Prevalence and Clinical Features of Sessile Serrated Polyps: A Systematic Review

Abstract

Background & Aims:

Sessile serrated polyps (SSPs) could account for a substantial proportion of colorectal cancers. We aimed to increase clarity on SSP prevalence and clinical features.

Methods:

We performed a systematic review of Medline, Web of Science, Embase, and Cochrane databases for original studies published in English since 2000. We included studies of different populations (United States general or similar), interventions (colonoscopy, autopsy), comparisons (world regions, alternative polyp definitions, adenoma), outcomes (prevalence, clinical features), and study designs (cross-sectional). Random-effects regression was used for meta-analysis where possible.

Results:

We identified 74 relevant colonoscopy studies. SSP prevalence varied by world region, from 2.6% in Asia (95% CI, 0–5.9%) to 10.5% in Australia (95% CI, 2.8%–18.2%). Prevalence values did not differ significantly between the United States and Europe (P=.51); the pooled prevalence was 4.6% (95% CI, 3.4%–5.8%), and SSPs accounted for 9.4% of polyps with malignant potential (95% CI, 6.6%–12.3%). Mean prevalence was higher when assessed through high-performance exams (9.1%; 95% CI, 4.0%–14.2%, P=.04) and with alternative definition of clinically relevant serrated polyps (12.3%; 95% CI, 9.3%–15.4%; P<.001). Increases in prevalence with age were not statistically significant, and prevalence did not differ significantly by sex. Compared with adenomas, a higher proportion of SSPs were solitary (69.0%; 95% CI, 45.9%–92.1%; P=.08), with diameters of 10 mm or more (19.3%; 95% CI, 12.4%–26.2%; P=.13), and were proximal (71.5%; 95% CI, 63.5%–79.5%; P=.009). Mean ages for detection of SSP without dysplasia, with any or low-grade dysplasia, and with high-grade dysplasia were 60.8 y, 65.6 y, and 70.2 y, respectively. The range for proportion of SSPs with dysplasia was 3.7%–42.9% across studies, possibly reflecting different study populations.

Conclusions:

In a systematic review, we found that sessile serrated polyps are relatively uncommon, compared with adenoma. More research is needed on appropriate diagnostic criteria, variations in detection, and long-term risk.

Graphical Abstract

INTRODUCTION

Colorectal cancer (CRC) is a leading cause of cancer deaths.¹ The disease develops primarily from precancerous lesions called adenomas.² In a series of seminal papers from the 1970s and 1980s,^{3, 4} this process was first characterized as relatively slow, with many adenomas resulting in relatively few cancers. Because adenomas are detectable and removable at endoscopy, CRC is highly amenable to screening, as was first demonstrated in the National Polyp Study in 1993.⁵

More recently, it has been suggested that CRC may also arise from an alternative pathway with serrated polyp precursors.^{6, 7} Because serrated polyps were generally considered innocuous until the World Health Organization (WHO) included them in its manual for tumor classification of the digestive system,⁸ relatively little is known about their epidemiology and risk.⁹ The serrated polyp family includes sessile serrated polyps (SSPs, also known as sessile serrated adenoma or -lesion), traditional serrated adenomas, and hyperplastic polyps. Hyperplastic polyps are the most common, and especially small distal hyperplastic polyps are not believed to harbor risk of malignancy. Traditional serrated adenomas are the most rare and can lead to CRC. SSPs appear to be substantially less common than adenomas,¹⁰ but their associated molecular features may be overrepresented among CRC cases,^{11, 12} suggesting potentially higher risk of CRC compared to adenomas.

In this study, we sought to increase clarity on the prevalence and clinical features of SSPs by systematically reviewing the relevant literature from the last two decades. Review outcomes included prevalence by age, sex, and clinical definition; the multiplicity, size, and localization; and the proportion with cytological dysplasia, the pathological bridge to cancer.

METHODS

General study design

The review was planned and designed in consultation with Stanford and Erasmus University librarians, and consisted of six steps: (i) defining scope; (ii) literature search (iii) literature review and selection; (iv) quality appraisal; (v) data abstraction; and, (vi) statistical analysis. Two different investigators performed steps (iii)–(v) independently, with disagreements resolved through consensus.

