Growth Rates and Histopathological Outcomes of Small (6-9 mm) Colorectal Polyps Based on CT Colonography Surveillance and Endoscopic Removal

Abstract

Background and Aims:

The natural history of small polyps is not well established and rests on limited evidence from barium enema studies decades ago. Patients with one or two small polyps (6-9 mm) at screening CTC are offered CTC surveillance at 3 years but may elect immediate colonoscopy. This practice allows direct observation of the growth of sub-centimeter polyps, with histopathological correlation in patients undergoing subsequent polypectomy.

Design:

Of 11,165 asymptomatic patients screened by CTC, 1,067 had one or two 6-9 mm polyps detected (with no polyps ≥10 mm). Of these, 314 (mean age, 57.4 years; M:F, 141:173; 375 total polyps) elected immediate colonoscopic polypectomy, and 382 (mean age 57.0 years; M:F, 217:165; 481 total polyps) for CTC surveillance over a mean of 4.7 years. Volumetric polyp growth was analyzed, with histopathological correlation for resected polyps. Polyp growth and regression were defined as volume change of ±20% per year, with rapid growth defined as +100% per year (annual volume doubling). Regression analysis was performed to evaluate predictors of advanced histopathology, defined as the presence of cancer, high-grade dysplasia (HGD), or villous components.

Results:

Of the 314 patients that underwent immediate polypectomy, 67.8% (213/314) harbored adenomas, 2.2% (7/314) with advanced histology; no polyps contained cancer or HGD. Of 382 patients that underwent CTC surveillance, 24.9% (95/382) had polyps that grew, while 62.0% (237/382) remained stable and 13.1% (50/382) regressed in size. Of the 58.6% (224/382) CTC surveillance patients that ultimately underwent colonoscopic resection, 87.1% (195/224) harbored adenomas, 12.9% (29/224) with advanced histology. Of CTC surveillance patients with growing polyps who underwent resection, 23.2% (19/82) harbored advanced histology versus 7.0% (10/142) with stable or regressing polyps (OR: 4.0; p<0.001), with even greater risk of advanced histology in those with rapid growth (63.6%, 14/22, OR: 25.4; p<0.001). Polyp growth, but not patient age/sex or polyp morphology/location were significant predictors of advanced histopathology.

Conclusion:

Small 6-9 mm polyps present overall low risk to patients, with polyp growth strongly associated with higher risk lesions. Most patients (75%) with small 6-9 mm polyps will see polyp stability or regression, with advanced histology seen in only 7%. The minority (25%) with small polyps that do grow have a 3-fold increased risk of advanced histology.

Introduction

Despite advances in screening and treatment, colorectal cancer (CRC) remains the second-leading overall cause of cancer mortality in the United States, resulting in over 50,000 deaths per year.[1] CRC begins as a benign precursor polyp and requires successive mutations or epigenetic events—typically taking many years--to become malignant.[2] This lead-time to malignancy, coupled with the preventive nature of polyp removal, is what has allowed screening to be effective in reducing overall CRC mortality over the last three decades.[3, 4]

Colorectal adenomas are regarded as precursor lesions to colorectal cancer. Distinction between low- and high-risk (advanced) adenomas is mainly based on polyp linear diameter (with an advanced adenoma threshold of 10 mm), but histopathological parameters are also correlated. Database studies have shown that sub-centimeter adenomas are nearly always benign, and patients harboring these polyps do not have a worse prognosis compared with larger/advanced adenomas.[5, 6] Conversely, adenoma detection rate has been shown to correlate with interval cancer,[7–9] leading to the assumption that higher adenoma detection rate is beneficial for CRC prevention, although adenomas are sub-centimeter (non-advanced) in more than 90% of cases. Consequently, there is still considerable debate about the relevance of small adenomas and the necessity surveillance examinations, including follow-up intervals.

While colonoscopy remains the primary modality for CRC screening, CT colonography (CTC) has emerged as an alternative method for CRC screening over the last 20 years.[10, 11] Previous studies of in vivo surveillance with conventional endoscopy and barium enema provide only limited direct observations of polyp natural history.[12–19] Although CTC is not currently utilized as a high-volume CRC screening tool in clinical practice, serial CTC is ideal for longitudinal in vivo investigation of unresected sub-centimeter polyps[20–22] due largely to its non-invasive nature, and may provide some additional insight into polyp natural history through direct observation.

