Prevalence of Colorectal Cancer and Advanced Adenoma in Patients with Acute Diverticulitis: Implications for Follow-Up Colonoscopy

Abstract

Background and Aims:

Guidelines recommend colonoscopy after an episode of diverticulitis to exclude neoplasia but the effectiveness of testing is uncertain. Patients with complicated diverticulitis may be at higher risk for neoplasia, but most patients have uncomplicated disease. We examined the incidence of colorectal cancer (CRC) and advanced adenoma (AA) in patients with diverticulitis compared with subjects undergoing screening colonoscopy.

Methods:

Computed tomography (CT) scans from January 1, 2008 to May 1, 2013 at the University of Pittsburgh Medical Center (UPMC) were reviewed to identify subjects with confirmed acute diverticulitis. Subsequent surgical, colonoscopy, and pathology reports were abstracted for AA and CRC diagnoses. Neoplasia incidence was compared with that reported for screening colonoscopy from a meta-analysis (N=68,324), and from colonoscopy exams at UPMC between 2013 and 2015 (N=28,573).

Results:

A total of 5167 abdominal/pelvic CT scan reports identified 978 patients with acute diverticulitis, among which 474 (48.5%) subjects had at least one colonoscopy or gastrointestinal surgery through April 2015. The CRC rate in patients with diverticulitis (13/474, 2.7%) was significantly higher (P<0.0001) compared with both the meta-analysis (0.8%) and UPMC (0.3%). The AA rate (19/474, 4.0%) was similar to the rate in the meta-analysis (5.0%, P=0.39) but significantly lower than at UPMC (7.7%, P=0.003). The incidence of AA or CRC in complicated diverticulitis (10/141, 7.1%) did not differ significantly (P=0.85) from the incidence of AA or CRC in uncomplicated diverticulitis (22/332, 6.6%).

Conclusions:

CRC after diverticulitis was significantly higher than that observed at screening colonoscopy and was not limited to complicated disease. Colonoscopy is advisable after the diagnosis of diverticulitis.

Introduction

Colonic diverticulosis is a common disease, defined by the presence of sac-like outpouchings of the colonic wall. Prevalence increases with age, and diverticula are found in approximately 65% of the population after age 80.¹ Diverticulitis, an inflammation and/or infection of a diverticulum, is the most common adverse event, occurring in anywhere between 5% and 25% of patients with diverticula.^{2, 3} In the modern era of an aging population, the incidence of diverticulitis is likely to rise. Nationwide inpatient analysis shows a 26% increase in rates of hospitalization for acute diverticulitis in the United States between 1998 and 2005.⁴

Computed tomography (CT) is the diagnostic modality of choice for acute diverticulitis,^{5, 6} with a sensitivity of 99%.^{5, 7} However, the absence of a mass lesion on CT scan does not exclude an underlying colonic neoplasm⁸ and CT findings in acute diverticulitis, such as thickening of the colonic wall, can be found in colorectal carcinoma (CRC).⁹ The incidence of colorectal neoplasm varies between 0.2%¹⁰ and 7.4%¹¹ of patients with acute diverticulitis.

Clinical guidelines recommend that colonoscopy be performed in patients with acute diverticulitis, especially if it is the first episode or no colonoscopy has been previously performed,^{8, 12} but studies have reached heterogenous conclusions.^{9, 13–16} The presence of ‘alarm’ symptoms such as anemia, rectal bleeding, unintended weight loss, abdominal pain, or a change in bowel habits, further strengthens the recommendation for colonoscopy after diverticulitis.¹⁷

The aim of this investigation is to examine the incidence of neoplasia, including advanced adenoma (AA) and CRC in subjects after diagnosis of diverticulitis and to compare the rates with subjects undergoing screening colonoscopy.

