SEARCH

SEARCH BY CITATION

Keywords:

  • colonoscopy;
  • adenoma;
  • colorectal neoplasia;
  • natural history

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The data of the National Polyp Study, a large longitudinal study on surveillance of adenoma patients, is used for testing assumptions on the adenoma-carcinoma sequence. The observed adenoma and colorectal cancer incidence in the National Polyp Study were compared with the simulated outcomes of the MISCAN-COLON model of epidemiology and control of colorectal cancer for the U.S. population based on expert opinion. Variants of this model were explored in order to identify assumptions on the adenoma-carcinoma sequence that are consistent with the study observations. The high observed adenoma detection rates at surveillance and low observed colorectal cancer incidence in the National Polyp Study could only be explained by assuming a high incidence rate of adenomas accompanied by regression of adenomas. The National Polyp Study data suggest that adenoma prevalence results from a dynamic process of both formation as well as regression of adenomas. This lowers the expectations for the effects of colorectal cancer screening strategies that focus on adenoma detection. © 2004 Wiley-Liss, Inc.

The evolution of colorectal cancer from a precursor lesion, the adenoma, was first reported in studies from St. Mark's Hospital in London and later designated by Morson et al.1, 2 as the adenoma-carcinoma sequence. Morson1 stated that the evolution of cancer from adenomas takes at least 5 years and may be more than 20 years. Introduction of colonoscopy provided an opportunity for clarifying this sequence because of its ability to examine the entire colon and remove polyps for pathologic examination. These pathology studies have in recent years been correlated with molecular genetic studies.3 The adenoma-carcinoma sequence is now well established as the major pathway for the development of colorectal cancer in the general population and in high-risk patients in families with familial adenomatous polyposis (FAP) and hereditary nonpolyposis colorectal cancer (HNPCC).2, 4 The epidemiology and natural history of adenomas are not only important for choosing the optimal follow-up policy after polypectomy, but also for evaluating endoscopic screening for colorectal adenomas and cancer. A better understanding of the dynamics of the adenoma-carcinoma progression would further clarify what can be expected of various colorectal cancer screening strategies that involve adenoma detection.

The National Polyp Study (NPS) was a longitudinal study that provided prospective data on the adenoma-carcinoma sequence and the effect of colonoscopic polypectomy. It was organized in 1978 and began to accrue patients in 1980. Its purpose was to evaluate more frequent and less frequent follow-up surveillance intervals in patients in whom newly diagnosed adenomas were removed. Removal of these adenomas resulted in a colorectal cancer incidence that was markedly lower than expected without polypectomy. In this report, we present a study of the natural history of the adenoma-carcinoma sequence, applying a microsimulation model for colorectal cancer epidemiology and control (MISCAN-COLON)5 to the data of the National Polyp Study.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

National Polyp Study data

The National Polyp Study was a randomized controlled trial of colonoscopic surveillance in patients who have had at least one adenoma removed.6 All patients referred for colonoscopy or polypectomy between November 1980 and February 1990 in 7 participating centers who did not have a family or personal history of familial polyposis, inflammatory bowel disease, or a personal history of polypectomy or colorectal cancer were eligible for enrollment in the study. A total of 9,112 subjects referred for colonoscopy were candidates for the study. Of these patients, 6,480 were not eligible for the National Polyp Study because they did not meet the eligibility criteria of having one or more adenomatous polyps; 4,763 were excluded because no polyps were found. Other excluded subjects were 776 with nonadenomatous polyps only, 549 with colorectal cancer, 149 with inflammatory bowel disease or other conditions and 35 with a sessile adenoma with a base larger than 3 cm. If the first colonoscopy was incomplete, a repeat colonoscopy was scheduled. The colon had to be cleared after 3 examinations and within 3 months for the patient to be part of the study. Two hundred and eight patients with incomplete initial examinations were excluded. Of the included patients, 13% had more than one colonoscopy. Of the 2,632 eligible patients, 1,214 patients were eligible but did not choose to participate and 1,418 patients consented to participate and were randomized to one of 2 arms. The eligible but nonrandomized and the eligible randomized adenoma patients had similar characteristics.6 All detected polyps were removed and a surveillance colonoscopy was offered to the randomized patients in arm A at 1, 3 and 6 years after initial colonoscopy and in arm B at 3 and 6 years after initial colonoscopy. If the colon was not cleared with high confidence at surveillance colonoscopy, the patient was scheduled for repeat colonoscopy. Mean follow-up time was 5.9 years. Five cancers were found during the trial (CI = 1.6–11.7; 2 in arm A and 3 in arm B), while 21 were expected based on the U.S. population with the same age and sex distribution7 and 43–48 were expected based on a comparison with 2 polyp-bearing cohorts without intervention.8, 9 All 5 cancers were asymptomatic malignant adenomas detected at surveillance colonoscopy.

