Risk Factor Associations with Adenomatous, Sessile Serrated, and Synchronous Adenomatous and Sessile Serrated Polyps

Colorectal cancer comprises a major portion of worldwide cancers; recent estimates rank colorectal cancer third in global cancer incidence (after lung and breast cancer) and second in mortality (1, 2). Until recently, sporadic colorectal cancer was thought to predominantly arise from the cancerous precursor adenomatous polyps (AP), also known as conventional adenomas (3). Less is known about the risk factors for sessile serrated polyps (SSP), also called sessile serrated adenomas, which are less common than APs and more difficult to detect during colonoscopy (3). Hyperplastic polyps (HP) are considered benign serrated polyps. The distinction between SSPs and HPs was appreciated only recently as awareness of the cancerous potential of SSPs grew; less is known about risk factors specific to SSPs (3, 4). Synchronous SSPs and APs (synch polyps) may signal increased risk for cancer progression (5, 6) but risk factors for their cooccurrence need investigation.

Modifiable risk factors have been implicated for colorectal cancer and polyps (7, 8). In men, up to 50% of colon cancers and 33% of distal colon adenomas are potentially preventable with a reduction of modifiable risk behaviors, including obesity, smoking, alcohol consumption, physical activity, red meat intake, and folic acid intake (7) and, in women, 37% of colon cancers are also estimated as preventable (8). Therefore, characterization of modifiable risk factors for precancerous lesions in various populations remains an important pursuit. While the metabolic risk factors associated with APs have been investigated, few studies have addressed SSPs or synch polyps in parallel with APs to understand the associations of these polyp types with metabolic risk factors.

Colorectal cancer trends have evolved with some countries experiencing a decrease due to improved screening, awareness, and preventative measures and other countries, an increase, likely reflective of the adoption of less favorable lifestyle habits, such as a Western diet (2, 9). The prevalence of noncommunicable diseases such as metabolic syndrome (MetS or MS), obesity, or diabetes have increased, augmenting the potential for additional health complications (10). APs have been associated with increased weight (11–13) or variably with diabetes (14) or other conditions such as MetS (15). Studies of risk factors for SSPs and synch polyps have been infrequent and incomplete (3, 6, 16–18).

We investigated the relationship between metabolic risk exposures and APs, SSPs, and synch polyps in a prospectively enrolled U.S. screening colonoscopy population. Herein, we assess the association of self-reported metabolic risk factors, including increased body mass index (BMI), diabetes, hypertension, and hyperlipidemia, on the risk of histologically distinct colon polyps.

Data for this study come from The Johns Hopkins Biofilm Colonoscopy Study approved by the Johns Hopkins Medical Institute (JHMI; Baltimore, MD) Institutional Review Board (IRB) for human research, IRB00094020 (principal investigators: C.L. Sears, F. Giardiello). Eligible participants were prospectively enrolled by written informed consent prior to colonoscopy between August 2016 and February 2020 at three sites: Digestive Diseases Associates in Wyomissing, PA; Greenspring Station Endoscopy Center in Lutherville, MD, and the Johns Hopkins White Marsh Endoscopy Center in Baltimore, MD. From August 2016 to February 2020, 2,091 individuals were enrolled in the study; of these, 40 were withdrawn from the study (38 had consent withdrawn by the supervising physician, one had a canceled colonoscopy at visit, and one participant withdrew consent); thus, 2,051 individuals were evaluated for inclusion in the current study. The design of the study is described elsewhere (19). Patient studies were conducted under the ethical guidelines outlined in the Belmont Report and U.S. Common Rule.

Participants included in this analysis were adults (ages 40–85) enrolled in the Johns Hopkins Biofilm Colonoscopy Study who had an intact colon and complete colonoscopy with visualization to the cecum. Participants with exclusively hyperplastic polyps, missing pathology on polyps, having rare pathologic outcomes, or missing information on some of the covariates of interest were removed from the analysis. Participants with no polyps at study visit, but with a reported history of polyps were also excluded. The Johns Hopkins Biofilm Colonoscopy Study excluded pregnant women, prisoners, individuals with inflammatory bowel disease, and individuals on blood thinners or antiplatelet drugs. Figure 1 shows the participant flow and reasons for exclusion from this analysis. The final study population included in this analysis was 1,370.