Our purpose in this 16-year prospective study is to report the growth patterns of small colorectal polyps identified and followed longitudinally with CTC, using novel volumetric assessment of individual polyps—many observed over multiple time points—with histopathological context and in comparison with polyps which proceed to resection without surveillance. For additional context, we report histopathological findings in a group of patients which were negative at initial screening CTC but found to develop de novo polyps on subsequent routine screening CTC.

Study Design

This retrospective longitudinal cohort study was conducted at a single academic institution. The study protocol was HIPAA-compliant. Ethics and Institutional Review Board (IRB) approval was obtained (University of Wisconsin School of Medicine and Public Health, IRB #2004-0104, #2014-1617, #2016-0418). An initial patient cohort provided written informed consent and established the low-risk nature of in vivo surveillance of small colorectal polyps.[20] The need for signed informed consent was waived for subsequent patients undergoing this clinical innovation pathway. This manuscript was completed following STROBE guidelines.[23]

Polyp Definitions, Terminology, and Screening Protocol

Colorectal polyp linear diameter is measured to the nearest millimeter (mm) and volume to the nearest cubic millimeter (mm³) at CTC, as previously described.[20, 22] Polyps ≥10 mm in diameter are defined as “large” and polyps 6-9 mm are defined as “small”. Polyps ≤5 mm are considered “diminutive” and are not routinely reported at CTC. A screening CTC where at least one large or small polyp is detected is “positive”, while “negative” screening CTC detects no large or small polyps, although diminutive lesions may be present. Patients with a negative CTC are offered 5-year follow-up screening CTC. Patients with a positive screening CTC are offered colonoscopy with polypectomy; if only one or two small polyps are detected an additional option of 3-year polyp surveillance CTC is offered.[24] This in vivo CTC surveillance was initially performed as part of a prospective trial,[20] but later transitioned to routine clinical practice. Initially, a more conservative 1-2 year surveillance interval was followed, but was expanded to three years after safety was demonstrated.[20] Polyps are defined as histologically “advanced” if they contain villous components (e.g., tubulovillous adenoma), high-grade dysplasia (HGD), or adenocarcinoma. Polyps are defined as “resolved” if they become clinically undetectable at a follow-up examination. Colonoscopy with polypectomy was advised as mandatory for any polyp ≥10 mm in diameter at any time. Polyp “growth” was defined as a volume increase of ≥20% per year,[20] with polyp “regression” defined as a volume decrease of ≥20% per year. Polyps not meeting either of these thresholds were defined as “stable”.

Study Groups

From April 1, 2004 through August 31, 2020 (197 months or 16.4 years), 11,165 total asymptomatic adult patients (mean age [±SD], 57.5 ± 8.3 years, M:F, 5,019:6,146) underwent clinical screening CTC at a single academic center, with screening for CRC/colorectal polyps as the sole exam indication. Of these, 1,977/11,165 (17.7%) were initially positive and 9,188/11,165 (82.3%) were initially negative. Of the 1,977 positive patients, 1,067/1,977 (54.0%) were eligible for CTC surveillance (no polyps ≥10 mm, 1-2 polyps 6-9 mm).[20, 24]

Of those eligible for CTC surveillance, 382/1,067 (35.8%) patients (mean age [±SD], 57.3 ± 7.1 years, M:F, 217:165) elected and underwent at least one additional CTC surveillance examination. There were 481 unique polyps identified among these 382 patients, which constitute the CTC surveillance group. The mean [±SD] surveillance interval was 4.7 ± 2.9 years per patient (range, 0.7-13.8 years), providing 1781.7 total patient-years of in vivo surveillance.

A further 314/1,067 (29.4%) patients (mean age [±SD], 57.4 ± 8.2 years, M:F, 141:173) elected and underwent same-day or short interval polypectomy, defined as polypectomy within one year of CTC. There were 375 unique polyps identified among these 314 patients, which constitute the immediate polypectomy group. There was no significant difference in age (p=0.863) between the CTC surveillance and immediate polypectomy groups, although there were significantly more men in the CTC surveillance group (58.2%) compared with the immediate polypectomy group (44.9%, p=0.002). The remaining 371/1,067 (34.8%) patients deferred follow-up or have follow-up pending as of the conclusion of the study. A complete flow diagram of the study cohort is provided in Figure 1.

Figure 1:

Flow diagram of study cohort.