Methods

We performed a retrospective analysis of patients diagnosed with CT-confirmed acute diverticulitis from January 2008 to May 2013 at 2 University of Pittsburgh Medical Center (UPMC) hospitals (UPMC Presbyterian and Shadyside). Patients were identified through search of an electronic medical record data repository that contains full-text medical records and integrates information from central transcription, laboratory, pharmacy, finance, administrative, and other departmental databases.¹⁸ To maintain patient confidentiality, all data were de-identified using an honest broker system. Criteria were met for exemption from informed consent, and the protocol was approved by the University of Pittsburgh’s Institutional Review Board. Subjects were passively followed via electronic medical records through April 2015.

For the period January 2008 to May 2013, there were 59,817 CT scan reports that were potentially consistent with a radiographic diagnosis of diverticulitis (ie, they contained the word “diverticulitis”). Additional qualifying phrases were applied to the search algorithm to exclude reports such as those including the phrase “without diverticulitis” and “no evidence of diverticulitis” leaving a total of 5167 reports for initial review. These reports were manually reviewed by a single reviewer. CT scan reports were extracted for type of contrast used (oral and/or intravenous), location of diverticulitis, the presence of colonic wall thickening, abscess, fistula, obstruction, colonic perforation, lymphadenopathy, colonic mass, or radiologist suspicion of cancer. A total of 1095 CT scans were identified as consistent with a diagnosis of diverticulitis representing 978 individual patients. For subjects with confirmed acute diverticulitis (N = 978), subsequent gastrointestinal surgical, colonoscopy, and pathology reports were retrieved and evaluated for the diagnosis of AA and CRC. AA was defined as any adenoma ≥10 mm, or with histologic findings of either high-grade dysplasia, or tubulovillous or villous histology. All subjects included in the cohort of 978 subjects were followed for at least two years.

Colonoscopy after acute diverticulitis in the literature

We conducted a search in both PubMed and the Cochrane library on February 11, 2018, using the following terms, “diverticulitis AND colonoscopy” and “diverticulitis AND colorectal cancer” and searched the bibliographies of studies for additional relevant investigations. English articles without time limitation were reviewed.

Screening Population for Comparison

The incidence of AA and CRC in the diverticulitis cohort was compared with the incidence observed in historical controls, using data from a meta-analysis of 68,324 patients undergoing screening colonoscopy,¹⁹ including 10 cohorts in a 12-year period (1994–2006). This meta-analysis reported an AA rate of 5% (95% Cl, 4%–6%) and CRC rate of 0.8% (95% Cl, 0.13%–2.97%). We also compared the AA and CRC incidence rates to a cohort undergoing screening colonoscopy at UPMC in western Pennsylvania (14 hospitals) from 2013 to 2015 (N = 28,573).²⁰ In this cohort, only endoscopists performing at least 200 colonoscopy examinations over 2 years were included to exclude part-time physicians or those for whom colonoscopy was a small component of clinical practice. In that sample, the AA rate was 7.7% and the CRC rate was 0.3%.

Statistical analyses

Differences between groups were assessed using t-tests or Wilcoxon-Mann-Whitney tests for continuous variables and Chi-square tests or the Fisher exact tests for categorical variables. All significance tests were two-sided; P values < 0.05 were considered significant. All analyses were performed using the SAS statistical software package (SAS version 9.4, SAS Institute Inc, Cary, NC, USA).

Results

Study Cohort

Review of the 5167 retrieved abdominal/pelvic CT scan reports confirmed the diagnosis of acute diverticulitis in 978 unique patients (Table 1). Mean age of these patients was 62.0 years, 75.2% were white, 45.5% were male, and 75.9% were diagnosed with uncomplicated disease, with the sigmoid colon the most commonly affected location (Table 1). Among the 978 patients, 474 subjects (48.5%) had at least one colonoscopy or gastrointestinal surgery through follow-up to April 30, 2015 (Figure 1). Median time between diagnosis and first procedure after diagnosis (either colonoscopy or surgery) was 116.0 days (range: 0–2407 days). Five hundred four subjects had no record of colonoscopy or gastrointestinal surgery in the UPMC system. Patients without a follow-up procedure after diagnosis were significantly older than those who had at least one colonoscopy or surgery (mean age ± SD, 63.0 ± 15.2 vs 61.0 ± 13.0, P = 0.03), were more likely to have had uncomplicated diverticulitis (81.4% vs 70.0%, P < 0.0001), and were less likely to have had a colonoscopy before the diagnosis of diverticulitis (30.4% vs 44.1%, P < 0.0001). No significant differences between the 2 groups were observed for sex and race.

