Influence of long-term colonoscopic surveillance on incidence of colorectal cancer and death from the disease in patients with precursors (adenomas)

Abstract

Case-control studies and short term prospective studies have suggested that selected groups of patients with precursors of colorectal cancer may benefit from colonoscopic surveillance after initial removal of adenomas. The aim of the present study was to demonstrate such a possible benefit from long term (1–24 years) colonoscopic surveillance in a population of patients with all types of adenomas regardless of size and way of removal. Two thousand and forty-one patients with a first time diagnosis of colorectal adenoma were included in prospective surveillance between year 1978 and 2002. All adenomas were considered. Incidence of CRC and mortality from CRC was calculated, using age, sex, and calendar specific number of person years of follow-up for comparison with the standard Danish population. CRC was found in 27 patients, the expected number being 41 (RR 0.65, 95% CI 0.43–0.95). Three of the 27 patients died from CRC, the expected number being 25 (RR 0.12, 95% CI 0.03–0.36). A total of 6 289 colonoscopies resulted in severe complications in 20 patients and two died from complications. Long-term colonoscopic surveillance may reduce incidence of CRC as well as mortality in patients with sporadic adenomas. The benefit is reduced to a minor degree by complications from surveillance

It is well-known that patients with sporadic colorectal adenomas are at increased risk of getting colorectal cancer (CRC) compared to the standard population [1]. It is less certain how much this risk may be reduced by initial removal of the adenomas [2], [3] and removal during surveillance [4]. Retrospective studies [5], [6] and prospective studies of short-term colonoscopic surveillance (3–6 years) in selected adenoma patients [7] suggest that incidence and mortality from CRC may be reduced, even to lower figures than those of a normal population [8].

A comparison of surveillance versus no surveillance has not been performed in randomised trials (RCT) for ethic reasons, but different lengths between surveillance examinations have been evaluated in RCT's [9], [10], measuring risk of new neoplasia (adenoma and CRC) as well as complications from surveillance. The main aim of the present study was to evaluate any benefits and drawbacks from long-term surveillance of a patient population with all types of adenomas.

Patients and methods

The 2 041 patients included from year 1978 to 2002 were between 24 and 76 years old (average 60.8 years for men and 60.1 for women) at the initial adenoma removal. Demographics are described in Table I. Most patients had colorectal symptoms, but they were rarely related to adenomas and 634 were asymptomatic, being recruited from a population screening trial [11] and with faecal occult bleeding indicating colonoscopy. The persons with adenomas detected during the colonoscopy were allocated to surveillance programmes according to the same criteria as symptomatic patients (Table II). Initially they were between 45 and 75 years of age and represented the same age and sex distribution as that of the uptake area (Funen) as well as that of Denmark. No patients over 76 years of age were included initially, but surveillance was continued beyond that age depending upon accompanying severe diseases. The university hospital serves the region of Funen, having a population being representative for Denmark [16]. No patients had a history of FAP, HNPCC or inflammatory bowel disease. Patients participating in chemoprevention trials were excluded [12].

Table I.  2 041 patients and their adenomas at initial removal

	Men	Women
Age (years)
24–29	3	7
30–39	15	19
40–49	106	87
50–59	352	251
60–69	500	340
70–76	211	150
Number of adenomas
1	691	592
2	289	171
3 or more	207	91
Location
Rectum & sigmoid	889	695
Above sigmoid	106	74
Both	192	85
Diameter (largest in mm)
1–10	702	509
11–20	361	235
> 20	124	110
Dysplasia (most severe)
Mild	674	490
Moderate	402	281
Severe	111	83
Structure (most villous)
Tubular	912	609
Tubolovillous	265	226
Villous	10	19
Sessile >5 mm	202	153
Advanced adenoma
(>10 mm, villous, severe dysplasia)	497	362
Polypectomy (most extensive procedure)
Hot biopsy	187	143
Sling	931	647
Piecemeal	26	14
Excision	15	26
Resection	23	22
Unknown	5	2