Model

The results of the MISCAN-COLON model are generated by microsimulation of individuals in whom adenomas and subsequent colorectal cancer may develop. Although the MISCAN-COLON model is originally designed for evaluation of population-based screening in an asymptomatic population,10 it can also be used to simulate surveillance after polypectomy. The output of the model consists of the adenoma and cancer detection rates at initial and surveillance colonoscopy and the effect of initial and surveillance colonoscopy on cancer incidence and mortality.

Parameter values in the expert-opinion-based model (expert MISCAN-COLON model) as presented in Table I have been established during 2 meetings at the U.S. National Cancer Institute.5, 10, 11 The disease stages that are distinguished in the modeled adenoma-carcinoma sequence are shown in Figure 1. It is assumed that adenomas are either nonprogressive and will never develop into cancer in a lifetime or progressive and are destined to develop into colorectal cancer. The average duration between incidence of a progressive adenoma and clinical diagnosis of cancer is assumed to be 20 years. The duration between adenoma incidence and preclinical colorectal cancer is assumed to be exponentially distributed with a mean of 16.4 years, while the duration of preclinical cancer is exponentially distributed with a mean of 3.6 years. It is assumed that polypectomy completely prevents growth of the polyp into cancer.

Table I. Main Assumptions in the Expert Miscan-Colon Model, Established in Expert Meetings at the National Cancer Institute in 1996 and 1997
ParameterValueBased on
  • 1

    This is not an explicit assumption, but a consequence of the assumptions about the adenoma incidence in the general population and the construction of the simulated NPS population as shown in Figure 2.

Adenoma incidence in general populationAge-dependentAdenoma prevalence in autopsy and colonoscopy studies of 15% in age group 50–59 to 33% in age group 70+,12, 13, 14, 15, 16 cancer incidence in SEER registry in 197817
 40–49 years; 0.9% per year
 50–59 years; 1.9% per year
 60–69 years; 3.3% per year
 70–79 years; 2.6% per year
Adenoma incidence in NPS population1Age-dependentResulting from the assumed adenoma incidence in the general population and the construction of the NPS population; see Figure 2
 40–49 years; 2.9% per year
 50–59 years; 6.1% per year
 60–69 years; 7.4% per year
 70–79 years; 5.9% per year
Duration distributions in preclinical stagesExponentialExpert opinion, other cancer models18, 19, 20
Mean duration of nonprogressive adenomasLifelongExpert opinion
Mean duration of progressive adenomas16.4 yearsExpert opinion
Mean duration of preclinical cancer3.6 yearsCancer detection rate at first screening and background cancer incidence in FOBT trials21, 22
Probability to develop cancer from removed adenoma0%Expert opinion
Probability that a new adenoma is progressiveDependent on age at onsetAdenoma prevalence in autopsy and colonoscopy studies of 15% in age group 50–59 to 33% in age group 70+,12, 13, 14, 15, 16 cancer incidence in SEER registry in 197817
 0–65 years; 14%
 65–100 years; linearly increasing from 14% to 96%
Distribution of risk for adenomas over the general populationGamma distributed, mean 1, variance 2Multiplicity distribution of adenomas in autopsy studies16
Sensitivity of fictitious screening test Adenoma ≤ 5 mm; 15%Adenoma size at initial polypectomy of all patients included in the National Polyp Study; see Table II
 Adenoma 6–9 mm; 28%
 Adenoma 10+ mm; 99%
 Cancer: 100%
Sensitivity of initial and surveillance colonoscopic examination Adenoma ≤ 5 mm; 80%Back-to-back colonoscopy studies23, 24, 25
 Adenoma 6–9 mm; 85% 
  Adenoma 10+ mm; 95% 
  Cancer; 95% 
Reach of initial and surveillance colonoscopic examination100%National Polyp Study design6
thumbnail image

Figure 1. Adenoma and cancer stages in the MISCAN-COLON model. Arrow A indicates the onset of adenomas. Arrow B (dotted) indicates the possibility of regression. In the expert model, it is assumed that adenomas do not regress. Arrow C indicates the moment of diagnosis of colorectal cancer. In the expert model, it is assumed that adenomas are either nonprogressive and will never develop into cancer in a lifetime or progressive and are destined to develop into colorectal cancer. Adenomas are categorized by size. It is assumed that 50% of nonprogressive adenomas will remain 6–9 mm and 50% will progress to the ≥ 10 mm stage. For progressive adenomas, it is assumed that 30% will develop through the sequence: ≤ 5 mm adenoma [RIGHTWARDS ARROW] 6–9 mm adenoma [RIGHTWARDS ARROW] preclinical colorectal cancer and that 70% will develop through the sequence ≤ 5 mm adenoma [RIGHTWARDS ARROW] 6–9 mm adenoma [RIGHTWARDS ARROW] ≥ 10 mm adenoma [RIGHTWARDS ARROW] preclinical colorectal cancer. For modeling reasons, in the models with regression of adenomas, it is assumed that regression occurs but only from adenomas 6–9 mm and ≥ 10 mm. The assumption that regression does not occur from adenomas ≤ 5 mm has no influence on the simulated adenoma detection rates and cancer rates in these models.