Scope of the review

We addressed several research questions under two main headers (Suppl.Table 1): (1) prevalence, and (2) clinical features of SSPs. Specific research questions under the first header included the prevalence by (1a) calendar year, (1b) world region, (1c) clinical definition, (1d) age/sex, (1e) indication for the prevalence assessment (e.g. CRC screening vs. follow-up of occult blood in stool), and (1f) quality of the examination (with high quality defined as examination with enhanced endoscopes or by providers from the upper quartile [minimum] of SSP detection rate); (1g) the fraction of SSPs among all potentially precancerous polyps (including adenomas, SSPs, or traditional serrated adenomas) was also assessed to gain insight into the relative proportions of polyps vs. CRCs with characteristics of the serrated pathway.¹³Specific research questions under the second header included polyp number (2a), size (2b), anatomic location (2c), co-existence with adenoma (2d), and presence (2e) and age at detection (2f) of cytological dysplasia.

Most research questions focused entirely on SSPs according to strict histopathological criteria.⁸ Since what is considered a “clinically relevant” serrated polyp, and pathologists’ attention to the difference between sessile serrated and hyperplastic polyps have all evolved over time, under research question 1c we also considered various more liberal definitions for clinically relevant serrated polyps as encountered in the literature, including large and/or proximal hyperplastic polyps in addition to the more strict histological definition. Under research question 1c only, we also considered inclusion of traditional serrated adenoma. Reported definitions were grouped into three different categories: 1) older terminology used before the 2010 WHO classification of tumors,⁸ consisting of large hyperplastic polyps (≥10mm diameter), proximal hyperplastic polyps (located proximal of the splenic flexure), and serrated adenoma (including but not distinguishing SSPs and traditional serrated adenoma); 2) strict terminology similar to histopathological criteria from the 2010 WHO classification manual, consisting of only SSP, and traditional serrated adenoma; and 3) more recently introduced composite definitions including clinically-relevant histology, size and location. The latter category combined SSP and traditional serrated adenoma, with either any large hyperplastic polyps (with large distal polyps excluded for some studies) or any proximal hyperplastic polyps (with large distal polyps included for some studies).

Literature search

Ovid Medline, Web of Science, Embase, and the Cochrane database were searched from Jan 1, 2000 through April 19, 2018 for original studies on serrated polyps written in English. General search terms were used to minimize the risk of missed information (Suppl.Methods I).

Literature review and selection

Covidence systematic review software was used to manage references, discard duplicates, review literature, and compare reviewer decisions. Selection of studies by each reviewer was based on scanning titles and abstract followed by potential full text review. Studies were selected according to predefined criteria for Population, Intervention, Comparison, Outcome, and Study Design (PICOS) (Suppl.Table 1). The target population was the U.S. general population, with data from other world regions included for research question 1a–b, and for other research questions if not statistically significantly different from the U.S. under 1b. The intervention was colonoscopy or autopsy with pathological assessment of findings. The comparators depended on the research question, and included prevalence data from other world regions and CRC incidence data¹⁴ for research question 1b, alternative definitions for clinically relevant serrated lesions for research question 1c, and adenoma data from 11 autopsy studies (identified elsewhere¹⁵ and not systematically reviewed here)^16–26 for research questions 1d and 2a–c. Outcome of interest was percentage of the population with SSPs or the percentage of SSPs with the feature of interest, relying on study criteria for diagnosis of SSPs. The study design was cross-sectional.

Quality appraisal

There is no standard instrument for quality appraisal of prevalence studies. Therefore, we modified a validated tool designed for appraisal of studies measuring the prevalence of lower back pain,²⁷ answering only the 8 of 10 questions deemed pertinent to our study design (Suppl.Methods II). Assessment criteria concerned study characteristics such as the method of sampling, the case definition, and the study instrument. Overall risk of bias was qualified low, moderate, or high depending on the level of confidence in the study estimates.

Data abstraction

Data were stored in a spreadsheet template designed to enable easy cross-reviewer comparison. We collected all relevant outcome data, as well as all study characteristics relevant to outcome interpretation or sub-analyses. For each outcome, both the numerator and denominator were recorded (e.g. the numbers of polyp patients and total study participants). Age- and sex-specific data were ascertained to the extent available.

