INTRODUCTION

Diverticular disease is highly prevalent in developed countries. The incidence increases with age and is uncommon among individuals younger than 40 years and the incidence is 50% among those older than 80 years of age. Approximately 25% of patients with diverticulosis will develop acute diverticulitis (AD), which may result in complications such as abscesses, fistulae or even colon perforation ^1,2. In view of its frequency and potential complications, both diverticulosis and diverticulitis now represent highly relevant clinical conditions in developed countries. Ongoing updates are required for its management based on the best evidence available ³.

There has been a long-standing debate on whether AD should be considered a clinical condition with a risk of advanced colonic neoplasm (ACN). Clinical guidelines ^4,5,6,7 recommend routine colonoscopy after a radiographically confirmed episode of AD in order to rule out the presence of advanced adenoma (AA) or colorectal cancer (CRC). This is due to the fact that the literature suggests that these lesions are more prevalent among patients with AD than in the general population and also represents a high risk for colorectal cancer that is undetected by imaging tests. However, this recommendation is based on evidence of a low quality. As it is based on older retrospective study reviews that have important selection biases and highly variable review designs.

CT is presently considered as the best imaging technique to accurately diagnose AD, with a sensitivity and specificity of 94% and 99%, respectively ⁸. Furthermore, it is an important tool to assess whether surgical management is required. However, retrospective studies report conflicting data with regard to the ability of CT to accurately identify ACN in the diverticulitis area. The AGA based their recommendations on the systematic review by Daniels ⁹, as it best described and delimited the biases present in published studies. However, the risk of an incorrect diagnosis with CT was high. In contrast, a recent study ¹⁰ prospectively compared CT and colonoscopy in the setting of AD. This study found a very high correlation between both techniques (kappa 0.829, p < 0.01, 95% CI), which suggests the futility of endoscopy after a CT scan that reveals no evidence of a malignancy.

On the other hand, the prevalence of ACN in recent systematic reviews is higher in patients diagnosed with acute complicated diverticulitis (ACD) (defined by the presence of abscess, perforation, fistula or stricture) than for those diagnosed with acute uncomplicated diverticulitis (AUD) using CT. This adds controversy to the suitability of colonoscopy for patients with AUD. Sharma ¹¹ performed a meta-analysis of eleven studies that included a total of 1,970 patients diagnosed with AD using CT. CRC was diagnosed in 1.6% of colonoscopies, 0.7% for AUD and 10.8% for ACD. De Vries ¹² performed a systematic review of a total of 2,490 patients with AUD, and CRC was diagnosed in 1.1% of cases and AA in 2.2% of patients. In the study by Daniels ³, there was an estimated 5% prevalence of ACN amongst a total of 1,976 patients with AUD, 1.5% for CRC and 3.8% for AA. Most recently reported retrospective studies ^13,14 seem to corroborate the data from these systematic reviews. Hence, the low prevalence of ACN in AUD is similar to that of the screened general population ¹⁵. This calls into question the present recommendation of a routine colonoscopy, that is reserved for patients where a CT scan cannot definitively exclude ACN and for those older than 50 years with no recent prior colonoscopy or with a prior low-quality colonoscopic procedure. In contrast, if the prevalence of ACN in AUD identified by CT is higher than that of the general population, routine colonoscopy would be recommended.

Importantly, the associated drawbacks of a colonoscopy for AD must be borne in mind. In the setting of acute inflammation, it may result in complications such as bleeding or perforation and should be considered as unsafe. Furthermore, inflammation may lead to a suboptimal colonoscopy preparation, reduce the chances of reaching the cecum and raise qualms for the endoscopist to excise the lesions. All of which will reduce endoscopy quality and increase the odds of a repeat procedure ^16,17. The add-on costs of unindicated endoscopy are also significant.

Thus, the goal of the present study was to determine the prevalence of ACN in AD and the factors that influence colonoscopy quality and safety. The aim was to identify which patients will benefit from the procedure and the best conditions to perform it.

METHODS

A retrospective observational study was performed. The study sample was collected from the database of the Digestive Endoscopy Unit, Hospital del Mar, Barcelona, Spain. All patients that underwent colonoscopy from 2005 to 2013 following a CT-based diagnosis of acute diverticulitis, where an endoscopy was indicated to assess a potential underlying CAN, were enrolled in the study. Patients that underwent surgery due to an unfavorable course of their inflammatory condition were not excluded, as a pathology study of the surgical sample was available. Only cases that underwent an endoscopy outside the two-year period following the CT-based diagnosis of AD were excluded.