Table II.  Planned surveillance according to adenoma morphology and time

Adenoma morphology	Years of inclusion	Interval strategy (months)	Patients
All pedunculated	1978–82 (RTC)*	6 versus 24 during	138
		5 years,
		then 12 versus 24
All pedunculated and small sessile and flat (≤5 mm)	1982–91 (RCT)*	24 versus 48	688
	1991–2002	36	836
Sessile >5 mm	1978–81	6 during 5 years then 12 6 versus 12 during 5 years then 12 versus 30	50
	1981–87 (RCT)*		117
	1988–2002 (RCT)*	12 versus 24	212
	Total		2041

*RCT: Randomisation of intervals

Large adenomas (>20 mm) were present in 234 patients and 355 had sessile adenomas above 5 mm. Surgical resection or excision became necessary in 86 (Table I). The classification of adenomas was done by one pathologist throughout the study [13] and was similar to that used in the National Polyp Study in USA [9]. Adenomas were considered advanced being >10 mm in diameter, villous or showing severe dysplasia (497 men and 362 women).

A colonoscopy was performed in all patients initially and was complete in 1 871; incomplete colonoscopy was supplemented by a double contrast barium enema (DCBE) in 148 patients, leaving 22 who first had documentation of clean colon (without neoplasia) at the following surveillance examination. In patients with multiple adenomas, large sessile adenomas removed by piecemeal technique or an unsatisfactory bowel preparation, colonoscopy was repeated within three months to obtain a clean colon, a prerequisite for inclusion. All were recruited and followed at the University Hospital of the county of Funen within the colorectal unit of the surgical department. The study was chaired by the second author during the 24 years. Intervals between planned colonoscopies varied between 6 and 48 months and were subjected to randomization in 1 155 patients (Table II). Early results from patients included from 1978 to 1991 have been published previously [14], [15], but no results have been published from the 836 patients followed every three years (Table II).

The relative risk (RR) of CRC and death from CRC in the total study population (2 041 patients) was calculated from 1978 to 2002 by dividing the observed number by the number expected in a standard Danish population with the same age and sex distribution [16]. Patients were considered at risk from initial adenoma removal until 1. November 2002 or to death before that time. The expected number in the normal population was determined by using the age, sex, and calendar specific number of person years of follow up for the study population in conjunction with age, sex, and calendar specific incidence and death rates for the standard Danish population. Consequently, the estimates of RR were adjusted for differences in the age, sex, and calendar specific rates. The 95% confidence intervals for RR were calculated from the cumulative Poisson distribution. If 1.0 was outside the interval, the observed and expected numbers were considered significantly different.

Lethal complications from the surveillance program were evaluated in the same way by comparing mortality rates from all causes using data from StatBank, Denmark. DK is a small country with a homogenous population and the Funen population represents one tenth of the whole population.

Compliance with surveillance was evaluated by dividing the number of person years from initial adenoma removal until the last surveillance examination by number of years until November 1, 2002 or death. No upper age limit was decided upon for ending surveillance, but accompanying severe disease was tantamount to ending the examinations; however, the patient was kept in the study.

Results

One hundred and fifteen of the original 2 041 patients had reached 24 years after inclusion at November 2002 (Table III). Colonoscopy had been performed 6 289 times and DCBE 998 times during 13 993 patient years of surveillance. The compliance was estimated to be 72.9% (8 132/11 155) in men and 76.3% (5 861/8 704) in women (overall 13 993/19 859) (Table V). Colonoscopy was complete in 95% (3 652/3 853) of the examinations in men and 92% (2 253/2 436) in women.

Table III.  2 041 patients under surveillance and cumulated number of patient years, colorectal examinations and neoplasia

	Time (months)
	24	48	72	96	120	192	240	288
Patients at risk	1979	1881	1737	1580	1362	687	315	115
Patients examined	1734	1624	1238	957	683	219	64	7
Years of surveillance	3412	6418	8821	10510	11754	13651	13946	13993
Colonoscopy	1333	3181	4139	4921	5398	6144	6265	6289
Complete	1209	2953	3865	4611	5056	5767	5881	5905
DCBE	199	487	641	781	853	969	992	998
Patients with adenomas	216	496	571	648	689	730	736	736
Patients with advanced adenomas	48	96	113	127	136	144	144	144
Patients with CRC	8	12	17	20	22	26	27	27

New adenomas were removed in 736 patients and 144 had advanced adenomas at least once (103 men and 41 women) (28 with severe dysplasia, 9 with villous adenomas and 131 with adenomas >10 mm). No microscopy was obtained in six of 210 polyps >10 mm in diameter and 25 of 758 polyps 6–10 mm in diameter, because they were lost.