Polyp diagnoses at the current colonoscopy were abstracted from the medical record to classify study participants. The precise location, size, diagnosis, and other characteristics of the colorectal polyps were collected from colonoscopy and pathology reports. To standardize polyp diagnosis, histopathology reports of all extracted or biopsied polyps were systematically reviewed by one gastroenterology physician member of the study team (S. Rifkin). Polyp cases were classified according to the presence, number, and synchronicity of HPs, SSPs, and APs. AP cases had one or more tubular, tubulovillous, or villous AP with or without dysplasia and with or without synchronous HPs. The SSP cases had one or more SSPs, with or without synchronous HPs. The synchronous cases have at lease one AP and one SSP with or without synchronous HPs. HPs were considered benign. Advanced polyps were defined as polyps that were greater than or equal to 1 cm in size, contained villous components and/or dysplasia. Advanced synchronous polyps were defined as having an advanced polyp of at least one AP or one SSP or both.

Controls were individuals without any evidence of polyps and without a reported history of polyps.

Participants were defined as having diabetes mellitus, hypertension, or hyperlipidemia if they self-reported these conditions in their medical histories. These medical histories were collected or confirmed at study visit by a member of the health care team. No information on the severity or length of these conditions was available. Height and weight measurements, either self-reported, abstracted from the medical record, or measured at enrollment, were used to calculate BMI. BMI categories utilized in analysis were <25 kg/m², ≥25–<30 kg/m², and ≥30 kg/m² defined as normal weight, overweight, and obese, respectively.

At colonoscopy, study participants were administered a questionnaire including information about sociodemographic factors such as race and employment status, medical and surgical history, basic dietary patterns, medication use [including antibiotics, nonsteroidal anti-inflammatory drugs (NSAIDs), proton pump inhibitors], risk factors for colorectal cancer including family history of colorectal cancer, smoking history, alcohol use, physical activity, and history of prior colonoscopy with pertinent findings.

To study the association between colorectal polyps and metabolic risk factors, covariates considered for adjustment were family history of colorectal cancer, dietary patterns of meat (including poultry) and alcohol consumption, use of NSAIDs, age, sex, race, smoking history, physical activity at work, and current physical activity. Smoking measures were assessed in pack years from participant report of the average number of cigarettes they smoked per day during the number of years that they reported smoking. An average was used to calculate pack years if participant provided a range. Pack-years smoking was then categorized into no pack-years, >0–10 pack-year smoker, 10–20 pack-year smoker, and >20 pack-year smoker. Finally, individuals who mentioned having exclusively smoked something other than cigarettes (cigars, pipe, marijuana, hookah) were combined into one category, though this list may not be exhaustive as this question was not explicitly asked. For NSAID usage, reports of regular usage of aspirin or other NSAIDs were combined.

For Table 1, stratification was performed by polyp type to compare included and potential covariates. Choice of considered predictors for adjustment were selected from prior studies of risk factors for colon polyps. For final selection of adjustment variables, results from univariate analyses, log likelihoods, and the Akaike information criterion were considered (20). Final adjustment covariates selected were age, sex, race, and smoking history.

The association of each metabolic risk exposure of interest with the polyp outcome was assessed individually using multinomial logistic regression to derive adjusted conditional odds ratios (aCOR) and 95% confidence intervals (95% CI) as an estimate of effect size.

Analysis of the relationship of the metabolic risk factors with advanced polyps was individually assessed using a multinomial logistic regression for each polyp type, where the outcomes were no polyps, nonadvanced polyps <1 cm, or at least one advanced polyp of >1 cm.

In current study, sex was adjusted for as a covariate and then a stratified analysis by sex was performed to test if any differences were observed.

In the final analyses, a P of <0.05 was considered statistically significant. All analyses were performed using Stata version 15.1.