Statistical analysis

Random-effects regression was used to meta-analyze outcome data where possible. T-tests (Wald) were used to examine the differences in outcomes against comparators defined in Suppl.Table 1 – first for prevalence by world region, then for other outcomes including all world regions with prevalence not statistically different from the U.S at a standard 5% significance level. Q- and I²-statistics were calculated to measure outcome heterogeneity across studies. Bar charts were used with forest plots to visualize meta-analysis results. Linear models and T-tests (Wald) were used to examine overall and region-specific trends in prevalence by calendar year. Log-linear models were fitted for overall and study-specific trends by age.

All analyses were performed using R statistical software version 3.6.2.

RESULTS

Study selection

The literature search returned 4,462 references through April 19, 2018 (Suppl.Figure 1). After duplicate removal, 2,123 studies were selected for review. During abstract review, 1,669 irrelevant studies were removed. During full text review, an additional 380 studies were excluded. In total, 74 studies on serrated polyps were included, of which 69 reported on prevalence,^28–96 and 21 reported on clinical features.^{28, 37, 39, 47, 50, 52, 55–57, 65, 67, 70, 86, 88, 91, 92, 97–101} Of the excluded studies, some were relevant but discarded due to overlap with larger studies from the same source population.^102–107 For the same reason, specific outcomes for some of the included studies were excluded.

Study characteristics and quality appraisal

All included studies used colonoscopy in combination with pathology review as the instrument for serrated polyp detection. No autopsy studies were identified. Study setting ranged from community hospital to academic medical center from four different world regions, U.S. (n=36), Europe (n=23), Asia (11), and Australia (n=4). Although inclusion criteria varied, studies often included exams for screening or symptoms, and excluded patients with a history of polypectomy, CRC, colectomy, familial risk syndromes, inflammatory bowel disease, or a recent colon examination. Generally, the study exams were performed between 2005–2015, diagnostic criteria were based on the WHO classification manual or various earlier sources, mean patient age was around 60 years, and average % female was around 50–60%. For details see Suppl.Table 2.

Since study enrollment was non-random (based on selection for colonoscopy instead of random sampling of the general population), and outcome assessment is operator-dependent (colonoscopy and pathology),⁴⁹ the risk of bias for the included studies was judged to be moderate to high (Suppl.Table 3).

Prevalence of sessile serrated polyps by study publication year and size

Few reports on the prevalence of SSPs were published before 2010, but the number of reports increased dramatically after the serrated polyp was first included in the 2010 WHO manual for tumor classification (Suppl.Figure 2).⁸ The range for reported prevalence of SSPs at colonoscopy across all studies was very broad (0.0, 95%CI 0.0%–0.1% in Cao et al.⁴⁰, up to 20.0%, 95%CI 17.3%–23.2% in Bettington et al.).³⁶ Despite an increasing awareness of SSPs, there was no significant upward trend in reported prevalence over time across all studies (+0.1% per year,95%CI −0.2%–0.4%; P=.49). However, trends varied by world region, and did increase among U.S. studies specifically (+0.3% per year, 95%CI 0.0%–0.7%; P=.03) (Suppl.Figure 2).

Prevalence of sessile serrated polyps by world region

The prevalence of SSPs at colonoscopy in the U.S. was 5.1% (95%CI 3.6%–6.5%) (Suppl.Figure 3–4). Prevalence was higher in Australia (10.5%, 95%CI 2.8%–18.2%; P=.03) but not statistically different in Asia (2.6%, 95%CI 0–5.9%; P=.14) and Europe (3.9%, 95%CI 2.0%–5.9%; P=.51) compared with the U.S. (Suppl.Figures 3–4). Excluding one extreme statistical outlier,⁴² mean prevalence for Asia was statistically different from the U.S. (0.9%, 95%CI 0.5%–1.3%; P=.004). The pattern in polyp prevalence across regions was qualitatively similar to the reported CRC incidence pattern,¹⁴ although relative prevalence differences appear larger (Suppl.Figure 3).

Mean overall prevalence of SSPs at colonoscopy across all U.S. and European studies was 4.6% (95%CI 3.4%–5.8%) (Figure 1).

Figure 1.

Prevalence of sessile serrated polyps across included studies.