Of patients who were negative (no polyps ≥6 mm) at initial CTC screening, 2,032/9,188 (22.1%) returned for routine 5-year follow-up CTC screening within the study period. Of these, 205/2,032 (10.1%) patients (mean [±SD] age at follow-up CTC, 61.9 ± 6.4 years, M:F, 108:97) were positive with at least one de novo polyp ≥6 mm. There were 252 unique de novo polyps among these 205 patients, which constitute the positive at follow-up screening group. The mean [±SD] interval to follow-up screening CTC among these patients was 6.5 ± 2.4 years (median, 5.6 years, range, 2.6-15.7 years). As would be expected, this positive after initial negative CTC group was significantly older than both the CTC surveillance and immediate polypectomy groups (p<0.001) but did not differ significantly by sex (p=.0340 and p=0.088, respectively).

Data Collection

The CTC technique used in our screening program has been previously described in detail.[25] Once acquired, CTC examinations were interpreted using dedicated standalone software with primary 3D polyp detection and 2D confirmation.[26] Polyps were prospectively identified during clinical CTC screening by an experienced abdominal radiologist, and were recorded in a database. Polyp linear diameter, segmental location within the colon (cecum, ascending, transverse, descending, sigmoid, or rectum), and morphology (sessile, flat, pedunculated) were prospectively recorded. Of note, diminutive polyps are not routinely reported during clinical screening.

As measurement of polyp volume is not currently standard of care during clinical screening, polyp volumes were measured in dedicated separate sessions by expert readers for the purpose of volumetric growth analysis. Polyp volume was measured using a semi-automated segmentation tool, which automatically detects the margins of the polyp, but allows for manual adjustment to ensure accurate volume measurement; this technique has been previously used.[20, 22, 27] Reviewers measuring polyp volume were blinded to any available resection or histopathological data. Polyp volume measurements were performed only on polyps which could be definitively identified on at least two CTC exams. Polyps resected during the study period underwent histopathologic evaluation by gastrointestinal pathologists. Pathologists were not privy to polyp growth data at the time of histopathological evaluation.

Data Analysis

Study data was collected and collated in Microsoft Excel (Microsoft, Redmond, WA, USA). Single factor ANOVA and the t-test were used, where appropriate, to assess for differences in continuous variables, and chi-squared analysis was used to assess for differences in categorical variables. A threshold of p=0.05 was used for statistical significance. Changes in polyp size over time (linear diameter and volume) were calculated for all polyps with at least two time points. For polyps with at least three time points, separate interval growth rates were calculated, in addition to overall growth rates. Basic descriptive and inferential statistics (including the t-test, ANOVA, and chi-squared tests) were performed using SPSS Statistics (IBM, Armonk, NY, USA) Ordinal regression was used to assess whether patient demographic or CTC factors in isolation or collectively are predictive of final polyp histology and was performed using R. (see Statistical Supplement). Missing data was not possible as initial inclusion criteria effectively stipulated complete cases.

Receiver operating characteristic (ROC) curve analysis of annualized polyp growth as a classifier of advanced histology at different growth thresholds was performed for resected polyps with volumetric growth data. Area under the curve (AUC) was calculated using the trapezoidal approximation method.[28, 29] As prior research suggests that rapid polyp growth may be a marker of advanced histology,[20] Youden index (Youden J-statistic) analysis[30] of the ROC curve at volumetric growth rate intervals of 5% per year was performed to evaluate growth thresholds. Subsequently, confusion matrix (2x2 table) analysis was performed to calculate diagnostic performance (sensitivity, specificity, etc) and odds ratios for polyp growth as a predictor of advanced histology.

Results

Polyp Baseline Characteristics

Among the 481 polyps (in 382 patients) in the CTC surveillance group, mean [±SD] polyp linear diameter and volume at initial CTC were 7.1 ± 1.1 mm (median, 7 mm, range, 6-9 mm) and 61.8 ± 49.0 mm³ (range, 9-366 mm³), respectively. Among the 375 polyps (in 314 patients) in in the immediate polypectomy group, mean [±SD] polyp linear diameter at CTC was similar at 7.1 ± 1.0 mm (median, 7 mm, range, 6-9 mm, p=1.00). No significant difference in polyp segmental location (p=0.230) or initial morphology (p=0.194) was seen between the CTC surveillance and immediate polypectomy groups.

