ACG Clinical Guideline: Epidemiology, Risk Factors, Patterns… : Official journal of the American College of Gastroenterology

The appearance of thickened bowel wall in CI usually results from subepithelial edema and/or hemorrhage and reflects the evolving degree of ischemia and the time between the ischemic insult and performance of CT (^{111, 112}). Balthazar et al. (¹¹¹) published a retrospective case series of 54 patients with CI. The study subdivided the CT findings into three morphologically distinct groups based upon the timing of the examination in relation to the evolving pathophysiology. A “wet” appearance was seen in 61% of cases and appeared as a loss of colonic haustra with varying degrees of pericolic streakiness; this correlated to the acute pathologic changes after reperfusion of ischemic bowel. A “dry” appearance was seen in 33% of cases and manifested as concentric and symmetric mild mural thickening with homogenous attenuation of the colon wall and minimal pericolic streakiness; this correlated to the progression of ischemic damage without reperfusion. The remaining 6% had circumferential intramural air consistent with pneumatosis coli and presumed colonic necrosis (¹¹¹). Of the overall cohort, 37% had moderate ascites and 24% had a “target sign” or “double halo” sign showing different attenuations of the layers of the bowel wall that corresponded to ischemia and reperfusion of the involved segment. This study attempted to characterize the most common findings in segments of colon affected by CI, but it lacked firm diagnostic criteria for CI, cohorts were relatively small, and there was no control group. Similar limitations are seen in the study of Romano et al. (¹¹²) that looked at patterns of CT signs in patients with early ischemia, late ischemia, and infarction. In this study, CI in 71 patients was classified based upon the phase of disease, and the authors described the “little rose” sign attributable to hyperdensity of the mucosa and edema of the submucosa as a sign of “early stage” CI, i.e., the “wet” appearance (¹¹²). Iacobellis et al. (¹¹⁹) retrospectively grouped CI in patients with pathologically proven disease and sequential CT imaging into acute (e.g., CT within 37 h of presentation), subacute (e.g., second CT between 37 h and 21 days), and chronic CI (e.g., third CT between 21 days and 2 months). When patients had no evidence of vascular occlusion, the acute phase was characterized by pericolic fluid (100%) and bowel wall thickening (28.1%); 86.4% had bowel wall thickening in the subacute phase. Bowel wall thickening was not frequently seen in patients with vascular obstruction in the acute or subacute phases (71.9%) and never in the chronic phase (¹¹⁹). None of these studies provided definitive criteria for diagnosis of CI using CT, although they offer some insights into potentially staging the ischemic process. Further investigation is required to characterize the most common signs of CI from each phase of the ischemic process and course of disease, to describe findings that reliably differentiate CI from other disease entities, and to determine what constellation of findings might be diagnostic of CI.

Adenocarcinoma of the colon is associated with CI and usually the segment of ischemic injury is proximal to the neoplasm. Tumoral segments can be differentiated from ischemic segments of colon by CT (^{120, 121}). Ko et al. (¹²²) looked at 20 patients who underwent CT, BE, and colonic resection because of adenocarcinoma of the colon with CI proximal to the tumor. On CT, the segments of CI had significantly less colon thickening (1.0 vs. 2.0 cm, P<0.05) and were longer (10.1 vs. 5.9 cm, P<0.05) than those with malignancy. In addition, although no statistical comparison was documented, the CI segments appeared to enhance more homogenously (70%), whereas the tumoral segments more frequently enhanced heterogeneously (60%) (¹²²). This study is flawed by its small number of patients, lack of delineation of other CI findings, and absence of a true control arm, but it still offers some insight into the differences on CT between malignancy and segmental CI. Given the association, although infrequent, between CI and distal malignancy (as well as other potentially obstructing lesions such as strictures and fecalomas), the colon should be screened for cancer with a colonoscopy several weeks following recovery from the ischemic insult, if the patient has not been screened recently. Screening colonoscopy should not be performed at the time of presentation with CI because a minimum of colon manipulation and distension is recommended to avoid further damage from the reduction in blood flow that occurs with colon insufflation.

The etiology of CI is usually not identified, but in patients with known risk factors, CT might provide significant support for the diagnosis. Byun et al. (¹²³) looked at 33 patients with lupus and acute abdominal pain: 79% of patients had evidence of ischemic bowel disease of whom 44% had ischemia involving the colon; the most frequent findings included engorgement of mesenteric vessels (79%), symmetric bowel wall thickening (74%), a target sign (67%), and multifocal bowel wall thickening (61.5%). Unfortunately, this study was small, did not compare its findings to a control arm, and lacked pathologic confirmation of the etiology of inflammation observed on imaging. Colonoscopy with biopsy remains the most accurate way to diagnose CI in patients with risk factors for disease, including vasculitides.