CRC was found in 14 men and 13 women (Table IV). Three of the patients considered to have CRC in stage Dukes’ A had cancer within a polyp removed by polypectomy into healthy tissue, presuming no lymphnode metastases were present. In total, twenty of the 27 CRC's had no lymphnode metastases (Dukes’ A + B); 16 patients had left sided tumours, 10 right sided and one had CRC's in both sides. Age at diagnosis was 70 years (median 70, range 54–81). CRC was detected at planned surveillance colonoscopy in 18, whereas six patients became symptomatic before; the remaining three did not follow the planned examinations and CRC was detected 59, 75 and 177 months respectively after the last surveillance examination demonstrating a clean colon. The three patients only had one examination demonstrating a clean colon before CRC was found, whereas the other 24 patients had 126 examinations covering the future tumour area (median 3.5, range 1–17). Eight CRC's (seven men and one woman) were detected within 12 months of a clean colon. Previous adenomas in the same area as the CRC were found in 20 patients and ten had advanced adenomas, which were removed by piecemeal technique in four. Eighteen of the 27 patients with CRC had more than one adenoma initially. The time from first adenoma until CRC varied between 3 and 240 months (median 49; Table III).

Table IV.  Patients with colorectal cancer during surveillance

Dukes’ stage	Number of patients
A	12 (10)*
B	8 (4)
C	5 (3)
Distant spread	2 (1)
Status at November 2002
Death from CRC	3
Death from other cause	4
Alive with CRC	3
Alive without CRC	17

*detected at planned examination in ()

At the end of the study three patients had died from CRC and three had recurrent CRC, but were alive (Table IV). None of the 27 patients with CRC died immediately postoperatively (<30 days). The relative risks of CRC and death from CRC as well as all causes within the study period are presented in Table V. The incidence was significantly reduced over all, but not in women per se. Mortality from CRC was reduced significantly in both sexes. The all cause mortality was not changed significantly, except for a small, but significant reduction in women. 569 of the 2 041 patients had died at November 2002.

Table V.  Incidence and mortality risks between 1978 and November 2002 in study group compared with standard Danish population

	Observed	Expected
Patient years	19 859
Men	11 155
Women	8 704
Patients with CRC	27	41.3
Men	14	26.1
Women	13	15.2
RR (95% CI)	0.65 (0.43–0.95)
Men	0.54 (0.29–0.90)
Women	0.86 (0.46–1.46)
Death from CRC	3	24.5
Men	2	15.5
Women	1	9.0
RR (95% CI)	0.12 (0.03–0.36)
Men	0.13 (0.02–0.47)
Women	0.11 (0.00–0.62)
Death from all causes	569	611.9
Men	381	393.0
Women	188	218.9
RR (95% CI)	0.93 (0.86–1.01)
Men	0.97 (0.87–1.07)
Women	0.86 (0.74–0.99)

Severe complications from surveillance examinations were seen in 20 patients (Table VI) and two died (one 66 year old man from cardiopulmonary complications after suture of a diagnostic perforation of colon with accompanying colostomy, the other, being diabetic and having chronic bronchitis, from a coronary occlusion immediately after colonoscopy with polypectomy). The remaining 18 patients were fully restituted including the three with peritoneal reaction (polypectomy syndrome), who were treated conservatively.