The Biofilm study was approved by the Johns Hopkins Hospital IRB: IRB00094020 (principal investigators: C.L. Sears, F.M. Giardiello).

All participants provided written informed consent.

Of 2,051 individuals completing the first study visit, 1,370 (66.8%) participants were included in the data analysis. Table 1 shows demographic characteristics of study participants, based on polyp classification. Of the 1,370 participants studied, there were 554 control participants (40.4%) without polyps and without a reported or unknown history of polyps, including 191 males and 363 females; 645 participants (47.1%) with APs, 325 males and 320 females; 101 participants (7.4%) with SSPs, 37 males and 64 females; and 70 participants (5.1%) with synch polyps, 40 males and 30 females. APs and synch polyps were more prevalent in males than in females (APs: 54.8% vs. 41.2%; synch: 6.7% vs. 3.9%). SSPs were more prevalent in females than in males (8.2% vs. 6.2%). At a power of 0.80 and a two-sided alpha of 0.05, the minimally detectable ORs for this study are 1.52, 2.58, 3.34 for APs, SSPs, and synch polyps, respectively.

Metabolic risk exposures included 39.9% of participants with hypertension (546 participants), 16.9% with hyperlipidemia (232 participants) and 13.3% with diabetes (182 participants). BMI measurements revealed 36.0% were overweight (493 participants), and 39.9% were obese (546 participants). Only 17.2% of participants had no metabolic risk factors whereas 38.5% of participants had one metabolic risk factor, 28.4%, two, 13.0%, three, and 2.9%, all four of the metabolic risk factors examined (BMI > 25 kg/m², hypertension, hyperlipidemia, diabetes). The differential impact of these risk factors on APs, SSPs, and synch polyps are described below.

Being obese or overweight or having diabetes was associated with an increased odds of APs compared with those with a normal BMI [aCOR for overweight and obese, 1.54 (1.11–2.13) and 2.09 (1.52–2.88), respectively] (Fig. 2A). The P value for trend for the association of increased BMI with APs was P < 0.001. Having diabetes versus no diabetes resulted in an aCOR of 1.59 (1.10–2.29; Fig. 2A). Hypertension or hyperlipidemia were not statistically significantly associated with an increased risk of APs (Fig. 2A).

After stratification by polyp size, advanced APs were associated with obesity [aCOR of 3.10 (1.53–6.26)] and with hypertension [aCOR of 2.32 (1.38–3.91)]. The association with hypertension was not seen when the polyp sizes were combined. Advanced APs were not associated with having diabetes or being overweight. (Table 2).

When stratified by sex (Fig. 3), diabetes was only associated with APs in females [aCOR of 1.83 (1.15–2.93)] whereas an overweight BMI was only associated with APs in males [aCOR of 2.35 (1.38–3.99)]. An obese BMI was associated with APs in both males and females [aCOR values of 3.01 (1.75–5.20) and 1.76 (1.20–2.60), respectively] (Fig. 3A).

No statistically significant associations between the metabolic risk exposures and SSPs were detected herein. This was true in an individual assessment of each metabolic risk factor (Fig. 2B), or when stratified by polyp size (Table 2), or by sex (Fig. 3B).

Finally, the odds of synch polyps increased in participants with hypertension versus those without hypertension [aCOR of 1.93 (1.14–3.25)] and in those with obesity versus a normal BMI [aCOR of 2.18 (1.06–4.48)]. Hyperlipidemia, diabetes, or being overweight were not associated with an increased risk of synch polyps in this population (Fig. 2C).

Stratifying synch polyps by size revealed that only advanced synch polyps were associated with hypertension [aCOR of 3.83 (1.25–11.7)] or an obese BMI [aCOR of 13.84 (1.37–139.74)]. Nonadvanced polyps did not display associations with the metabolic risk factors examined (Table 2).

No statistically significant associations were seen for synch polyps when stratified by sex (Fig. 3C).