Among the 252 de novo polyps (in 205 patients) that developed in the positive at follow-up screening group, mean [±SD] polyp linear diameter was 8.5 ± 4.5 mm (median, 7 mm, range 6-40 mm), which was significantly larger than either the CTC surveillance or immediate polypectomy groups (p<0.001 for both).There was no significant difference between this group and the CTC surveillance group or the immediate polypectomy group for polyp segmental location (p=0.572 and p=0.058, respectively) or initial polyp morphology (p=0.109 and p=0.138, respectively). Polyp baseline characteristics, including segmental location and initial morphology, are summarized in Table 1.

Table 1:

Polyp Baseline Characteristics

Study Group	CTC Surveillance		Immediate Polypectomy		Positive at Follow-up Screen
Total Polyps (N)	481		375		252
Colorectal Segment Location	N	%	N	%	N	%
Cecum	52	10.8	38	10.1	25	9.9
Ascending	111	23.1	79	21.1	69	27.4
Transverse	87	18.1	52	13.9	45	17.9
Descending	38	7.9	24	6.4	20	7.9
Sigmoid	118	24.5	113	30.1	65	25.8
Rectum	75	15.6	69	18.4	28	11.1
Morphology at Initial CTC	N	%	N	%	N	%
Sessile	355	73.8	277	73.9	171	67.9
Flat	104	21.6	71	18.9	61	24.2
Pedunculated	22	4.6	27	7.2	20	7.9

Polyp Characteristics at Resection

Among the 375 polyps in 314 patients in the immediate polypectomy group, all (100%, 375/375) polyps underwent resection and histopathologic evaluation. The per polyp prevalence of adenomatous histology was 65.9% (247/375) with histologically advanced rate of 1.9% (7/375), corresponding to per patient rates of 67.8% (213/314) and 2.2% (7/314), respectively. No adenocarcinomas or polyps with HGD were found in the immediate polypectomy group. Polyp outcome characteristics for this group, including histopathology results, are summarized in Table 2.

Table 2:

Polyp Characteristics at Resection

Study Group	CTC Surveillance		Immediate Polypectomy		Positive at Follow-up Screen
Resected Polyps (N)	286		375		188
Morphology at Final CTC	N	%	N	%	N	%
Sessile	187	57.8	277	73.6	123	60.1
Flat	67	19.5	71	18.8	51	32.4
Pedunculated	18	4.8	27	7.6	14	7.4
Diminutive (<6 mm)	14	8.9	0	0	0	0
Histopathology	N	%	N	%	N	%
Adenocarcinoma	2	0.5	0	0	0	0
Tubulovillous adenoma (TVA)	28*	9.3	7	1.9	17	9.0
Tubular adenoma (TA)	160*	57.1	209	55.7	96	51.1
Sessile serrated polyp (SSP)	51	15.6	31	8.3	38	20.2
Traditional serrated adenoma (TSA)	2	0.5	0	0	0	0
Hyperplastic	38	11.7	61	16.3	21	11.2
Other benign non-neoplastic	5	2.4	60	16.0	16	8.5
Not retrieved	0	2.9	7	1.9	0	0

^*One tubulovillous adenoma and one tubular adenoma in the CTC surveillance group harbored high-grade dysplasia

Among the 481 polyps in 382 patients in the CTC surveillance group, 286/481 (59.5%) polyps in 224 patients underwent resection and histopathologic evaluation by the conclusion of the study period, while 152/481 (31.6%) remained detectable but unresected and 43/481 (8.9%) resolved (became undetectable at CTC). Polyp outcome characteristics for the 286 polyps that underwent resection, including final morphology and histopathology, are summarized in Table 2. Per polyp prevalence of adenomatous histology was 85.0% (243/286) with histologically advanced rate of 10.8% (31/286), corresponding to per patient rates of 87.1% (195/224) and 12.9% (29/224), respectively, all significantly greater than the immediate polypectomy cohort (P<0.001 for all). Two adenocarcinomas were found in the CTC surveillance group.