Download figure to PowerPoint

thumbnail image

Figure 2. Construction of the simulated National Polyp Study population. Simulated compliance with screening and initial colonoscopic examination is 100%. Simulated compliance with surveillance colonoscopic examinations is less than 100% and is based on the observed compliance in the National Polyp Study.

Download figure to PowerPoint

Table II. Characteristics at Initial Polypectomy of All Patients and Their Adenomatous Polyps Included in the National Polyp Study, as Observed in the National Polyp Study and as Simulated in the Expert Miscan-Colon Model
CharacteristicObservedSimulated
  • n = 1,418.

  • 1

    Forty-four patients with polyps classified as adenomas by the local pathologist were classified as nonadenomas by the review pathologists and were excluded from the National Polyp Study cohort simulated in this modeling study (n = 1,374).

Age  
 < 5013%11%
 50–5928%27%
 60–6939%40%
 70–7918%20%
 80+2%3%
Adenoma size1  
 ≤5 mm27%27%
 6–9 mm18%17%
 ≥10 mm55%56%
No. of adenomas1  
 157%61%
 222%23%
 ≥320%16%
Site of largest adenoma1  
 Distal colon64%61%
 Proximal colon36%39%

If all individuals have equal risk for adenomas, i.e., adenomas are randomly distributed over the population, the resulting adenoma multiplicity is Poisson-distributed. However, autopsy studies show a larger-than-Poisson variation,16 probably because of variation in genetic and environmental factors. The model accounts for the heterogeneity in adenoma multiplicity by drawing a risk index for each individual. The individual adenoma incidence rate is equal to the individual risk index multiplied by the age-specific adenoma incidence rate. This risk index is drawn from a gamma distribution with a mean of 1 and a variance of 2, which is chosen to fit the multiplicity distribution of adenomas in autopsy studies.26 The probability that a new adenoma is progressive is age-dependent but does not depend on the individual risk index. The age-specific adenoma incidence and the probability that an adenoma is progressive is chosen to fit observed U.S. cancer incidence in 1978 before the introduction of screening17 and prevalence of adenomas in autopsy and colonoscopy studies.12, 13, 14, 15, 16

The participants in the National Polyp Study had adenomas diagnosed and removed. The MISCAN-COLON model is adapted to this situation by applying a fictitious screening test to the general population to select individuals with adenomas detected at diagnostic colonoscopy. The fictitious screening test can be regarded as a mixture of fecal occult blood tests (FOBT) and sigmoidoscopy. These individuals constituted the simulated trial population. Figure 2 shows how the simulated National Polyp Study population was constructed. Like in the National Polyp Study, simulated individuals with colorectal cancer diagnosed at the diagnostic colonoscopy were excluded from the trial population. The sensitivity of the fictitious screening test was adjusted to reproduce the age distribution, the distribution of adenomas over the distal and proximal colon and the size and multiplicity distribution of adenomas at initial polypectomy in the National Polyp Study.

In the National Polyp Study, incomplete initial and surveillance colonoscopies were followed by repeat colonoscopy until the bowel was cleared with high confidence. For the purpose of modeling this clinical situation, we define a colonoscopic examination as a series of one or more colonoscopies in a short time period of which at least one reaches the cecum and the examination is considered to be of high confidence. Then in the model, it is assumed that all initial and surveillance colonoscopic examinations consist of one colonoscopy and that the reach of the colonoscopy is 100%, i.e., the complete bowel is visualized so that subsequently none of the patients need to undergo a second colonoscopy. Sensitivity of the initial and surveillance colonoscopic examinations is based on tandem studies of colonoscopy and increases from 80% for adenomas ≤ 5 mm to 95% for preclinical cancer.23, 24, 25 The simulated population of 5 independent simulations of 30,000 each (5 × 30,000 = 150,000) was designed to be approximately 100 times as large as the observed National Polyp Study cohort of 1,418 patients to minimize chance variation in the simulation results. The National Polyp Study surveillance schema and the observed compliance rates per arm and round are applied to the simulated trial population.

Analysis

Outcomes of the model are the simulated number of cancers during the trial and the simulated number of surveillance examinations at which adenomas are detected. The model further differentiates between cancers that are detected by a surveillance examination and those that are interval-detected. These cancers are further subdivided into those originating from adenomas or preclinical cancers missed at initial colonoscopy, and those in newly developed, fast-progressing lesions.