Table 1:

Characteristics of the Diverticulitis Cohort (N = 978)

Age (year), Mean (±SD)	62.0 (±14.2)
Sex, N (%)
Male	445 (45.5)
Female	533 (54.5)
Race, N (%)
White	735 (75.2)
Black	168 (17.2)
Other	9 (0.9)
Unknown	66 (6.7)
Diverticulitis type, N (%)
Uncomplicated	742 (75.9)
Complicated1	235 (24.0)
Unknown	1 (0.1)
Diverticulitis location, N (%)
Ascending colon	39 (4.0)
Ascending/transverse colon	3 (0.3)
Transverse colon	12 (1.2)
Transverse/Descending colon	4 (0.4)
Descending colon	124 (12.7)
Descending/sigmoid colon	65 (6.6)
Sigmoid colon	730 (74.6)
Unknown	1 (0.1)
Colonic wall thickening, N (%)
Yes	582 (59.5)
No/Not mentioned	396 (40.5)
Abscess, N (%)
Yes	148 (15.1)
No/Not mentioned	830 (84.9)
Fistula, N (%)
Yes	22 (2.3)
No/Not mentioned	956 (97.7)
Obstruction, N (%)
Yes	7 (0.7)
No/Not mentioned	971 (99.3)
Perforation, N (%)
Yes	115 (11.8)
No/Not mentioned	863 (88.2)
Lymphadenopathy, N (%)
Yes	53 (5.4)
No/Not mentioned	925 (94.6)
Mass, N (%)
Yes	21 (2.2)
No/Not mentioned	957 (97.8)
Suspicion of cancer, N (%)
Yes	32 (3.3)
No/Not mentioned	946 (96.7)

^1.Complicated diverticulitis is defined as diverticulitis associated with abscess, fistula, bleeding, obstruction, or perforation.

Figure 1:

Study flowchart

Incidence of Neoplasia

Table 2 lists the incidence rates observed for AA and CRC in our study cohort of 474 subjects with follow-up colonoscopy or surgery. Nineteen subjects were diagnosed with AA (4.0%) and 13 subjects with CRC (2.7%). Median time between diverticulitis diagnosis and detection of neoplasia was 213.5 days (range: 0–1552 days). The AA rate observed in our cohort was similar to the rate observed in historical data from a meta-analysis of patients undergoing screening colonoscopy(P = 0.39), but the CRC rate was significantly higher (P < 0.0001) (Table 2). Compared with subjects undergoing screening colonoscopy at UPMC from 2013 to 2015, the rate observed for AA was significantly lower whereas the rate for CRC was significantly higher (P = 0.003 and P < 0.0001, respectively) (Table 2). There was no significant difference in the cumulative rate of neoplasia (AA + CRC) compared with either the meta-analysis or to UPMC from 2013 to 2015. Ten subjects with AA or CRC in our cohort (31.3%) had complicated diverticulitis; 131 subjects without AA or CRC (29.6%) had complicated diverticulitis. The incidence of AA or CRC in patients with complicated diverticulitis (10/141, 7.1%) did not differ significantly (P = 0.85) from the incidence of AA or CRC in patients with uncomplicated diverticulitis (22/332, 6.6%).