Among the 252 de novo polyps in 205 patients in the positive at follow-up screening group, 188/252 (74.6%) polyps in 154 patients underwent resection by the conclusion of the study period, with the remainder under continued CTC surveillance. Per polyp prevalence of adenomatous histology was 80.3% (151/188) with histologically advanced rate of 9.0% (17/188), corresponding to per patient rates of 81.2% (125/154) and 10.4% (16/154), respectively. These rates were significantly higher than the immediate polypectomy group (p<0.001 for all), but not significantly different than the CTC surveillance group in terms of per polyp or per patient adenoma rate (p=0.210 and p=0.146, respectively) or advanced adenoma rate (p=0.641 and p=0.519, respectively). No adenocarcinomas or polyps with HGD were found in the positive at follow-up screening group. Polyp outcome characteristics for this group, including histopathology results, are summarized in Table 2.

Polyp Growth

Among all 481 polyps in 382 patients in the CTC surveillance group, mean [±SD] annual volume change was +8.3 ± 57.6 mm³ or +17 ± 91% per year during the surveillance interval and mean annual change in polyp linear diameter during the surveillance interval was 0.0 ± 1.3 mm or 0 ± 18% per year. Among the 382 patients (481 total polyps) in the CTC surveillance group, polyp growth (volume increase ≥20% per year) was seen in 24.9% (95/382), polyp regression (volume decrease ≥20% per year) in 13.1% (50/382) and polyp stability in 62.0% (237/382). Polyps demonstrating rapid growth (annual volume increase of ≥100% per year, i.e., annual volume doubling) were seen in 5.2% (20/382). There was no significant difference seen in either patient age (p=0.678) or sex (p=0.264) among groups with growing, stable, or regressing polyps. Among growing polyps in the CTC surveillance group, 42.3% (44/104) attained linear diameter ≥10 mm at the time of resection, including both adenocarcinomas and polyps with HGD.

Among patients with polyps that grew (annual volume increase ≥20%) who underwent resection, 95.1% (78/82) harbored adenomas, 23.2% (19/82) with advanced histology, including two adenocarcinomas and two additional polyps with HGD (one tubular adenoma and one tubulovillous adenoma). Among patients with polyps that remained stable or regressed who underwent resection, there were significantly fewer adenomas (82.4%, 117/142, p=0.007) and polyps with advanced histology (7.0%, 10/142, p<0.001), all of which were tubulovillous adenomas and none with HGD.

Compared with stable or regressing polyps, the odds ratio that a patient with growing polyps harbors an adenoma is 4.2, 95% CI [1.4-12.4], with an odds ratio of harboring advanced histology of 4.0, 95% CI [1.8-9.1].

Of the resected polyps that underwent rapid growth (annual volume increase ≥100% per year), 64% (14/22) were histologically advanced, including both adenocarcinomas, both polyps with HGD (one tubulovillous adenoma and one tubular adenoma), and 7 additional tubulovillous adenomas. An example of rapid growth in an adenocarcinoma is shown in Figure 2. The remaining 36% (8/22) consisted of 7 tubular adenomas and 1 traditional serrated adenoma, none with HGD. Only 7% (17/264) of resected polyps with annual volume increase <100% per year were histologically advanced (all tubulovillous adenomas, none with HGD); the odds ratio that a polyp has advanced histology at this rapid growth threshold of annual volume doubling relative to a polyp below this growth threshold is 25.4, 95% CI [9.4-69.0].

Figure 2:

Initial screening CTC and 3-year surveillance CTC in 59-year-old man (at initial CTC). 2D and 3D images form CTC screening in 2013 show a sessile 6-mm sigmoid polyp. At 3-year CTC surveillance, the polyp demonstrated rapid growth and appeared more mass-like. The large polyp was malignant at histopathology (adenocarcinoma), one of only two cancers that developed during the study.

Receiver Operating Characteristic Curve Analysis

The ROC curve for mean annualized polyp volume change (% volume change per year) as a predictor of advanced histology at varying growth rate thresholds among the 286 resected polyps with volumetric growth data is shown in Figure 3, with AUC of 0.801. A maximum Youden Index value of 0.444 was obtained at a volumetric growth threshold of +20% per year. Applying this threshold to the sample of 286 resected polyps in the CTC surveillance group for prediction of advanced histology yielded sensitivity 74%, specificity 70%, positive predictive value 23%, negative predictive value 96%, and accuracy of 71%, with odds ratio of 6.8, 95% CI [2.9-15.8]. At ordinal regression, volumetric polyp growth was associated with advanced histology, but patient age, patient sex, polyp morphology, and polyp segmental location were not.