Table VI.  Severe complications from surveillance

	Number of patients
Diagnostic colonic perforation (one death)	12
Therapeutic colonic perforation	3
Polypectomy syndrome	3
Severe bleeding after polypectomy	1
Coronary occlusion (lethal)	1

Discussion

We found that a population of patients with all types of adenomas, subjected to initial removal and following colonoscopic surveillance had a significant reduction of incidence (35%) of CRC as well as mortality (88%) from CRC compared to a standard population. The smaller, insignificant reduction of incidence in women may be related to the lesser compliance and fewer complete colonoscopies, but missed polyps were not more frequent in women and no woman got CRC after use of the piecemeal technique. Advanced adenomas were less frequent in women and were only seen in the same area as the CRC in one of the 13 women, who developed CRC in contrast to nine of the 14 men. Also, the period from the colon was considered clean the last time before diagnosis of CRC was not shorter in women (median 32 months) than in men (median 15 months). Another explanation might be that other pathways than the adenoma-carcinoma sequence could be more frequent in women [17].

The reduction of incidence contributed to a minor degree to the reduction in mortality, which already was demonstrated when half of the present patients were included in 1993 [15]. The reduction in mortality was mainly due to the more favourable stage of CRC. Usually, no more than 10% are Dukes’ A in the Danish population in contrast to the 12/27 in the present surveillance population. Polypectomy or local surgical excision was performed in five of the 12 patients with Dukes’ A CRC, contributing to the low postoperative mortality.

The trend towards a lower overall mortality than in the standard population may be explained by the selection of patients for surveillance, having no severe accompanying diseases at time of inclusion. Surveillance was not without complications, but the lethal ones compromised the reduction in mortality from CRC to a very minor degree.

The prototypes of colonoscopes changed during the 24 years of study, suggesting that severe complications may occur less often in the future. Using a less extensive examination like flexible sigmoidoscopy would have eliminated the chance of detecting ten of the present 27 cancers. The more recently introduced colonography (virtual colonoscopy) would have eliminated complications, but with the drawback that no polyps could be removed, making a following colonoscopy necessary. Furthermore, colonography may be less sensitive for detection of flat adenomas which are diagnosed in increasing numbers with different colonoscopic imaging techniques.

The present study does not allow for any strict guidelines to how often colonoscopy should be done. Prognostic factors for recurrent adenomas, like size and broad base, multiplicity and villousness were taken into consideration before deciding frequency of surveillance [10], [14], using shortest intervals (Table II) in those with large sessile adenomas. Also, multiplicity indicated repeated colonoscopy within 3 months of initial colonoscopy to secure a clean colon. Intervals longer than 4 years were believed to carry a lower compliance than the present and were therefore avoided. However, longer intervals and even no surveillance have been suggested in patients with single, less advanced adenomas [18–20]. A study from UK [21], based on similar patients like the present, has suggested that surveillance intervals of 5 years may be as effective as two years considering recurrent adenomas. No conclusion could be reached about risk of CRC and intervals. We very recently [22] found a higher risk of CRC using surveillance intervals of 4 years instead of 2 in patients with previous pedunculated and small flat and sessile adenomas. Even shorter intervals than 2 years were suggested in patients with larger sessile adenomas.

The German guidelines [19] are based on long term prospective surveillance, but in a simplistic program and without any documentation for decrease of incidence of CRC and death rates. An Italian study [23] has reported reduced incidence (RR 0.34) of CRC in a multicentre trial compared to a normal population, but initially excluding polyps >3 cm in diameter which also is the criticism of the National Polyp Study of USA [24]. The latter was a short-term surveillance program in contrast to the former [23], but also demonstrated a reduced incidence of CRC compared to a normal population, supporting the value of surveillance.

A systemic review of the literature will be less than optimal because of considerable differences in age at inclusion, type of adenomas, duration of surveillance, and lack of information about incidence of CRC and mortality. However, new population trials performed over decades are unrealistic. The present results suggest in spite of limitations that adenoma surveillance may be of value in countries with high risk of CRC, and it becomes increasingly important, when screening of the general population above 50 years of age is recommended. The efficacy of screening for CRC with fecal occult blood tests has been proven in several prospective randomised trials [25]. The present results were obtained in a population where screening average risk persons played a substantial role during 17 of the 24 years of study, one third of the adenoma patients being recruited because of occult faecal bleeding. Colonoscopy has to be used, but development of more specific markers for risk of CRC may reduce the large number of colonoscopies, reducing the number of side effects, which might prove fatal in patients with complicating diseases.