Table 2:

Advanced Adenoma and Colorectal Cancer Incidence after Diverticulitis Compared with Screening Colonoscopy

	Current Sample N (%)	Meta-analysis19 N = 68,324 %	P value (Current vs meta-analysis)	UPMC20 N = 28,573 %	P value (Current vs UPMC)
Advanced adenoma (AA)	19/474 (4.0)	5.0	0.39	7.7	0.003
Colorectal cancer (CRC)	13/474 (2.7)	0.8	< 0.0001	0.3	< 0.0001
Cumulative (AA + CRC)	32/474 (6.8)	5.8	0.38	8.0	0.32

Because all included subjects had at least 2 years of follow-up and because differences in the incidence of CRC could be affected by remote development of neoplasia, we repeated the analyses limiting our cohort to patients who had at least 1 colonoscopy or surgery within 2 years of diagnosis. In total, 369 subjects (77.8%) had a colonoscopy or gastrointestinal surgery within 2 years of diagnosis. In this subset, 13 subjects (3.5%) were diagnosed with AA and 12 subjects (3.3%) with CRC. Median time between diverticulitis diagnosis and detection of neoplasia was 130 days (range: 0–668 days). The AA rate observed in this subset was similar to the rate observed in the historical data from the meta-analysis (P = 0.19) and significantly lower than the rate observed in the UPMC screening colonoscopy cohort (P = 0.003). The rate of CRC was higher than the rate observed in both the meta-analysis and in the UPMC screening colonoscopy cohort (P < 0.0001). Seven subjects with AA or CRC in this subset (28.0%) had complicated diverticulitis; 119 subjects without AA or CRC (35.6%) had complicated diverticulitis. The incidence of AA or CRC in patients with complicated diverticulitis (7/126, 5.6%) did not differ significantly (P = 0.66) from the incidence of AA or CRC in patients with uncomplicated diverticulitis (18/243, 7.4%).

In patients with CRC, all of the cases of diverticulitis were in the sigmoid (12/13) or in the descending colon (1/13). In 9 out of 13 cases (69.2%), the diverticulitis was diagnosed in the same location as the tumor. In 2 cases, the tumors were in the right side of the colon and in 2 cases in the rectum.

Discussion

Results from this study suggest that colonoscopy is advisable after the diagnosis of diverticulitis as significantly more CRC was detected after diverticulitis than among subjects undergoing screening colonoscopy. Moreover, patients with uncomplicated diverticulitis had a similar incidence of advanced neoplasia as those with complicated diverticulitis, hence our findings support recommending colonoscopy after both complicated and uncomplicated diverticulitis.

It is well appreciated that CT scans can fail to detect CRC in the setting of diverticulitis.²¹ In fact, the performance of colonoscopy after the diagnosis of diverticulitis emanates from the concern that colon cancer can simulate or be obscured by the diagnosis of diverticulitis on CT scan.⁹

Although multiple studies have investigated the role and yield of colonic evaluation after acute diverticulitis,^{9, 13–16} heterogenous conclusions have resulted. In a systematic review of 10 studies and 771 patients, the prevalence of CRC in diverticulitis was 2.1% versus an estimated U.S. SEER rate of 0.68%. The authors concluded, however, that the cancer incidence in diverticulitis was overestimated due to selection bias, with higher-risk subjects more likely to undergo further evaluation.⁹ In a meta-analysis of 11 studies in 7 countries and 1970 total patients, no increased risk of cancer was observed in 1497 patients with uncomplicated diverticulitis, whereas an increased risk was observed in 79 patients with complicated disease.¹⁶ Similarly, a meta-analysis of uncomplicated diverticulitis in 9 studies and 2490 patients of whom 1468 had a follow-up colonoscopy examination, found a prevalence of cancer of 1.16% and AA of 2.2% and concluded that routine follow-up colonoscopy in uncomplicated diverticulitis was not needed.¹⁴ A systematic review in nearly 1800 patients with left-sided diverticulitis reported a CRC rate of 1.5%. They calculated that 122 colonoscopy examinations in patients with left-sided diverticulitis were needed to detect one CRC.¹³ And, in the most recent and comprehensive meta-analysis including 31 studies and 50,445 patients, the pooled CRC rate was 1.9%.¹⁵ Although lower than the CRC rate of 2.7% in our cohort, this rate is still substantially higher than we observed in our control group populations undergoing screening colonoscopy. They identified a substantially higher rate of CRC in complicated diverticulitis (7.9%) compared with uncomplicated diverticulitis (1.3%) and concluded that colonoscopy should be considered after acute diverticulitis.¹⁵ Again, even the rate observed in uncomplicated diverticulitis (1.3%) is higher than the cancer rate we observed at screening colonoscopy.