Elucidating the risk factors associated with different polyp types may help in identifying individuals at greater risk clinically for these polyps and assist physicians in conversations encouraging patients to undergo screening colonoscopy. Our study used well-defined polyp categories to elucidate differences in four metabolic risk factors for three types of polyps. We identified that associations of metabolic risk exposures differed between histologically distinct polyp types when assessing each exposure independently adjusted for age, sex, race, and smoking status. Being overweight, obese, or diabetic independently associated with APs, while hypertension associated exclusively with advanced APs. Metabolic risk factors were not significantly associated with SSPs, suggesting that metabolic risk factors impact AP but not SSP development. Consistent with the AP alone data, obesity or hypertension increased the risk of synch polyps, in particular, advanced synch polyps. Thus, synch polyps behaved more like APs versus SSPs, but possibly with stronger risk.

Herein increased BMI, and particularly obesity, was a strong risk factor for APs. Although the associations for BMI and APs vary in the literature, our results are consistent with assessment via meta-analyses (11–13) whether using physician-measured BMI (11) or self-reported measures (12, 13). Our study assessed BMI mostly using self-reported medical histories. As compared with those with normal weight, being overweight increased the odds of APs by approximately 50% and being obese more than doubled the odds of APs (∼100% increase; P_trend <0.001). This dose-response relationship was also seen in the meta-analysis by Okabayashi and colleagues, which included self-reported measures, but not seen in the meta-analysis by Wong and colleagues, which included studies with only physician-measured BMI (11, 12). In contrast, in the meta-analysis by Ben and colleagues, adenoma risk increased only with obese BMI but not overweight BMI (13). Our effect estimations for BMI were higher than in the meta-analyses despite similar categorization of BMI although the CIs overlapped (11–13). One reason our estimates may differ include the decision to remove individuals with a reported history of polyps from our control group, potentially diminishing homogeneity between our cases and controls and strengthening the association between obesity and APs.

Additional differences among the studies exist. In the Wong and colleagues meta-analysis, the odds of APs was higher in females with a BMI >25 kg/m² (vs. females with a normal BMI; ref. 11), whereas we found males to have a higher odds of APs for both overweight and obese BMI. The meta-analysis by Ben and colleagues found no significant differences in the associations by sex (13), and the meta-analysis by Okabayashi and colleagues found overweight and obesity to be significant risk factors only for females (12). While it is not clear why AP risk differed by sex in our population, menopausal status, not explicitly collected in our study, may play a role, with a potentially greater adenoma risk being seen in premenopausal women (12).

Our study associated diabetes with APs [aCOR of 1.59 (1.11–2.13)]. This association was not seen for advanced APs herein, and stratified analysis by sex found this association only in females. The association between diabetes and APs has been noted previously and detected by meta-analysis in both nonadvanced and advanced adenomas; however, the sample size was too limited to assess risk by sex via meta-analysis (14). In contrast, other data in women did find an association with adenomas and diabetes (21). Other studies have associated diabetes with having multiple polyps (22) or have found an association with adenomas and diabetes in diabetic subgroups such as in younger populations (23) or in those with other comorbidities such as chronic kidney disease (24). However, not all studies have detected an association between diabetes and APs and differences in underlying population characteristics and/or study design (retrospective vs. prospective) may be responsible (14). Potentially consistent with an association of diabetes and polyp risk, metformin, a medication commonly used to treat type 2 diabetes, has been investigated as a treatment for colorectal adenomas; two meta-analyses revealed that metformin use was associated with reduced risk of colorectal adenoma (25, 26). Metformin's protective effects may vary by population subtypes; it may potentially be more protective in populations with higher diabetes prevalence (26). Thus, as yet, ill-defined differences amongst study populations likely contribute to adenoma risk with diabetes.

