The advanced adenoma

The currently accepted evolution of colon cancer is via the “adenoma-carcinoma sequence,” a term first coined by Jackman and Mayo in 1951, and further developed and refined by pathologists including Morson, Muto, and Bussey.

The model for progression from normal epithelium to adenoma to carcinoma is a series of genetic mutations. Colonic neoplasms are thought to arise as a result of mutational activation of oncogenes (ras gene on chromosome 12p) coupled with the mutational inactivation or loss of tumor suppression genes (familial adenomatous polyposis gene on chromosome 5q, p53 gene on chromosome 17p, and DCC gene on chromosome 18q). These mutations act at several steps in the progression from normal epithelium to hyperproliferative epithelium to early, intermediate, and late adenoma, and finally to carcinoma.

The degree of dysplasia is highly correlated with risk of malignancy because on a genetic and cellular level, increasing atypia leads to a stepwise progression toward carcinoma. The presence of high-grade dysplasia is therefore a strong predictor of which adenomas will go on to become carcinomas.

There is a direct, though nonlinear, relationship between the size of an adenoma and its likelihood to harbor malignancy. In a colonoscopy series of a nonscreening population, the prevalence of carcinoma in adenomas greater than 2 cm in size was 19.4%, whereas that in adenomas of less than 1 cm was 0.07%. These data reflect an upper bound for the prevalence of carcinoma, as the prevalence of carcinoma and dysplasia is observed to be lower across all size ranges in an asymptomatic screening population compared to a nonscreening cohort. Other OC-based clinical series have confirmed the relationship between size and grade of dysplasia. Accordingly, it is well established that adenoma size correlates with the degree of cellular dysplasia. Some investigators have also proposed a direct relationship between increasing villous component seen on histology and the degree of dysplasia. At present, the presence of high-grade dysplasia, adenoma size, and villous characteristics are the primary characteristics used to risk-stratify a polyp with respect to its propensity to harbor or develop malignancy.

Screening studies have targeted the removal of adenomas that have the highest potential of developing into colorectal carcinoma. These “advanced adenomas” are typically defined by the presence of any of the following 3 criteria: high-grade dysplasia, size 10 mm or larger, or a substantial (>25%) villous component (ie, tubulovillous or villous adenomas). Lesions demonstrating these characteristics are thought to be at high risk of developing into colorectal carcinomas compared with their less advanced counterparts. In an asymptomatic screening population, the overall prevalence of advanced adenoma or carcinoma (collectively termed advanced neoplasia) of any size is 3% to 4% ; the prevalence of subcentimeter advanced neoplasia is only 0.3%.

The diminutive polyp

In CTC, polyps are stratified by size as less than 6 mm (diminutive), 6 to 9 mm (small, or sometimes referred to as ‘intermediate’), and equal to or greater than 10 mm (large). Because polyp histology cannot be reliably characterized in vivo, size remains the most useful characteristic to assess the carcinomatous potential of a polyp before its resection. Thus, screening guidelines are developed based on size estimates and neoplasia prevalence estimates obtained from large clinical series. Several studies have established that approximately 50% of diminutive polyps are adenomas while the other 50% are nonneoplastic. The prevalence of advanced histology in diminutive polyps is quite low and the accurance of their size measurement is limited, as discussed next.

The diminutive polyp

In CTC, polyps are stratified by size as less than 6 mm (diminutive), 6 to 9 mm (small, or sometimes referred to as ‘intermediate’), and equal to or greater than 10 mm (large). Because polyp histology cannot be reliably characterized in vivo, size remains the most useful characteristic to assess the carcinomatous potential of a polyp before its resection. Thus, screening guidelines are developed based on size estimates and neoplasia prevalence estimates obtained from large clinical series. Several studies have established that approximately 50% of diminutive polyps are adenomas while the other 50% are nonneoplastic. The prevalence of advanced histology in diminutive polyps is quite low and the accurance of their size measurement is limited, as discussed next.

Measurement of polyps by OC and CTC

It is worth mentioning the nature of OC and CTC measurement because prior studies providing the framework for our current thinking on polyp size and risk of malignancy rely on these size measurements, in particular OC measurements, as the reference standard.

The use of any formal measurement device is variable in both clinical and research OC practice; no single measurement method has been consistently employed to estimate size of polyps observed at OC. Measurement methods employed in research trials of colonoscopy include visual estimation, linear probe, and open biopsy forceps; the latter, for example, having been employed in the National Polyp Study. Considerable variability has been reported concerning the accuracy of OC measurements, including both under- and overestimation of polyp size. Compared with visual estimation and linear probe, Gopalswamy and colleagues demonstrated that open biopsy forceps was the least accurate polyp measurement tool, with a 12.3% mean difference from the immediate post-resection (nonprocessed specimen) reference polyp size. This gap was further accentuated in polyps less than 6 mm in size, which demonstrated a 27.9% difference from the reference polyp size.