Recent guidelines do recommend colonoscopy after diverticulitis.⁸ The reason for heterogeneity among the many studies is multifactorial, due to different lengths of follow-up time, different populations under study (complicated, uncomplicated, or left-sided only diverticulitis), different methods of colonic evaluation (colonoscopy vs registry follow-up), differences in the percentage of patients who are lost to follow-up, and different control populations for comparison.

In Table 3, we summarize studies since 2014 evaluating the relationship of CRC and acute diverticulitis. Again, heterogeneous study designs and differing cohorts and methodologies for study reflect the difficulty in arriving at a uniform conclusion. A large study from the Danish National Registry²² identified 1723 patients (4.3%) with CRC and a diagnosis of diverticulitis, compared with a rate of 2.3% CRC in a matched control group, for an incidence rate ratio of 1.86 (P < 0.001). But the case group included CRC cases diagnosed before the onset of diverticulitis, and hence their conclusion that diverticulitis is associated with CRC, does not directly inform the question of whether a patient diagnosed with diverticulitis needs a subsequent evaluation. Using a national insurance administrative database in Taiwan, Huang et al²³ reported a significantly higher risk of CRC with diverticulitis, adjusted for age, sex, and comorbidities (adjusted hazard ratio: 4.2, P < 0.0001), but the results were based on an administrative data set, with limited clinical detail. In contrast, smaller case-control studies show no relationship between CRC and diverticulitis,^{10, 24, 25} whereas a study from Switzerland in left-sided disease²⁶ and a small study from Korea¹¹ demonstrate otherwise. These studies vary in eligibility criteria, length of follow-up, and in the control groups used for comparison.

Table 3:

Recent Case-Control Studies of the Risk of CRC after Diverticulitis

	Diverticulitis				Controls				Statistical Comparison
	N	Age	CRC, n (%) or rate/PYO	Comments	N	Age	CRC, n (%)	Characteristics
Mortenson* 2017, Denmark22	40,496	NR	1723 (4.3)	Lifetime risk: includes CRC before diagnosis of diverticulitis	404,960	NR	9127 (2.3)	Danish National Registry	IRR: 1.86 (95% CI, 1.8–2.0, P<0.001)
Daniels 2015, Netherlands25	401	57	5 (1.2)	Uncomplicated left-sided diverticulitis	1,426	60	9 (0.6)	Screening colonoscopy	OR: 1.3 (95% CI, 0.4–4.4, P=0.7)
Meyer* 2015, Switzerland26	506	67	11 (2.1)	Left-sided diverticulitis	NR	NR	Expected N = 0.25 cancers	Geneva population-based cancer registry	SIR: 44 (95% CI, 23–76, P<0.001)
Alexandersson 2014, Iceland24	282	58	2 (0.7)	Uncomplicated diverticulitis	NR	NR	Expected N = 0.83 cancers	Icelandic Cancer Registry	SIR: 2.4 (95% CI, 0.3–8.7, P=NS)
Choi* 2014, Korea11	149	49	11 (7.4)		298	NR	2 (0.7)	Screening colonoscopy	OR: 11.8 (95% CI, 2.6–53.9, P=0.001)
Huang* 2014, Taiwan23	28,909	56	695 (49/10K PYO)		182,488	56	1305 (15/10K PYO)	National Insurance Database	HR: 4.2 (95% CI, 3.8–4.6, P<0.0001)
Lecleire 2014, France10	404	61	1 (0.2)	Hospitalized diverticulitis	404	61	1 (0.25)	Screening colonoscopy	P=NS

Cl: Confidence interval; CRC: Colorectal cancer; HR: Hazard ratio; IRR: Incidence rate ratio; N: Number; NR: Not reported; NS: Not significant; OR: Odds ratio; PYO: Person-years observed; SIR: Standardized incidence ratio;