Mechanisms behind these associations are not fully understood, but several have been proposed (11). For example, by meta-analysis, levels of adiponectin, an adipokine protein hormone important in limiting glucose levels, were decreased in individuals with adenomas and colorectal cancer (27); this suggests greater exposure of the colon epithelium to glucose excess. Elevated insulin levels, commonly resulting from obesity or other similar morbidities, and the insulin to IGF axis may create a favorable environment for formation of colorectal tumors (28, 29). Furthermore, the inflammatory state found in obesity may contribute by increasing tumorigenic cytokines like IL6 (28). Kim and colleagues considered the neutrophil-to-lymphocyte ratio (NLR) in association with colorectal adenomas as an inflammation marker revealing a strong independent association between higher NLR and colorectal adenomas after controlling for age, gender, BMI, smoking, alcohol, and exercise (30). Similarly, elevated C-reactive protein, a marker of inflammation, has been associated with adenomas (31). Understanding mechanisms contributing to polyp development is key to providing insight into approaches for modifying lifestyle factors to lower polyp risk and identify risk factors for SSPs.

Unexpectedly, herein, hypertension was associated with advanced, but not nonadvanced, APs. Hypertension more than doubled the odds of advanced APs. In limited data, other researchers have also found this association albeit in small studies or in association with AP recurrence (32, 33). Of interest, Huang and colleagues also found that hypertension was significantly associated with advanced, but not nonadvanced, APs (34). A few studies suggest that antihypertensive medications modify the relationship between polyps and hypertension; however, outcomes are variable, such as increased risk associated with antihypertensive medications (35); β-blockers associated with decreased risk of colon cancer (36, 37); or no impact of calcium channel blockers on colon cancer risk (38). Collectively, our study adds to the plausibility of the association of hypertension and AP risk, especially in advanced APs. However, further study, including mechanisms, is warranted.

Our data display a distinct difference in metabolic risk factor associations between APs and SSPs, where no associations with SSPs were identified. Information on risk factors for SSPs are limited because risk factors for serrated polyps overall, which include the more common HPs, were typically studied as a whole before the serrated pathway to colorectal cancer was better understood (3). Studies have been hampered by small populations of SSPs (32). Because HPs are benign, the challenges with elucidating an understanding of SSP risk factors in studies that combine HPs and SSPs has been noted (16). A meta-analysis evaluating risk factors for serrated polyps overall found BMI, among other nonmetabolic risk factors, to be associated, but in a substudy investigating risk factors specifically for SSPs this association was not seen (39). Consistent with our analyses, even when metabolic risk factors were included in analyses, to date, only nonmetabolic risk factors for SSPs were reported (18, 40–42). Whether sample sizes have been too limited, or there is just a weaker effect of metabolic risk factors on SSPs, needs further study.

Herein obesity or hypertension was associated with increased risk of synch polyps, mirroring associations we observed for APs, although with stronger point estimates. Few studies have examined individuals with synch polyps. Limited observations suggest that synch polyps associate with increased risk of advanced polyps (6) and potentially associate with smoking and increased BMI, although the study finding this association grouped traditional serrated adenomas (TSA) with SSPs as a proxy for SSPs (17). Our study adds to a framework associating increased BMI with synch polyps. However, further study is needed to understand risk factors for synch polyps.

There were several limitations in the current study. The study data are cross-sectional and therefore information on temporality could not be ascertained. Metabolic risk factor assignments were based on self-reported medical histories, were not verified by prior medical records, and lacked quantitative precision (e.g., current HbA1C levels) as well as information on length and severity of these conditions. While these self-reported medical histories leave open the possibility of misclassification, we do not believe that there is differential misclassification among those with versus those without polyps and misclassifications would most likely have weakened (e.g., underreporting of prior conditions and/or excess weight) rather than strengthened the estimates presented in this study. Although a strength of our study was a larger population sample size of individuals with SSPs and synch polyps than in other studies, the sample size of these populations combined with metabolic risk factors remained small. For example, there were only six participants that had both diabetes and SSPs. Finally, the study participants were majority white, limiting generalizability.

Strengths of this study include information on prior polyp history which allowed for exclusion of participants with a previous history of polyps from the control group. This exclusion of individuals with a history of polyps has been done in the literature, though not consistently (32, 43–46). This allowed for sharpening of the categories of those with and without polyps, because a previous history of polyps is associated with an increased risk for polyp recurrence (47, 48). Furthermore, our populations of SSPs and synch polyps is larger than a number of previous studies and, importantly, our inclusion of SSPs was limited to histopathologically confirmed SSPs and did not include benign HPs or rare TSAs.