U.S. and European studies. Squares=means; square size=population size; whiskers=95% CIs; polygon=pooled mean, random-effects meta-regression; ***=significant heterogeneity, P<.001.

The mean fraction of all potentially precancerous polyps with sessile serrated histology in U.S. and European studies was 9.4% (95%CI 6.6%–12.3%) (Figure 2).

Figure 2.

Fraction of polyps with sessile serrated histology across included studies.

U.S. and European studies. Squares=means; square size=population size; whiskers=95% CIs; polygon=pooled mean, random-effects meta-regression; ***=significant heterogeneity, P<.001. Precancerous lesions include conventional adenoma, SSPs, and traditional serrated adenoma.

Even within U.S. and European studies there was substantial heterogeneity in estimates, with study-specific prevalence ranging from as low as 0.4% (95%CI 0.0%–0.8%)⁶⁴ to as high as 13.8% (95%CI 9.0%–18.6%; Q=1716, P<.001) (Figure 1),⁶⁸ and with the fraction of polyps with sessile serrated histology ranging from 3.9% (95%CI 3.3%–4.6%)³⁷ to 25.4% (95%CI 19.5%–31.3%; Q=2430, P<.001) (Figure 2).⁶⁸

Prevalence according to definition for clinically relevant serrated polyps

Under older terminology, the prevalence of clinically relevant serrated polyps in U.S. and European studies varied from 1.8% (95%CI 1.3%–2.2%) for large hyperplastic polyp, to 3.3% (95%CI 1.6%–5.0%) for serrated adenoma, to 8.5% (95%CI 6.7%–10.4%) for proximal serrated polyp (Figure 3, Suppl.Figure 5). The latter was significantly higher than the prevalence with the strict SSP definition (4.6%, 95%CI 3.4%–5.8%; P<.001).

Figure 3.

Prevalence of serrated polyps according to clinical definition.

U.S and European studies. Whiskers=95% confidence intervals; ***=significant difference vs. SSP prevalence, P<.001. The “clinically relevant” grouping includes definitions combining sessile serrated, traditional serrated polyps and hyperplastic polyps by size (>5 or >10mm) and/or location (proximal). See Suppl.Figure 5 for corresponding Forest plots and more detail on polyp definitions.

Under strict terminology, the prevalence of SSPs and traditional serrated adenomas combined (4.4%, 95%CI 2.9%–5.9%) was somewhat lower than the prevalence of only SSPs (Figure 3), but this difference was not significant (P=.93) and was due to study selection, since not all studies reported on both histologies. The prevalence of SSPs alone in the studies reporting on both SSPs and traditional serrated adenoma was 4.1% (95%CI 2.6%–5.5%), suggesting a net prevalence for traditional serrated adenoma of approximately 0.3%.

Under more liberal definitions for clinically-relevant histology, size and location, the prevalence estimates were higher than under the strict SSP definition, ranging from 5.6% (95%CI 3.3%–7.9%; P=.44) for definitions based primarily on histology and large size, to 12.3% (95%CI 9.3%–15.4%; P<.001) for definitions based primarily on histology and proximal location (Figure 3).

Prevalence of sessile serrated polyps according to exam indication and quality

The prevalence of SSPs in U.S. and European studies was not related to exam indication, i.e. whether the colonoscopy detecting the polyps was performed for screening in asymptomatic adults vs. for screening or other indications (4.3%, 95%CI 2.4%–6.1% vs. 4.5%, 95%CI 2.9%–6.0%, P=.89) (Figure 4, Suppl.Figure 6). Although prevalence at a patient’s initial screening exam was higher (5.3%, 95%CI 0%–11.0%; P=.73), this estimate was based on only 3,818 exams from two studies, with wide confidence intervals (Figure 4).

Figure 4.

Sessile serrated polyp prevalence according to exam indication and quality level.

U.S. and European studies. Whiskers= 95% confidence intervals; *=significant difference vs. prevalence in unselected exams, P<.05. Good–Excellent bowel preparation=by Aronchick scale, 7–9 Boston Bowel Preparation Score, or split-dose; high-quality examinations=enhanced endoscopes or provider from upper quartile of sessile serrated polyp detection. See Suppl.Figure 6 for Forest plots and definitions.