In CTC, a polyp is measured along its single largest diameter, excluding the stalk, using a software-based caliper tool deployed by all vendors of CTC interpretation software. The software-based caliper tool can be used on either the 2-dimensional (2D) multiplanar displays or the 3-dimensional (3D) (endoluminal) displays now universally available for CTC interpretation software. CTC measurements are performed within the context of a calibrated reference system established by the computed tomography (CT) gantry and scan parameters. When performed in an axial plane, these measurements are precise to approximately 0.7 mm (based on the standard parameters encountered for CTC imaging: an axial 512 × 512 pixel reconstruction matrix in a 35-cm field of view). Increasing use of isotropic voxel imaging, a technique available in current generation scanners, permits preservation of similar high spatial resolution and precision for all CT measurements, including 3D-rendered (endoluminal) and nonaxial plane measurements.

These technical distinctions are further borne out in recent data comparing size measurements of polyps by OC and CTC; investigators observed that mean error for 2D-CTC polyp measurements was significantly lower than that for OC. In sum, CTC provides the ability to make extremely precise size measurements of colonic lesions; however, the accepted prevalence estimates for polyp dysplasia are largely based on optical colonoscopy, which as usually practiced demonstrates relatively poor size measurement accuracy for polyps in the diminutive (<6 mm) size range.

Prevalence of advanced adenoma and carcinoma in diminutive polyps

In current guidelines for colon screening and surveillance, only adenomatous polyps are of concern with respect to colorectal cancer risk; hyperplastic polyps, hamartomatous/juvenile polyps, mucosal tags, and lipomas confer no increased cancer risk.

However, it should be noted that in addition to the traditional adenoma-carcinoma sequence, the microsatellite-instability pathway is postulated as an alternate pathway of carcinogenesis in which a hyperplastic polyp may progress to a sessile serrated polyp, then to a dysplastic serrated polyp (also known as a serrated adenoma), and finally to colorectal carcinoma. Any malignant risk consequent to this transformation is still associated with the development of an adenomatous precursor lesion; hence, it remains valid to say that hyperplastic lesions themselves are not the target of screening. The transformation and growth rate of serrated adenomas is less well defined in comparison to corresponding data for other adenomatous lesions. However, observed pathologic data suggest that small and diminutive sessile serrated adenomas demonstrate histologic abnormalities at rates comparable to conventional small and diminutive adenomas and hence confer similar clinical risk. In Church’s study of small polyps, he assigned serrated adenomas to the same category as tubular adenomas, assuming a similar risk for both lesions.

Correctly estimating the prevalence of different polyp histologies is complicated by the varied ways in which prevalence data has been reported. Literature concerning colon polyps has described the prevalence of advanced neoplasia in both screening and nonscreening cohorts. Prevalence data have further variously been reported as a fraction of all detected polyps versus as a fraction of adenomas only. As mentioned previously, the prevalence of advanced neoplasia is higher in nonscreening, symptomatic cohorts compared with a screening, asymptomatic population (the latter being the target population for this discussion). The prevalence of advanced features would also be higher when reported as a percentage of only adenomas rather than of all detected polyps. Because neither the radiologist nor the gastroenterologist can prospectively determine which polyps are adenomas, it makes the most practical sense to evaluate the prevalence of advanced neoplasia as a fraction of all polyps for the purposes of determining a reporting threshold in a screening population.

In nonscreening cohorts, the rate of advanced neoplasia has been reported in the National Polyp Study as approximately 2.0% among resected diminutive adenomas, and in a study by Church as 2.1% among resected diminutive polyps. In a surgical series of colonic adenoma specimens, the prevalence of cancer in diminutive adenomas was reported as 1.3%. These figures certainly overestimate the risk when considering a nonscreening population and when considering all polyps.

In contrast, Odom et al. found only 1 case of carcinoma among 2851 resected diminutive adenomas (0.04%) in a retrospective analysis of a nonscreening population.

In a recent OC study of a screening cohort, Lieberman et al. found 63 advanced adenomas out of 3744 diminutive polyps (1.7%). In this study, serrated adenomas were also included as advanced lesions; however, the pathology literature suggests that the risk of a serrated adenoma is similar to that of a tubular adenoma. If serrated adenomas are not considered advanced adenomas, as the literature suggests, then the prevalence of advanced adenomas in the study would further decrease to 1.2%. There was 1 case of carcinoma among the 3744 diminutive polyps (0.03%).

The results reported by Kim et al. in a comparison of parallel, concurrent screening programs using CTC and OC further support the conclusion that the prevalence of advanced neoplasia in diminutive polyps is negligible. These investigators prospectively evaluated the experience of 3120 CTC and 3163 OC asymptomatic screening subjects and observed that the prevalence of advanced neoplasia was similar in both arms, even though the CTC program did not report diminutive polyps of less than 6 mm. Of 2995 polyps removed, 2006 were diminutive, and of these only 4 were advanced adenomas (0.2% of all diminutive polyps). None of the polyps less than 6 mm harbored carcinoma. The 18 cases of cancer reported in the series were all in polyps 10 mm or larger. The rate of advanced adenoma for polyps 6 to 9 mm was 2.0% and for polyps 10 mm or larger was 8.6%. Combined, these data reinforce the conclusion, as has been incorporated into clinical guidelines, that the prevalence of advanced neoplasia in diminutive polyps encountered for asymptomatic colon screening is less than 1%, and that asymptomatic screening should target the identification and possible removal of advanced adenomas.