The observed cancer and adenoma incidence rates in the trial were compared with the rates as simulated by the MISCAN-COLON model based on expert opinion. In case of discrepancies between observed and simulated results, we varied a few pivotal assumptions in order to search for models that are consistent with observed results. Parameters that were varied are the adenoma incidence in the trial population, the duration distribution of progressive adenomas, the spontaneous regression rates of nonprogressive adenomas and the sensitivity of colonoscopy.

The goodness of fit of each set of model assumptions is evaluated by the deviance, which compares 5 outcomes of the model with the observed National Polyp Study results. The results that are included in the deviance are the number of surveillance-detected (asymptomatic) cancers (observed in the National Polyp Study: 5), the number of interval cancers (observed: 0), the number of surveillance examinations with adenomas in arm A at the first surveillance examination (observed: 150 in 545 examinations), in arm A at the second surveillance examination (observed: 73 in 338 examinations) and in arm B at the first surveillance examination (observed: 137 in 428 examinations). The deviance is defined as

  • equation image

where ki is the observed number of occurrences for outcome i, ni is the observed number of participants for the cancer results and the number of examinations for the adenoma results, pi = ki/ni is the observed rate and λi is the simulated rate. A low deviance indicates a good fit with the National Polyp Study data. If the deviance is higher than 11.07, the simulated results are significantly different from the observed results in the National Polyp Study.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

MISCAN-COLON model based on expert opinion

Table II shows that the MISCAN-COLON model based on expert opinion simulated a cohort that successfully reproduced the characteristics at initial polypectomy of the National Polyp Study population. However, this expert MISCAN-COLON model simulates a cancer incidence during the surveillance period of 1.5 per 1,000 person-years, which is more than twice as high as the observed 0.6 (95% CI = 0.2–1.4), while it simulates an 18% adenoma detection rate at surveillance examinations, which is considerably lower than the observed 27% (95% CI = 25–30%; see model A in Table III). Of the simulated cancer incidence in the first 6 years after initial polypectomy, 61% is caused by cancers developed from new progressive adenomas, 22% is caused by missed adenomas that progressed into cancer and 18% is from preclinical cancers missed at initial colonoscopy. Of the simulated cancers, 40% is found at surveillance colonoscopy, and 60% are diagnosed because of symptoms, while in the National Polyp Study all 5 incident cancers were detected at surveillance. The overall goodness of fit of the expert model is poor, mainly caused by the poor fit of adenoma detection rates.

Table III. Cancer Incidence Rate and Proportion of Surveillance Examinations with Adenomas as Observed in the National Polyp Study Population and as Simulated with the Expert Miscan-Colon Model and Several Model Variants
 Cancer rate per 1,000 person-yearsProportion of surveillance examinations with adenomas
Surveillance detectedInterval detectedAllArm A, year 1Arm A, year 3Arm B, year 3AllDeviance
  • 1

    95% confidence interval of 0.2–1.4.

  • 2

    95% confidence interval of 0.25–0.30.

Observed0.60.00.610.280.220.320.272 
 Model A, expert MISCAN-COLON assumptions0.60.91.50.170.130.250.1884
Assumptions intended to raise the adenoma detection rate        
 Model B, low adenoma sensitivity (60%)1.31.72.90.280.190.310.2732
 Model C, high adenoma incidence1.01.42.40.260.210.360.2828
 Model D, high adenoma incidence and spontaneous regression0.50.71.10.210.250.340.2624
Assumptions intended to reduce the cancer incidence rate        
 Model E, no fast-growing adenomas (constant duration of 20 years)0.30.30.60.160.110.230.17104
 Model F, high cancer sensitivity (100%)0.50.71.20.170.120.240.1883
Assumptions intended to fit both the cancer incidence and adenoma detection rate        
 Model G, no fast-growing adenomas, high adenoma incidence and spontaneous regression0.20.20.40.210.260.350.2727
 Model H, high cancer sensitivity, high adenoma incidence and spontaneous regression0.40.61.00.220.250.340.2623

Natural history assumptions that better explain observed rates

Higher model-simulated adenoma detection rates than in the expert MISCAN-COLON model can be achieved with a lower sensitivity of colonoscopy for adenomas or a higher adenoma incidence.

Low sensitivity for adenomas.

The sensitivity for adenomas has to be extremely low in order to simulate the observed adenoma detection rates in the National Polyp Study, which conflicts with the low observed cancer incidence (model B in Table III).

Higher adenoma incidence.