RESULTS

Two hundred and sixteen patients were included in the study: 103 (48%) were female and 113 (52%) male, with a median age of 59 ± 15.2 years. Of these, 158 (73%) had AUD and 58 (27%) had ACD; 31 (14%) with an abscess, 27 (12%) with intra-abdominal free air and two (1%) had an ileocolic fistula. A total of 18 patients (8%) had experienced a previous episode of AD. Only ten (5%) patients had undergone a high-quality endoscopic procedure during the previous ten years. AD was predominantly located in the sigmoid colon (197, 91%) and also in the descending colon (13, 6%), cecum (four, 2%) and ascending colon (two, 1%). No differences were found between the ACD and AUD subgroups for any variable, except for surgery due to an unfavorable clinical course. Fifteen (25%) surgeries were performed in the ACD group and 0 (0%) in the AUD group, p = 0.001. There was a total of 13 sigmoidectomies and two non-resective surgeries; one simple suture with a cavity lavage and one debridement and Penrose drain (Table 1).

Table 1 Population characteristics. Subgroup comparison between complicated and uncomplicated acute diverticulitis 

ACD: acute complicated diverticulitis; AUD: acute uncomplicated diverticulitis; AD: acute diverticulitis; LGIE: lower gastrointestinal endoscopy. Quality endoscopy was defined as an endoscopy that meets the requirements set forth by the Asociación Española de Gastroenterología (15). Clinical outcome is ok when no surgery was required.

A total of 106 lesions were identified in 44 (20.3%) patients via colonoscopy. These included 45 non-advanced adenomas in 25 patients (17.1% [AUD/ACD: 17.9/15%, p = 0.3]) and 15 ACNs in 12 patients (5.6% [AUD/ACD: 3.2/11.7%, p = 0.02]). CRC was identified in two cases (0.9% [AUD/ACD: 0/3.4%, p = 0.07]) and AAs in ten patients (4.5% [AUD/ACD: 3.2/8.6%, p = 0.1]). Among non-adematous lesions, 44 hyperplastic polyps and two sessile serrated polyps were diagnosed. All were < 1 cm, proximally located and none were associated with the segment affected by AD (Fig. 1).

Fig. 1 Lesions identified by colonoscopy (HP: hyperplastic polyp; SSP: sessile serrated polyp; NAA: non-advanced adenoma; AA: advanced adenoma; CRC: colorectal cancer; ACN: advanced colonic neoplasm). 

Of the 15 ACNs, six were located in the sigmoid, all in the area of AD; two T4 stage CRC of 5.8 cm and 6.1 cm long, respectively and four AAs with a mean size of 16 mm. Four ACNs were found in the descending colon with a mean size of 11 mm; only one was located in an AD-affected area. The remaining ACNs were identified in the transverse and right colon, none were associated with AD areas and their mean size was 13 mm. With regard to the seven cases of ACN found in association with AD, CT only detected the presence of AD and did not identify the additional lesion.

Age > 50 years (0% vs 7.8%, p = 0.02) and the presence of ACD (3.2 vs 11.7%, p = 0.02) were the sole factors associated with an increased risk of ACN. No lesions were found in the surgical specimen or ACN during the subsequent endoscopic procedure in patients that underwent surgery due to an unfavorable clinical course.

A total of 143 (66.2%) patients met the quality criteria; ACD/AUD: 37 (61.7%)/106 (67.9%), p = 0.4. A complete colonoscopy was performed in 190 cases (88%), the preparation was effective in 162 cases (75%) and all lesions < 2 cm were excised during 32 of 41 colonoscopies with polyps (78%). The quality improved progressively (Fig. 2), the time elapsed from AD diagnosis and ROC curve analysis revealed that the optimal cut-off for endoscopy quality was 3.7 weeks for reaching the cecum (p = 0.001) and 9.5 weeks for preparation effectiveness (p = 0.012). Of the 73 (33.8%) low-quality endoscopies, only 14 (19.1%) were repeated. All repeat procedures were complete (100% of patients) and effective preparation was achieved in 40% of patients. The diagnostic yield was low, and only one AA, three non-advanced adenoma (NAAs) and no CRC were identified.

Fig. 2 Time from diagnosis with acute diverticulitis using CT to lower GI endoscopy. 

Early endoscopic procedures were performed in 18 patients (8.3%); seven (12%) with ACD (five with an abscess, one with free air, one with an ileocolic fistula) and eleven (6.9%) with AUD. No major complications were reported after colonoscopy, regardless of whether they were carried out early or late.