Prevalence was higher when patients were examined by high polyp detectors or with enhanced endoscopes (9.1%, 95%CI 4.0%–14.2% for these high-performance exams, vs. 4.7%, 95%CI 3.5%–5.9% for unselected exams; P=.04) (Figure 4, Suppl.Figure 6). Prevalence was also higher for exams with good to excellent bowel preparation (8.2%, 95%CI 3.3%–13.0%), though the difference vs. all exams was not significant (P=.10) (Figure 4).

Prevalence of sessile serrated polyps by age and sex

Few U.S and European studies reported on the prevalence of SSPs by age and sex. The age pattern for SSPs contrasted with that for adenomatous polyps (Figure 5). SSP prevalence increased with age, but the relative increases were smaller than that for adenoma, and did not reach statistical significance (relatively, +1.9% per age year, 95%CI %–0.1%–3.9% for two studies providing a denominator [P=.06], vs. +2.7%, 95%CI 2.0%–3.3% for adenoma [P<.001]). Across studies, the relative age-increases declined over time (Suppl.Figure 7, and Suppl.Tables 4–5).

Figure 5.

Prevalence of sessile serrated vs. adenomatous polyps by age.

Dots=study observations by age; colors=different U.S. and European studies (Suppl.Table 4–5); size=patient denominator; dashed line=fitted log-linear trends for ages 40–100 years measuring relative increase by age-year; ***=significant trend, P<.001. Some studies in A provided no patient denominator, and were excluded from trends and represented with small constant N.

The ratio for the reported prevalence of SSPs in men vs. women of 1.20 (4.9%, 95%CI 2.8%–7.0% for men vs. 4.1%, 95%CI 2.7%–5.5% for women; P=.54) (Suppl.Figures 8–9) was comparable to the ratio of 1.16 for adenoma (37.6%, 95%CI 29.7%–45.4% for men, vs. 32.4%, 95%CI 25.2%–39.6% for women; P=.35) (Suppl.Table 6).

Number, size and location of sessile serrated polyps

The number, size, and localization of sessile serrated vs. adenomatous polyps in U.S. and European studies is presented in Figure 6.

Figure 6.

Number, size, and localization of sessile serrated vs. adenomatous polyps.

U.S. and European studies. See Suppl.Figures 10–12 for Forest plots and Suppl.Tables 7–9 for adenoma data. Whiskers=95% confidence intervals; **=significant differences for SSP vs. adenoma, P<.01 (***P<.001); Size=small (0–5mm), medium (6–9mm), or large (10+mm); location=proximal (cecum, ascending colon, hepatic flexure, transverse colon), distal (splenic flexure, descending colon, sigmoid colon), or rectal.

Compared with adenomatous polyps, SSPs were characterized by a relatively lower multiplicity, with 69.0% (95%CI 45.9–92.1%; P=.08) (Suppl.Figure 10) of SSP patients vs. 43.3% (95%CI 38.0%–48.6%) (Suppl.Table 7) of adenoma patients having only one polyp, and 14.3% (95%CI 0–31.3%; P=.07) vs. 34.0% (95%CI 28.0%–40.0%) having 3 or more, respectively.

SSPs were somewhat larger on average than adenomas, with 19.3% (95%CI 12.4%–26.2%; P=.13) (Suppl.Figure 11) vs. 13.2% (95%CI 10.4%–16.1%) (Suppl.Table 8) having a diameter of more than 10mm.

Compared with adenoma, more SSPs were located proximal to the splenic flexure: 71.5% (95%CI 63.5%–79.5%; P=.009) (Suppl.Figure 12), vs. 59.3% (95%CI 54.2%–64.3%) (Suppl.Table 9). The proportion of polyps located in the rectum was similar (P=.75) and less than 10% for both types of polyps.

Sessile serrated polyps and synchronous adenoma

Among two studies reporting coexistence of adenomas and SSPs, 49.4% (95%CI 45.4%–53.9%) of SSP patients had adenomas (Suppl.Figure 13), notably more than the average prevalence of adenoma in autopsy studies (Suppl.Figure 8).