The simulated adenoma detection rates are more in agreement with the observations when the adenoma incidence rate in the patients referred for colonoscopy is doubled (model C in Table III). The resulting adenoma prevalence is not in agreement anymore with the adenoma prevalence in the unscreened general population, which is about 33% in the 70+ age category.12, 13, 14, 15, 16, 27 However, the National Polyp Study cohort is a selected population with an adenoma incidence that may be higher than in the general population. A serious problem is that increasing the adenoma incidence also increases the risk for cancers, thus further increasing the already too high simulated cancer incidence (from 1.5 to 2.4 per 1,000 person-years compared to 0.6 observed). This could theoretically be resolved by restricting the increase in incidence to nonprogressive adenomas. However, this would make the cancer risk in patients with multiple adenomas similar to the cancer risk in patients with only one adenoma, which is not consistent with published data that show that adenoma multiplicity is a risk factor for colorectal cancer.28, 29 Therefore, assuming a higher adenoma incidence that is associated with a higher adenoma prevalence has to be rejected.

Higher adenoma incidence combined with regression.

The high adenoma detection rates in the National Polyp Study can also be explained by assuming high adenoma incidence compensated by spontaneous regression of (nonprogressive) adenomas. If spontaneous regression occurs regularly, adenoma incidence can be high while adenoma prevalence agrees with observed prevalence, even in older-age groups where adenoma prevalence and multiplicity hardly increase according to autopsy and colonoscopy studies.

Although it is generally assumed that adenomas grow into cancer or remain in the colon until death, spontaneous regression or washout of adenomas has been reported.30, 31, 32, 33 In the observational study of Knoernschild,31 the mucosa near asymptomatic benign polyps was tattooed in 257 patients. After follow-up of 3–5 years, the polyp had completely disappeared in 18% of the patients. Table III shows the results of model D in which nonprogressive adenomas disappear on average after 5 years with an exponentially distributed duration. The adenoma incidence is 3 to 5 times higher than in the expert MISCAN-COLON model. Between the ages 55 years and 84 years, the incidence is approximately 10% per year in the general population with a peak in age group 70–74 years of 16% per year. This model variant results in adenoma detection rates that are more in agreement with the National Polyp Study observations. In this model, the simulated colorectal cancer risk in the National Polyp Study population is approximately 30% lower than in the expert MISCAN-COLON model. This is a consequence of the assumption of spontaneous regression of nonprogressive adenomas combined with the increased incidence of nonprogressive adenomas to compensate for the decrease in adenoma prevalence caused by regression. This explains why the simulated colorectal cancer incidence is lower than in the expert MISCAN-COLON and not significantly different from the observed colorectal cancer incidence (model D in Table III).

In summary, a high adenoma incidence combined with spontaneous regression of adenomas is the only explanation of the observed adenoma detection rate that does not increase simulated cancer incidence and even decreases the simulated cancer incidence. The deviances of models B, C and D are all lower than the deviance of the expert MISCAN-COLON model, which indicates that models B, C and D are more in agreement with the National Polyp Study results. The simulated results of models B and C are still significantly different from the observed results (p < 0.05) mainly due to the difference in interval-detected cancers. The simulated results of model D are significantly different from the observed results (p < 0.05) due to the difference in interval-detected cancers and the simulated adenoma detection rate in arm A, year 1.

To decrease the simulated cancer incidence further, we explored 2 possibilities for lowering the cancer incidence: no fast-growing progressive adenomas, and a high sensitivity of colonoscopy for cancer.

No fast-growing adenomas.

In the expert MISCAN-COLON model, approximately 30% of the progressive adenomas will develop into cancer within 6 years due to the exponentially distributed duration of 20 years on average. With fewer fast-growing progressive adenomas than assumed in the expert MISCAN-COLON model, cancers from new polyps will not surface in the first years after polypectomy and thus the incidence will remain low. As an example, model E in Table III is a model in which adenomas do not develop into cancer within the trial period, i.e., there are no fast-growing (within 6 years) progressive adenomas. Under these assumptions, none of the incident cancers in the first 6 years after polypectomy are newly developed, 74% develop from cancers missed at initial colonoscopy and 26% develop from adenomas missed at initial colonoscopy. The percentage of cancers developing from missed adenomas is small, because most missed adenomas are small and take more than 6 years to develop. In this simulation, 51% of the cancers are diagnosed at surveillance examinations, compared to 39% in the expert model. The simulated cancer incidence rate decreases from 1.5 to 0.6 per 1,000 years, which is equal to the observed rate.

Higher sensitivity for cancer.

The assumed sensitivity of colonoscopy for cancer is 95% in the expert model, based on a retrospective study of colonoscopic sensitivity for cancer.25 However, in the National Polyp Study, sigmoidoscopy or barium enema was performed as the reason for referral for colonoscopy in 25% and 44% of the patients, respectively,34 and often additional colonoscopies were performed to resolve cases, which gives extra opportunities to detect cancer in these patients. Raising the sensitivity for preclinical cancer in the MISCAN-COLON model from 95% to 100% reduces the incidence from missed cancers. But because missed cancers cause only 18% of the cancer cases in the expert model, the decrease in cancer incidence is modest, from 1.5 to 1.2 per 1,000 years (model F in Table III).