In conclusion, our findings suggest that histopathologically distinct colon polyp precursors to colorectal cancer differ in metabolic risk factors associations, which we postulate, reflect promotion of different pathways leading to colorectal cancer. Key is a distinction between APs and SSPs, with the former, but not the latter, associated with metabolic risk factors that may differ by sex. Obesity appears to be the strongest risk factor for both nonadvanced and advanced APs, while being overweight or having diabetes are also notable risk factors for nonadvanced APs. Hypertension may be a specific risk factor for advanced APs. However, larger studies are needed to assess these associations and, in particular, associations with SSPs and synch polyps. Our results suggest that self-reported medical history provides valuable insight to polyp risk, potentially enabling the use of larger retrospective studies of colonoscopy populations to assess knowledge gaps. As data accrue, more aggressive colonoscopy screening, critical to colorectal cancer prevention (49, 50), may be considered in populations of individuals with metabolic risk factors and modifiable lifestyle risk factors.

C.L. Sears receives grant support from Bristol Myers Squibb and Janssen. No disclosures were reported by the other authors.

C.N. Santiago: Data curation, investigation, formal analysis, methodology, writing–original draft, writing–review and editing. S. Rifkin: Conceptualization, formal analysis, supervision, methodology, writing–review and editing. J. Drewes: Writing–review and editing. G. Mullin: Investigation, writing–review and editing. E. Spence: Data curation, formal analysis, investigation, writing–review and editing. L.M. Hylind: Supervision, investigation, writing–review and editing. J.J. Gills: Supervision, investigation, writing–review and editing. D. Kafonek: Supervision, investigation, writing–review and editing. D.M. Cromwell: Supervision, investigation, writing–review and editing. L. La Luna: Supervision, investigation, writing–review and editing. F. Giardello: Conceptualization, supervision, funding acquisition, investigation, project administration, writing–review and editing. C.L. Sears: Conceptualization, formal analysis, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing.

This study was supported by grants R01CA196845 (C.L. Santiago, F. Giardello, C.N.Santiago, J. Drewes, G. Mullin, E. Spence, D. Kafonek, D.M. Cromwell, L. LaLuna), Bloomberg Philanthropies (C.L. Santiago, J.J. Gills), T32DK007632 (S. Rifkin), intramural funds (L.M. Hylind) and the Johns Hopkins Cancer Center Support Grant, NCI P30CA006973 (C.L. Santiago, F. Giardello). All data from R01CA196845 (Johns Hopkins) were stored in Research Electronic Data Capture (REDCap). The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the NCI or the NIH. The authors thank all members of the Sears laboratory for assistance with the Biofilm Colonoscopy Study.

Biofilm Study Consortium authors: Courtney Stevens (Johns Hopkins University School of Medicine, Baltimore, MD), Brent Tabisz (Johns Hopkins University School of Medicine), Marshall Bedine (Green Spring Station Endoscopy, Lutherville-Timonium, MD), Eduardo Gonzalez-Velez (Johns Hopkins University School of Medicine, Baltimore, MD), Hazel Marie Galon Veloso (Johns Hopkins University School of Medicine, Baltimore, MD), Pamela Schearer (Reading Hospital, Tower Health, Reading, PA), Stacy Gerhart (Digestive Diseases Associates, Reading, Wyomissing, PA), Amy Schiller (Digestive Disease Associates, Reading, Wyomissing, PA), Karin Donato (Digestive Disease Associates, Reading, Wyomissing, PA), Randi Sweigart (Digestive Disease Associates, Reading, Wyomissing, PA), John Altomare (Digestive Disease Associates, Reading, Wyomissing, PA), Nirav Shah (Digestive Disease Associates, Reading, Wyomissing, PA), Christopher Ibrahim (Digestive Disease Associates, Reading, Wyomissing, PA), Ravi Ghanta (Digestive Disease Associates, Reading, Wyomissing, PA).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.