Figure 3:

Receiver operating characteristic (ROC) curve for annualized polyp volumetric growth rate as a classifier of advanced histology for resected polyps. Area under the curve (AUC) is 0.801, which is considered excellent.[29] Of note, the maximum Youden index (0.444) occurs at an annualized polyp volumetric growth rate of 20% per year (solid arrow), confirming this value as a reasonable choice for growth threshold. A near-maximum Youden index (0.441) also occurs near the growth rate of 100% per year (dashed arrow), representing annual volume doubling.

Discussion

Colorectal cancer is a prevalent and clinically important human disease, but largely preventable with screening detection and removal of relevant precursor polyps. Although CRC is well studied and clinical screening programs have operated for decades, data on colorectal polyps are largely restricted to cross-sectional studies, with only limited longitudinal data. This is due in part to the nature of colonoscopy and the potentially premalignant nature of colorectal polyps, leading to the long- established “universal polypectomy” standard of care. Although it is widely recognized that most diminutive and small colorectal polyps will never progress to cancer,[6, 31–34] there is little direct observational data on the growth patterns of these polyps.

Early investigations used modalities now considered imprecise and outdated, such as barium enema,[17, 19] or were invasive yet limited in scope, such as sigmoidoscopy.[12, 13, 16] The few longitudinal studies employing colonoscopy were principally focused on cancer outcomes in patients who have had prior polypectomy or CRC surgery (as opposed to a true screening population) and offer little specific information on individual polyp growth, morphology, or histology beyond cancer.[14, 15, 18, 35] Many of these early studies are now over 25 years old,[13–17, 19] only offer a maximum follow-up interval of 1-3 years,[12–16, 18, 19] or focused solely on large polyps.[17] Using CTC as a tool to monitor polyp growth, we are able to offer the largest longitudinal study of small (6-9 mm) colorectal polyps to date, including over 1,100 polyps (nearly 850 with histopathologic correlation) in over 800 patients during a study period spanning 16 years. We are further able to leverage the ability of CTC to take detailed size measurements, including volume measurements, at multiple time points. With this data, we can make several important observations about the growth and behavior of small colorectal polyps over time.

As suggested by prior research,[6, 20, 31, 33, 36, 37] we confirm that most small 6-9 mm polyps remain stable or regress in size, with only 25% of patients in the CTC surveillance group in this study experiencing polyp growth during an average surveillance period of 4.7 years. These patients accounted for an outsized share (37%) of CTC surveillance patients ultimately undergoing colonoscopy and polyp resection. Unsurprisingly, the per polyp and per patient adenoma detection rates were significantly higher in this group than patients who underwent immediate colonoscopy and polypectomy, likely due to a combination of selection of patients with growing polyps and a greater amount of time for polyps to continue growing. Further, our results also confirm that small polyps represent low risk to the patient.[6] In the immediate polypectomy group, the per patient rate of advanced histology for small 6-9 mm polyps was low at 2%, without any HGD or adenocarcinoma (all histologically advanced polyps were tubulovillous adenomas). Among CTC surveillance patients with stable or regressing polyps (polyps that remained small) that underwent resection, the advanced histology rate was only 7%—again without HGD or adenocarcinoma—despite an average of nearly five additional years in vivo. These results supporting overall low risk for small (6-9 mm) polyps presumably also extends the even smaller category of diminutive (>6 mm) polyps, which are not routinely reported at CTC and have been previously reported to be low risk.[6, 31, 33]

ROC curve analysis (Figure 2) illustrates the ability of polyp growth to predict advanced histology, with excellent AUC[29] (0.801) and Youden index supporting 20% volume increase per year as a threshold for polyp growth. Among patients with polyps that grew during CTC surveillance who ultimately underwent resection, 95% harbored adenomas with an advanced histology rate of 23%, including both adenocarcinomas and both polyps with HGD (one tubulovillous adenoma and one tubular adenoma) found in this study. The higher risk of clinically meaningful polyps in this group relative to surveillance patients with stable or regressing polyps is further supported by the odds ratios for presence of adenomas (OR 4.2) and advanced histology (OR 4.0) and suggests that polypectomy is warranted in patients with small but growing polyps. This is especially true for polyps that exhibit rapid growth (annual doubling of polyp volume), where the odds ratio of advanced histology is 25.4 relative to polyps that do not meet this threshold. Additionally, polyps which were found to arise de novo in patients with an initial negative screening CTC had adenoma and advanced histology rates that were significantly greater than those immediately resected after being found at initial screening CTC; as such, it may be reasonable to view these interval developing polyps with greater suspicion regardless of size.