Neither of these 2 assumptions that decrease the cancer incidence in the study period affect the simulated adenoma detection rates, which remain too low. Because the deviance largely depends on the adenoma detection rates, the deviances of model E and F are comparable or higher than the expert MISCAN-COLON model (model A). Furthermore, neither assumption explains the fraction of asymptomatic cancers in the National Polyp Study. All observed cancers (n = 5) were asymptomatic in the National Polyp Study. Even in a model that assumes 100% cancer sensitivity and a long constant duration, only 68% of the simulated cancers are asymptomatic. This is caused by incomplete compliance of the participants, which leads to some symptomatic cancers detected in nonresponders.

Table III shows the results of the MISCAN-COLON model with high adenoma incidence and spontaneous adenoma regression, combined with no fast-growing progressive adenomas (model G), and combined with 100% colonoscopic sensitivity for cancer (model H). The simulated cancer incidence in model G is more than 60% reduced compared to model F, while the simulated cancer incidence in model H is only slightly reduced compared to model F. In model D, the simulated cancer incidence is higher than observed, but not significantly different. If that model is modified with the assumption of no fast growing adenomas (model G), the simulated cancer incidence is lower than observed, but again not significantly different. Thus, models that include the assumption of development and regression of adenomas and in which the percentage of the progressive adenomas that will develop into cancer within 6 years is between 0% and 30% are consistent with the observed cancer incidence. The deviances of model G and H are comparable with the deviance of model D.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The assumptions of the expert MISCAN-COLON model were developed in collaboration with the U.S. National Cancer Institute in meetings of a group of colorectal cancer experts.5 In order to clarify the natural history of the adenoma-carcinoma sequence, a few pivotal natural history assumptions made by this group were modified to determine which natural history assumptions best fit the National Polyp Study observed data. The expert MISCAN-COLON model predicted a higher cancer incidence and lower adenoma detection rates than observed in the National Polyp Study. In order to have the highest concordance between the model results and the National Polyp Study results, a new factor had to be introduced, i.e., adenoma regression.

High adenoma incidence combined with regression accounted for the high percentage of patients with adenomas at surveillance, without losing its consistency with adenoma prevalence data from autopsy studies and without increasing the colorectal cancer incidence during the study. The high incidence is supported by a study of repeat colonoscopy that estimated that the 1-year adenoma incidence rate is 11%.35 The assumption that adenomas spontaneously regress is supported by previous findings in short-term studies that adenomas may disappear or regress in size31, 32, 36 and by the observation that the adenoma prevalence does not increase after age 70.12, 13, 14, 15, 16 A recent study of celecoxib in patients with familial adenomatous polyposis reported adenoma regression in the control group. In this study, a tattoo was placed at baseline endoscopy near a small area with a high density of polyps. Repeat endoscopy was performed 6 months later and the number of polyps at the tattooed area in the placebo group was 4.5% less than at baseline endoscopy.37 The Telemark study recently reported that adenoma prevalence in patients who had undergone sigmoidoscopy 13 years before was not significantly lower than in the patient without previous sigmoidoscopy. The authors mention adenoma regression as one of the possible explanations.38

In an attempt to identify a group of adenoma patients at high risk for colorectal cancer, the adenoma stages in the MISCAN-COLON model could be extended with grade of dysplasia and histology. However, to do so, one should make assumptions about the relationship between size, grade of dysplasia and histology in the adenoma-carcinoma sequence and this relationship is not well known.

The cost-effectiveness of repeat sigmoidoscopy or colonoscopy colorectal cancer screening in the general population has been studied using models.39, 40, 41 None of these included the assumption that adenomas regress and adenoma incidence is accordingly high. Regression combined with high adenoma incidence makes adenoma detection as a strategy for colorectal cancer prevention less favorable, because more adenomas will develop in the population after polypectomy. This means that many individuals will have adenomas at surveillance colonoscopy, which will result in polypectomy and pathology. This will increase the costs of endoscopic screening and surveillance. Furthermore, it is likely that more individuals will develop at least one adenoma. Thus, in repeat screening rounds, many individuals without previous adenomas will have adenomas detected. This will result in more surveillance colonoscopies. There is less difference in risk level between individuals with and without adenomas, making each single surveillance colonoscopy less effective.