^*:Statistically significant

Some studies have reported the incidence of CRC after excluding patients who underwent surgery.^{17, 21, 27–35} Complicated diverticulitis carries a significant propensity for surgical intervention. Excluding surgical cases would reduce the true incidence of CRC associated with diverticulitis. We included colorectal cancers in patients who underwent colonoscopy or surgery because the clinical question we are addressing is whether a patient with a radiologic diagnosis of diverticulitis should undergo a colonoscopy. To answer that question, all cases of CRC identified subsequent to the diagnosis of diverticulitis should be included. Our data show that although the majority of cancers (69.2%) were detected in the same segment interpreted as harboring diverticulitis, cancers can also be detected distally and proximally to the segment of diverticulitis. Moreover, our data demonstrate the presence of CRC in both uncomplicated and complicated diverticulitis.

Some studies have identified CRC cases through cancer registries to decrease the likelihood of underreporting of the CRC rate in subjects without colonoscopy during the study period.^{22–24, 26, 36} We followed subjects in the medical record for up to seven years after diagnosis to search for additional subsequent cancers.

We did not observe a higher rate of AA in subjects after diverticulitis compared with those undergoing screening colonoscopy in the meta-analysis, and we saw a lower prevalence of advanced adenoma in subjects with diverticulitis compared with those undergoing screening at UPMC from 2013 to 2015. However, the higher detection rate for advanced adenoma in more recent years compared with that from 2008 when our study began, may be attributable to the overall increase in adenoma detection that has accompanied attention to adenoma detection rates.³⁷ Our significant differences were limited to higher rates of CRC in patients with diverticulitis.

There are several strengths to the current investigation. We comprehensively reviewed an unselected population of abdominal CT scan reports to identify patients with incident acute diverticulitis. Hence, our sample is a representative population of confirmed diverticulitis. Moreover, we used an extensive longitudinal medical record system to track for evidence of subsequent AA or CRC. We compared findings in the postdiverticulitis patients with 2 different control populations, a large meta-analysis of over 64,000 patients, and a UPMC population undergoing screening colonoscopy. The UPMC population is particularly noteworthy because it sampled all subjects undergoing screening colonoscopy over a 2-year period in 14 hospitals in UPMC, all in western Pennsylvania, an appropriate reference population to our patients with diverticulitis.

Limitations include that this is a retrospective study of medical records at a single institution. As a quaternary care center, it is possible that our patient population represents more complicated cases referred for a higher level of care. A disproportionate number of complicated patients might have inflated the finding of CRC, in comparison to a general population. However, our rate of complicated diverticulitis is consistent with other studies,^{26, 29, 30} and we saw no difference in complicated diverticulitis versus uncomplicated in rates of colonic advanced neoplasia. A substantial portion of our sample did not have a follow-up procedure; thus, their incidence of neoplasia was unknown. It is possible they were diagnosed with disease elsewhere or alternatively, were lower risk, and had they been tested, would have lowered our CRC estimates. Those without a follow-up procedure were similar in gender compared with those who did undergo testing, and if anything, they might have been at an even higher risk of CRC compared with the group with follow-up, being older, and less likely to have had a previous colonoscopy. We did not consider the patients’ clinical findings, comorbidities, prior history of colorectal surgery or acute diverticulitis, or a family history of CRC, factors that could have affected the incidence of CRC in our cohort. We included follow-up for up to 7 years after diagnosis, and it is possible that some of the CRCs observed developed well after the diagnosis of diverticulitis. However, even when truncating the follow-up to within 2 years of diagnosis, the results were similar.

In conclusion, the rate of CRC after diverticulitis was significantly higher than the rate observed at screening colonoscopy and was similar in complicated versus uncomplicated diverticular disease. Colonoscopy is advisable after the diagnosis of diverticulitis.

Abbreviations:

AA

advanced adenoma

CRC

colorectal cancer

CT

computed tomography