These findings have very important potential clinical relevance for a screening program undertaking polyp surveillance, as they would support a strategy of focusing on growing polyps (and especially fast-growing polyps) as targets for resection; this strategy would likely be most effective when paired with resection of polyps reaching a certain size threshold regardless of growth rate. While colonoscopic polypectomy is generally a safe and well-tolerated procedure, it is not without risk nor cost.[38–42] By focusing on growing polyps, a greater number of patients could be spared the costs and risks of polypectomy while safely continuing to monitor polyps for future growth. As an example, of the 1,977 patients who had a positive initial CTC screening study (at least one polyp ≥6 mm), less than half (46%) had large (≥10 mm) or more than two small (6-9 mm) polyps which would warrant immediately polypectomy, while the remaining 54% (1,067 patients) were eligible for CTC surveillance with only one or two small (6-9 mm) polyps. Assuming the 25% growth rate over nearly five years observed in this study, the strategy of proceeding to polypectomy for patients with growing polyps while maintaining CTC surveillance at the recommended three-year interval for patients with stable or regressing polyps would obviate the need for polypectomy in up to 800 (75% of 1,067) patients, which is approximately 40% of the 1,977 positive patients in this study or 7% of the initial screening cohort of 11,165 patients.

This study has several limitations. Given the real-world standards of care and limitations of a clinical screening program, there is some unavoidable selection bias in our study group. Patients with large polyps or more than two small polyps at the time of initial screening are not eligible for CTC surveillance, and a substantial portion of patients who were eligible for CTC surveillance instead chose immediate polypectomy, deferred follow-up, or were lost to follow-up. Despite this bias, our cohort represents, to date, the largest and longest longitudinal study of in vivo colorectal polyps. Of all polyps in the study cohort, a quarter remained unresected and hence histopathologically unclassified at the conclusion of our 16-year study; a majority of these polyps remain under CTC surveillance because of their indolent growth behavior, and consequently may be presumed to be of little risk to the patient. We recognize that polyp volume measurement is not currently standard of care for colorectal cancer screening, and not widely available. However, we feel that prior and current data is clear that volume represents a more informative surrogate for polyp size than linear diameter. From a clinical standpoint, although volumetric measurement is not widely feasible at CTC or colonoscopy screening at this time, we would encourage screeners to be cognizant of their subjective impressions of polyp bulk, as polyps appearing bulkier are likely to have a greater chance of being clinically important. Finally, beyond histopathologic endpoints, the genetic and epigenetic composition of resected polyps may correlate with volumetric growth rate,[43] which we plan to investigate in future radiogenomic correlation.

In conclusion, this 16-year natural history study represents the largest and most comprehensive longitudinal evaluation of colorectal polyps to date. We confirm that small polyps represent little overall risk to the patient and may be safely monitored, potentially sparing unnecessary polypectomy and associated risks and costs. Finally, small colorectal polyps which grow at follow-up and those which arise de novo following initial screening are more likely to be histopathologically advanced, which may be of value in clinical decision-making.

Summary.

1. What is already known about this subject?

• Prior research—largely cross-sectional and epidemiologic studies—shows that most small colorectal polyps will never progress to cancer.

• However, there is little direct observational or longitudinal data on the growth of colorectal polyps, and most prior studies have relied on outdated modalities, have limited follow-up, or examine non-screening populations.

2. What this study adds:

• A minority of 6-9 mm polyps will grow in a 5-year period. However, this 25% of growing lesions will present with a 3-fold increased risk of advanced histology as compared with stable or regressing polyps.

• The risk of advanced histology in 6-9 mm polyps increases with the speed of growth, being up to 10-fold in rapidly growing lesions.

• Compared with patients undergoing immediate polypectomy for small 6-9 mm polyps, patients who entered CTC surveillance had a significantly higher prevalence of adenomatous (87% vs 68%) and histologically advanced (13% vs 2%) polyps.

• Colorectal polyps may regress or resolve without intervention, occurring in approximately 13% of small 6-9 mm polyps in this study.

3. How this study might affect research, practice, or policy:

• Most smaller colorectal polyps harbor non-advanced histology and have a low tendency to progress to advanced lesions or cancer. A small fraction will exhibit more aggressive behaviour in terms of growth and advanced histology, substantiating asymmetric evolution between non-advanced and advanced adenomas.