The National Polyp Study provided the opportunity to examine the dynamics of the natural history of the adenoma-carcinoma sequence. The outcome suggests that the adenoma-carcinoma sequence is a dynamic process of formation and regression of adenomas. This has negative consequences for the effects and costs expected from endoscopic colorectal cancer screening and surveillance of adenoma patients.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors thank Martin Brown, project officer of the National Cancer Institute.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  • 1
    Morson BC. Genesis of colorectal cancer. Clin Gastroenterol 1976; 5: 50525.
  • 2
    Morson BC. The evolution of colorectal carcinoma. Clin Radiol 1984; 35: 42531.
  • 3
    Vogelstein B, Fearon ER, Hamilton SR, Kern SE, Preisinger AC, Leppert M, Nakamura Y, White R, Smits AM, Bos JL. Genetic alterations during colorectal-tumor development. N Engl J Med 1988; 319: 52532.
  • 4
    Muto T, Bussey HJ, Morson BC. The evolution of cancer of the colon and rectum. Cancer 1975; 36: 225170.
  • 5
    Loeve F, Boer R, van Oortmarssen GJ, van Ballegooijen M, Habbema JDF. The MISCAN-COLON simulation model for the evaluation of colorectal cancer screening. Comput Biomed Res 1999; 32: 1333.
  • 6
    Winawer SJ, Zauber AG, O'Brien MJ, Ho MN, Gottlieb L, Sternberg SS, Waye JD, Bond J, Schapiro M, Stewart ET, Panish J, Ackroyd F, et al. Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps: the National Polyp Study Workgroup. N Engl J Med 1993; 328: 9016.
  • 7
    Winawer SJ, Zauber AG, Ho MN, O'Brien MJ, Gottlieb LS, Sternberg SS, Waye JD, Schapiro M, Bond JH, Panish JF, Ackroyd F, Shike M, et al. Prevention of colorectal cancer by colonoscopic polypectomy: the National Polyp Study Workgroup. N Engl J Med 1993; 329: 197781.
  • 8
    Atkin WS, Morson BC, Cuzick J. Long-term risk of colorectal cancer after excision of rectosigmoid adenomas. N Engl J Med 1992; 326: 65862.
  • 9
    Stryker SJ, Wolff BG, Culp CE, Libbe SD, Ilstrup DM, MacCarty RL. Natural history of untreated colonic polyps. Gastroenterology 1987; 93: 100913.
  • 10
    Loeve F, Brown ML, Boer R, van Ballegooijen M, van Oortmarssen GJ, Habbema JDF. Endoscopic colorectal cancer screening: a cost-saving analysis. J Natl Cancer Inst 2000; 92: 55763.
  • 11
    Loeve F, Boer R, van Ballegooijen M, van Oortmarssen GJ, Habbema JDF. Final report MISCAN-COLON microsimulation model for colorectal cancer: report to the National Cancer Institute Project No. NO1-CN55186. Rotterdam: Department of Public Health, Erasmus University, 1998.
  • 12
    Johnson DA, Gurney MS, Volpe RJ, Jones DM, VanNess MM, Chobanian SJ, Avalos JC, Buck JL, Kooyman G, Cattau EL Jr. A prospective study of the prevalence of colonic neoplasms in asymptomatic patients with an age-related risk. Am J Gastroenterol 1990; 85: 96974.
  • 13
    DiSario JA, Foutch PG, Mai HD, Pardy K, Manne RK. Prevalence and malignant potential of colorectal polyps in asymptomatic, average-risk men. Am J Gastroenterol 1991; 86: 9415.
  • 14
    Lieberman DA, Smith FW. Screening for colon malignancy with colonoscopy. Am J Gastroenterol 1991; 86: 94651.
  • 15
    Rex DK, Lehman GA, Hawes RH, Ulbright TM, Smith JJ. Screening colonoscopy in asymptomatic average-risk persons with negative fecal occult blood tests. Gastroenterology 1991; 100: 647.
  • 16
    Koretz RL. Malignant polyps: are they sheep in wolves' clothing? Ann Intern Med 1993; 118: 638.
  • 17
    National Cancer Institute, DCCPC SRP, Cancer Statistics Branch. Surveillance, epidemiology, and end results (SEER) program public use CD-ROM (1973–1998). Released April 2001, based on the August 2000 submission. Bethesda, MD: National Cancer Institute, 2001.
  • 18
    Walter SD, Day NE. Estimation of the duration of a pre-clinical disease state using screening data. Am J Epidemiol 1983; 118: 86586.
  • 19
    Gyrd-Hansen D, Sogaard J, Kronborg O. Analysis of screening data: colorectal cancer. Int J Epidemiol 1997; 26: 117281.
  • 20
    Launoy G, Smith TC, Duffy SW, Bouvier V. Colorectal cancer mass-screening: estimation of faecal occult blood test sensitivity, taking into account cancer mean sojourn time. Int J Cancer 1997; 73: 2204.
  • 21