Presence and age at detection of cytological dysplasia in sessile serrated polyps

The proportion of SSPs with reported cytological dysplasia ranged from 3.7 to 42.9% across four U.S. and four European studies (Figure 7), with an overall mean of 13.9% (95%CI 4.8%–23.0%). Although both U.S and European studies consisted of a mix of studies including exams for screening or other indications (Suppl.Table 2), the proportion of SSPs with cytological dysplasia differed between U.S. and European studies (7.8%, 95%CI, 2.9%–12.6% vs. 20.6%, 95%CI 3.6%–37.5%; P=.15) (Suppl.Figure 3), possibly due to relatively more nonscreening exams in Europe (Suppl.Table 10). Excluding one extreme statistical outlier,³⁷ the proportion of SSPs with dysplasia decreased to 9.7% (95%CI, 4.5%–14.9%).

Figure 7.

Proportion of sessile serrated polyps with cytological dysplasia across studies.*

U.S. and European studies, including nonscreening indications. Squares=means; square size=population size; whiskers=95% CIs; polygon=pooled mean, random-effects meta-regression; ***=significant heterogeneity, P<.001.U.S. 7.8% (95%CI, 2.9%–12.6%) vs. European studies 20.6% (95%CI 3.6%–37.5%) (P=.15). For some studies, the relative prevalence of dysplastic vs. all SSPs was used as approximation (Suppl.Table 10). Excluding one outlier,³⁷ the proportion decreased to 9.7% (95%CI, 4.5%–14.9%; Europe 13.1% (95%CI, 0.6%–25.5%).

SSPs with dysplasia tended to be reported at higher ages than SSPs without dysplasia (Suppl.Figure 14, Suppl.Table 11). Among four U.S. and European studies reporting age at diagnosis, SSPs without dysplasia were detected at an average age of 60.8 years, SSPs with low-grade or any dysplasia at an average age of 65.6 years, and SSPs with high-grade dysplasia at an average age of 70.2 years.

DISCUSSION

We performed a systematic review including meta-analysis of studies published since 2000 on the prevalence and clinical features of sessile serrated polyps (SSPs) to provide perspective into the potential importance of these lesions. The overall prevalence of SSPs in colonoscopy studies varied around the world. In studies from the U.S. and Europe, based on strict histological definitions, the prevalence was 4.6% (95%CI 3.4%–5.8%), and SSPs accounted for 9.4% (95% CI 6.6%–12.3%) of potentially precancerous lesions. Prevalence estimates were as high as 9.1% when based on high-quality colonoscopy examinations, and 12.3% when also counting potentially clinically-relevant hyperplastic polyps. Compared with adenoma, SSPs more often presented solitary, with large size and in the proximal colon. SSPs with dysplasia and high-grade dysplasia were reported at approximately 5 and 10 years older average ages, respectively, than non-dysplastic SSPs.

There is no consensus in the literature on which serrated polyps are clinically relevant, and this has consequences for the estimates of prevalence and risk. Histologically, SSPs and hyperplastic polyps may be difficult to distinguish,^{49, 53} creating the potential for misclassification under strict histological definitions that consider SSPs and traditional serrated adenoma as the only clinically relevant serrated polyps.⁸ More liberal definitions have been proposed to address this, but studies differ on whether these should include large,^{33, 57, 89} proximal,^{59, 81} or both large and proximal hyperplastic polyps.^{32, 78, 97} We report separate results based on older terminology, strict definitions, and more liberal definitions to shed light on what can be a confusing subject.

We found no significant association between SSP detection and calendar year when all studies were pooled. However, lower SSP rates in Asia, including in large recent studies, affected the pooled estimates. Among U.S. studies, there was a significant upward trend (+0.3%-point/year), consistent with reports from specific settings.^{53, 65} Differences in prevalence between Asia and the U.S. may be due to practice variation, or more fundamental causes. The lower reported SSP rates in Asia are consistent with CRC incidence data (Suppl.Figure 3B) and race-specific data from the U.S.⁶⁵ Further research on these differences could help elucidate SSP etiology.

SSP prevalence increased with age, but the increases did not quite meet the traditional criteria for statistical significance, unlike increases for adenoma. The relationship between SSP prevalence and age may have evolved over time by period or cohort effects (Suppl.Figure 7), such as more prior colonoscopies with polypectomy in older patients, or an increased risk of SSPs in younger adults. An increased risk in younger persons would synchronize with documented young-onset CRC increases,¹⁰⁹ although those increases were mostly confined to the rectum, where SSPs are uncommon.