DISCUSSION

The overall prevalence of ACN in our population was similar to that reported by other retrospective series. With regard to the assessment of findings according to AD type, our results are also similar to those of recently reported reviews ^9,13,14. They also support the recommendation of performing routine endoscopies in patients with ACD, as ACN is significantly more common in association with AD than with AUD. Therefore, according to our series, the AUD group should be considered as low risk for ACN and routine endoscopy would have an uncertain indication. However, we must bear in mind the fact that this is based on retrospective data. Our population has a selection bias as it includes patients diagnosed with AD who never underwent an endoscopy and were therefore not included in the study. This may overestimate the prevalence of ACN. On the other hand, colonoscopy quality was low, which may underestimate the prevalence. When the findings of the study were analyzed according to age (0% ACN in patients < 50 years), colonoscopy was not cost-effective in younger patients. These patients have a low risk of ACN, although this must be ascertained in further prospective studies.

The fact that ACN associated with AD was identified in seven patients via colonoscopy but was not identified by CT is an important observation. The fact that the images were obtained in an emergency room with a non-multi-detector CT scanner and not reviewed by a radiologist with expertise in abdominal conditions may have played a role. The jury remains out with regard to the diagnostic accuracy of CT for ACN and this is partly why the clinical guidelines still recommend colonoscopy for patients with AD. There is no doubt that prospective studies using standardized multi-detector CT imaging and assessment by a specialist radiologist will narrow down the indication of colonoscopy for AD in the near future.

A vitally important and scarcely reported matter in the literature is the quality of endoscopy in the setting of AD. The presence of a large bowel inflammatory condition may be a presumed limiting factor for endoscopy quality. This was observed in our population, where only 143 (66.2%) patients met endoscopy quality criteria. This is significantly lower than that obtained from screening colonoscopies (> 95%) in our hospital from 2010 onwards. With regard to colon preparation, the low quality observed may result from an altered intestinal motility and laxative intolerance in an inflamed colon. However, other factors such as the use of an unfractionated administration in our patients or prolonged fasting (which could not be assessed retrospectively) may also have played a role. With regard to the completeness of the endoscopy, the low quality may be related to the presence of acute inflammatory strictures, pain and the difficulty in advancing the endoscope through an inflamed area amongst other factors. The failure to excise lesions might be associated with the presence of large bowel inflammation. The mean age was 59 years, which was similar to that reported by Daniels, very little quality data were included but the overall quality was deemed low for endoscopies in patients with AD.

Obviously, a retrospective study cannot appropriately assess all these factors. However, the time from endoscopy indication to endoscopy completion was an analyzable variable in our population and may play a relevant role in endoscopy quality in the clinical setting of colonic inflammatory conditions such as AD. Presumably, the farther away from the site of the inflammation, the fewer limitations there are for endoscopy quality. In our study population, the association between time and quality was confirmed. A waiting period of one month since the diagnosis is advisable in order to achieve a complete colonoscopy. A two month waiting period is recommended in order to achieve an optimal preparation. Hence, we conclude that at least two months should elapse from diagnosis to endoscopy for AD. If this recommendation is validated in a prospective analysis, it would be primarily directed to low-risk patients with no ACN-suspicious lesions on CT. Obviously, high-risk patients or patients with a radiographically suspected ACN should undergo the procedure without delay; quality is less relevant in these cases.

Finally, another important aspect in the context of AD is the safety of the endoscopy in the presence of colonic inflammation and potential complications. Early colonoscopy has been reported to be safe for AD in the absence of intra-abdominal free air ¹⁹. In our study, the number of early endoscopy procedures within two weeks of diagnosis was low (8.3%). However, no complications were seen both in early and later endoscopic examinations. Therefore, we conclude that colonoscopy for AD is safe, even when performed early in the course of the disease. While there were no complications in patients with intra-abdominal free air, even with early endoscopies, the small number of patients assessed prevents us from drawing solid conclusions.

To conclude, we may say that the number of ACNs identified during colonoscopies after an AD episode is non-negligible in our setting. The colonoscopy yield was higher for ACD versus AUD in a proportion that renders routine colonoscopy advisable in ACD, whereas its recommendation remains unclear for AUD. Furthermore, the quality of endoscopy in this context is low and likely limited by a number of factors, although the time from diagnosis with AD to colonoscopy does influences quality. Waiting a minimum of two months from diagnosis to colonoscopy would be acceptable in low-risk patients with no suspicion of CAN in order to improve quality. Finally, colonoscopy in the AD setting seems a safe procedure that could be safely indicated early in cases of radiographically suspected ACN.

As a final reflection, we think that the present recommendations regarding endoscopy indication, quality and safety for AD are based on data from retrospective studies. While our results will help assess the suitability of a colonoscopy indication in our setting, further prospective studies are needed in less biased populations in order to assess the manifold factors that may play a role in the safety and quality of our endoscopic practice.