Contribution of villous histology to advanced adenomas

The National Polyp Study demonstrated that adenoma size and percentage of villous component were both correlated with the presence of high-grade dysplasia. The investigators stratified diminutive adenomas (<6 mm in size) into 4 histologic subtypes: 0% villous component (tubular adenomas), 1% to 25% villous component, 26% to 75% villous component, and greater than 76% villous component. This study showed a progressively increasing percentage of high-grade dysplasia with each successive histologic category: 0.7% (n = 8) of diminutive purely tubular adenomas, 4.7% (n = 2) of diminutive adenomas with 1–25% villous component, 10.0% (n = 1) of diminutive adenomas with 26–75% villous component, and 20% (n = 1) of diminutive adenomas with greater than 75% villous component demonstrated high-grade dysplasia.

This observed association between increasing villous component and high-grade dysplasia is one reason that some have used villous component as an independent risk marker of developing carcinoma. However, it is important to keep in perspective the extremely low prevalence both of high-grade dysplasia and of substantial villous histology in diminutive polyps when making this association. In the National Polyp Study, only 12 of 1270 diminutive adenomas (0.9%) demonstrated high-grade dysplasia, and only approximately 15 of 1270 demonstrated a greater than 25% villous component (1.2%). Only 2 diminutive adenomas demonstrated both high-grade dysplasia and substantial villous histology (0.2%). As a fraction of all polyps encountered , these prevalence data would be even lower.

The traditional definition of advanced adenomas has included all polyps with 25% or greater villous component, and it is not clear from the literature why this categorization is so inclusive when the National Polyp Study data demonstrate that adenomas with 25% villous component and adenomas with 75% villous component demonstrate different prevalences of high-grade dysplasia. These data have led to recent debate in the pathology literature suggesting that the cancer risk ascribed to a polyp based on the presence of villous features is significant only when villous features comprise nearly all of the adenoma.

Further obscuring the assessment of clinical significance of villous features in diminutive lesions is the considerable interobserver variability among pathologists in determining both the presence and percentage of villous features within a polyp. Using the National Polyp Study data, the authors analyzed interobserver variability using kappa statistics among the central study pathologists and the pathologists at the respective institutions where the polyps were originally obtained. All the pathologists used predefined World Health Organization criteria for villous features and high-grade dysplasia. There was a correlation of greater than 90% among pathologists with respect to determination of high-grade dysplasia; however, there was substantially less agreement with respect to the determination of the presence or absence of villous component, despite the presence of preestablished criteria (O’Brien MJ [Department of Pathology, Boston Medical Center, Boston, MA], personal communication, October 8, 2009).

A similar comparison was undertaken in the Multicenter Study on Colorectal Adenomas, in which 4 gastrointestinal pathologists reviewed and classified 100 polyps. The histologic classification of adenomas was significantly different among all 4 pathologists, with median kappa values of 0.50, 0.15, and 0.36 for the diagnosis of tubular, tubulovillous, and villous adenomas, respectively. Along similar lines, in a Danish study comparing the interobserver variability among 3 pathologists in their evaluation of adenomas, the investigators observed only a 61% agreement among the 3 pathologists in characterizing adenoma histology in terms of villous component. Finally, a study comparing the accuracy of community pathologists to a reference-standard consensus demonstrated that tubular adenoma was incorrectly upgraded to tubulovillous or villous adenoma in 35% of readings.

In summary, there is a lack of concordance in interpreting the presence and degree of villous change and few data establishing that tubulovillous histology in a diminutive adenoma independently confers a clinically significant risk. It may be prudent to reconsider whether diminutive lesions with tubulovillous features and no observed high-grade dysplasia should be included as advanced adenomas.

The accurate clinical categorization of these lesions is important because it underlies the assumed clinical impetus to report or resect diminutive lesions, the vast majority of which demonstrate benign histology by any criteria. If only those lesions that contain high-grade dysplasia or carcinoma, regardless of villous histology, are included in the designation of an advanced lesion, then the percentage of advanced adenomas among diminutive lesions would be reduced by nearly half, further challenging the prudence of reporting diminutive lesions. For example, in Church’s nonscreening OC study, inclusion of only those lesions with high-grade dysplasia or carcinoma would decrease the prevalence of advanced adenomas from 2.1% to 0.9% of all diminutive polyps, and in the National Polyp Study from 2.0% to 0.9% of all diminutive adenomas. When considering the more applicable screening study by Lieberman et al. the prevalence of advanced adenoma would decrease from 1.2% to 0.05% (2 cases of high-grade dysplasia or carcinoma among 3744 diminutive polyps).

In the authors’ view, the combination of: (1) low prevalence of high-grade dysplasia and substantial villous component in diminutive polyps; (2) poor measurement characteristics of diminutive polyps; (3) low interobserver agreement for the presence and percentage of villous histology; and (4) limited outcome data highlighting the clinical risk of diminutive tubulovillous polyps without high-grade dysplasia all suggest that the classification of tiny tubulovillous polyps as “high-risk” based on the presence of villous component alone represents an overestimation of risk.