Prevalence was not associated with symptomatic indications. This is consistent with previous studies,^{65, 108} and suggests SSPs may cause fewer symptoms than adenoma.

There was substantial heterogeneity in SSP prevalence across studies, which may be partly due to study biases. Studies included in this review were judged to have moderate to high risk of bias due to patient selection, detection bias, and possible misclassification. No studies ascertained unselected samples from the general population. In the U.S., undergoing a colonoscopy is associated with socioeconomic indicators, such as health literacy and insurance status.¹¹¹ Patients with a previous screening history were usually included, which could be associated with lower SSP prevalence either due to polypectomy earlier in life or selection of healthy individuals if patients with a polypectomy earlier in life were excluded (Figure 5). Although colonoscopy is the gold standard for colon examination, it may have missed a substantial proportion of serrated polyps (Figure 4),¹¹² particularly in earlier years (Suppl.Figure 2). Thus, the true prevalence of serrated polyps should be considered higher than what is reported in colonoscopy studies, since it must be corrected by the miss rate. Finally, it may be difficult for pathologists to distinguish sessile serrated from hyperplastic polyps. Both the endoscopic detection and pathology assessment of SSPs require training and verification of quality to ensure adequate diagnosis.

Dysplasia occurred in 3.7%–42.9% of SSPs across studies in our review. This wide range must be interpreted with caution (Suppl.Table 10). It may reflect small sample sizes, and population differences in risk behavior (e.g. smoking) or presence of symptoms signaling dysplastic but not non-dysplastic SSPs. Further research should determine the true prevalence of dysplastic SSPs in unselected populations and the effect of risk factors including smoking.

Strengths of this study include the systematic character, the sensitive search strategy, and the broad scope encompassing many relevant questions regarding SSP epidemiology. The study also has limitations. First, we did not attempt to reclassify hyperplastic polyps from the era before awareness of the SSP.¹¹³ While this might reduce the risk of detection bias and contamination, it is not feasible to perform histopathology reassessment of all specimens in these older studies. Few of the reviewed autopsy studies provided detailed information on hyperplastic polyps. Second, although our review addressed several questions related to SSP risk, e.g. regarding size and presence of dysplasia, we could not directly assess risk of malignancy. Our literature search suggests that very few studies with longitudinal patient follow-up exist. A large trial following patients after SSP removal is recruiting, but results are not expected for >10 years,¹¹⁴ and there is no plan to observe SSPs unresected. Observational data suggest no higher post-polypectomy risk for SSPs vs. adenomas, while features like large size and dysplasia, which we estimated to each occur in up to 10–15% of SSPs, may increase risk.^115–117

In conclusion, although SSP prevalence may be underestimated by colonoscopy studies with inherent miss rates, SSPs appear to be relatively uncommon compared with adenomas. More standardized diagnostic criteria for clinically relevant serrated polyps, training, and quality verification for endoscopists and pathologists are needed to ensure SSP detection and removal. Additional research is needed to increase clarity on actual prevalence by age, variation across settings, and long-term risk with and without removal.

LAY SUMMARY

In a review of research articles, we found the reported prevalence of sessile serrated polyps to vary based on world region, diagnostic criteria, and examination quality, but there was not much difference among age groups or between sexes.

WHAT YOU NEED TO KNOW

Background & Context:

Sessile serrated polyps (SSPs) might account for 30% of colorectal cancers, but there is uncertainty about SSP prevalence and clinical features.

New findings:

In a systematic review, we found the reported prevalence of SSPs to be less than 5%, on average, with limited variation by age and sex. However, prevalence varies with study location, diagnostic criteria, and examination quality. SSPs are often solitary, large, and proximal; the proportion that have dysplasia, based on cytologic analysis, varied widely across studies, possibly reflecting population differences.

Limitations:

Cancer risk could not be assessed.

Impact:

Standardized diagnostic criteria, training, and quality verification for endoscopists and pathologists should be considered to ensure adequate SSP detection, diagnosis and removal. Additional research is needed to determine differences in prevalence with age and among different locations, and long-term risk.

ABBREVIATIONS:

CRC

colorectal cancer

PICOS

population, intervention, comparison, outcome, study design

SSP

sessile serrated polyp

WHO

World Health Organization