    Hardcastle JD, Thomas WM, Chamberlain J, Pye G, Sheffield J, James PD, Balfour TW, Amar SS, Armitage NC, Moss SM. Randomised, controlled trial of faecal occult blood screening for colorectal cancer: results for first 107,349 subjects. Lancet 1989; 1: 11604.
  • 22
    Kronborg O, Fenger C, Olsen J, Bech K, Søndergaard O. Repeated screening for colorectal cancer with fecal occult blood test: a prospective randomized study at Funen, Denmark. Scand J Gastroenterol 1989; 24: 599606.
  • 23
    Hixson LJ, Fennerty MB, Sampliner RE, Garewal HS. Prospective blinded trial of the colonoscopic miss-rate of large colorectal polyps. Gastrointest Endosc 1991; 37: 1257.
  • 24
    Rex DK, Cutler CS, Lemmel GT, Rahmani EY, Clark DW, Helper DJ, Lehman GA, Mark DG. Colonoscopic miss rates of adenomas determined by back-to-back colonoscopies. Gastroenterology 1997; 112: 248.
  • 25
    Rex DK, Rahmani EY, Haseman JH, Lemmel GT, Kaster S, Buckley JS. Relative sensitivity of colonoscopy and barium enema for detection of colorectal cancer in clinical practice. Gastroenterology 1997; 112: 1723.
  • 26
    Diggle PJ, Liang K, Zeger SL. Analysis of longitudinal data. Oxford: Clarendon Press. 1994.
  • 27
    Levin TR, Palitz A, Grossman S, Conell C, Finkler L, Ackerson L, Rumore G, Selby JV. Predicting advanced proximal colonic neoplasia with screening sigmoidoscopy. JAMA 1999; 281: 16117.
  • 28
    Lotfi AM, Spencer RJ, Ilstrup DM, Melton LJ III. Colorectal polyps and the risk of subsequent carcinoma. Mayo Clin Proc 1986; 61: 33743.
  • 29
    Bertario L, Russo A, Sala P, Pizzetti P, Ballardini G, Andreola S, Spinelli P. Risk of colorectal cancer following colonoscopic polypectomy. Tumori 1999; 85: 15762.
  • 30
    Cole JW, McKalen A, Powell J. The role of ileal contents in the spontaneous regression of rectal adenomas. Dis Colon Rectum 1961; 4: 4138.
  • 31
    Knoernschild H. Growth rate and malignant potential of colonic polyps: early results. Surg Forum 1963; 14: 13742.
  • 32
    Hoff G, Foerster A, Vatn MH, Sauar J, Larsen S. Epidemiology of polyps in the rectum and colon: recovery and evaluation of unresected polyps 2 years after detection. Scand J Gastroenterol 1986; 21: 85362.
  • 33
    Giardiello FM, Hamilton SR, Krush AJ, Piantadosi S, Hylind LM, Celano P, Booker SV, Robinson CR, Offerhaus GJ. Treatment of colonic and rectal adenomas with sulindac in familial adenomatous polyposis. N Engl J Med 1993; 328: 13136.
  • 34
    Winawer SJ, Zauber AG, O'Brien MJ, Gottlieb LS, Sternberg SS, Stewart ET, Bond JH, Schapiro M, Panish JF, Waye JD, Kurtz RC, Shike M, et al. The National Polyp Study: design, methods, and characteristics of patients with newly diagnosed polyps. Cancer 1992; 70 (Suppl 5): 123645.
  • 35
    Bensen S, Mott LA, Dain B, Rothstein R, Baron J. The colonoscopic miss rate and true one-year recurrence of colorectal neoplastic polyps: Polyp Prevention Study group. Am J Gastroenterol 1999; 94: 1949.
    Direct Link:
  • 36
    Hofstad B, Vatn MH, Andersen SN, Huitfeldt HS, Rognum T, Larsen S, Osnes M. Growth of colorectal polyps: redetection and evaluation of unresected polyps for a period of three years. Gut 1996; 39: 44956.
  • 37
    Steinbach G, Lynch PM, Phillips RK, Wallace MH, Hawk E, Gordon GB, Wakabayashi N, Saunders B, Shen Y, Fujimura T, Su LK, Levin B. The effect of celecoxib, a cyclooxygenase-2 inhibitor, in familial adenomatous polyposis. N Engl J Med 2000; 342: 194652.
  • 38
    Thiis-Evensen E, Hoff GS, Sauar J, Majak BM, Vatn MH. The effect of attending a flexible sigmoidoscopic screening program on the prevalence of colorectal adenomas at 13-year follow-up. Am J Gastroenterol 2001; 96: 19017.
    Direct Link:
  • 39
    Frazier AL, Colditz GA, Fuchs CS, Kuntz KM. Cost-effectiveness of screening for colorectal cancer in the general population. JAMA 2000; 284: 195461.
  • 40
    Khandker RK, Dulski JD, Kilpatrick JB, Ellis RP, Mitchell JB, Baine WB. A decision model and cost-effectiveness analysis of colorectal cancer screening and surveillance guidelines for average-risk adults. Int J Technol Assess Health Care 2000; 16: 799810.
  • 41
    Sonnenberg A, Delco F, Inadomi JM. Cost-effectiveness of colonoscopy in screening for colorectal cancer. Ann Intern Med 2000; 133: 57384.