CI is usually nonocclusive in nature, but CT (or CTA) can be used to identify whether or not vascular occlusions are present. Unfortunately, the IMA is occluded in up to 10% of asymptomatic patients over 60 years of age and hence the significance of such a finding is uncertain (¹²⁴). With emerging technology, contrast-enhanced CT can sometimes assess patency of the celiac, SMA, and IMA without devoted CTA; this usually requires direct communication with the radiology team to review the imaging specifically with the goal of vascular assessment. Data do not exist on the sensitivity and specificity of CT for assessing vascular occlusion in CI, but a meta-analysis including three prospective and three retrospective studies that looked at CT in the diagnosis of mesenteric arterial embolism, mesenteric arterial thrombosis, and mesenteric venous thrombosis (i.e., primary vascular AMI) found the sensitivity and specificity for diagnosis to be 93.3% and 95.9%, respectively (¹¹³). These six studies were performed in tertiary care referral centers in industrialized countries and, therefore, their high diagnostic accuracy might not reflect the accuracy of studies done in small community hospitals or practices; regardless, CT (or CTA) offers a high sensitivity and specificity for vascular assessment in patients with CI.

In contrast to AMI in which conventional mesenteric angiography or CTA plays an essential role, vascular imaging studies are not indicated in most patients with suspected CI because by the time of presentation, colon blood flow has usually returned to normal and the observed changes are not from ongoing ischemia but rather reflect the ischemic insult with or without reperfusion injury. In patients with potentially severe, complicated disease, such as those with IRCI or when the clinician cannot distinguish between AMI and CI (e.g., in patients with severe abdominal pain and little or no bleeding), a dedicated multiphasic CT is indicated. In patients with IRCI, CTA should be performed either initially or before discharge, as the colonic episode may be the “heralding” indication of SMA disease and possibly impending AMI. Multiphasic CT scanning, including a CT without oral contrast but with intravenous contrast and attention to the angiographic arterial and portal venous phases, has become standard for the diagnosis of AMI (^{125, 126, 127}). One retrospective study that compared triple-phase CT scanning in 39 patients with pathologically confirmed AMI and 41 patients with suspected AMI who ultimately had another diagnosis showed that the specificity for detection of SMA thrombus or occlusion approached 100%, although the sensitivity was poor (26%). Similarly, for celiac and IMA thrombus, the specificity was estimated to be 98%, and the sensitivity was 5%. This study also showed that the diagnostic use of oral contrast during CT offered minimal additional benefit to the two vascular phases (¹¹⁴). Although this study contained small numbers of patients, it had a control group and characterized each phase of assessment, thus offering valuable data. In any patient in whom the presentation of CI may be a heralding sign of AMI (e.g., IRCI), and the multiphasic CT is negative for vascular occlusive disease, traditional splanchnic angiography should be considered for further assessment because a negative triple-phase CT does not absolutely exclude vascular obstruction as the cause. It is important to note that there are no available published data yet analyzing IRCI as a “heralding” indication of SMA disease. However, our experience has revealed that a select group of patients with IRCI may present again within weeks, months, or even a few years after the heralding presentation of IRCI with catastrophic episodes of AMI, resulting in loss of most of their small bowel or death. A limited but increasing number of patients with IRCI have undergone angiography and had occluding thrombi found in the SMA; stenting and/or thrombectomy resulted in good outcomes. Conversely, some patients with SMA thrombi did not undergo angiography or stenting yet had a positive outcome as well. Although there are no data to yet support our recommendation, we believe that the benefit of preventing a potentially catastrophic ischemic insult to the small intestine (AMI) far outweighs the risks of a CTA.

Pneumoperitoneum, portal venous gas, and pneumatosis are ominous findings in patients suspected of CI. In one Canadian study of 14 total patients who underwent colectomy with pathology showing transmural ischemia, CT showed that 100% had fat stranding, 71% had pneumatosis, and 36% had portal venous gas (¹²⁸). However, the presence of pneumatosis and portal venous gas does not necessarily indicate nonviable bowel. In one small study of 23 patients with either small or large bowel ischemia, pneumatosis and portomesenteric venous gas were associated with transmural bowel infarction in 78% and 81% of patients respectively (¹²⁹). When patients with an obstructing malignancy presented with a large bowel obstruction and CT showed cecal pneumatosis, surgery revealed transmural necrosis of the cecum in 26% of patients (6 of 23 total patients) (¹³⁰). Milone et al. (¹¹⁵) assessed CT findings in 121 patients with transmural colon necrosis and 87 with limited colon necrosis. The presence of pneumatosis plus portomesenteric gas showed good specificity (83%) for transmural infarction, but very low sensitivity (17%); the concomitant presence of the two findings did not achieve statistical significance in predicting transmural CI (¹¹⁵). These studies are limited by their sample size and retrospective nature, but they do show that patients with CI who also have CT findings of pneumatosis and/or portomesenteric venous gas are likely to have severe disease, but not necessarily with full transmural necrosis. Nonetheless, in the clinical setting of CI, one of these findings is still an indication for prompt operation. Although CT has become the standard of care in assessing patients with suspected CI, both US and MRI also offer some diagnostic utility. In one retrospective case series of 58 patients with CI evaluated by US, the findings were as follows: symmetric bowel wall thickening (100%), segmental involvement of the colon (80%), preservation of colon wall stratification (66%), altered pericolic fat (28%), free fluid (19%), and pneumatosis (1.7%). When this cohort was subdivided into transmural necrosis and nontransmural necrosis, the only significant difference observed was the presence of abnormal pericolonic fat, which was seen more frequently in patients with transmural disease (75% vs. 20%, P<0.01) (¹³¹). In another study of 66 patients presenting to the emergency department with nonspecific abdominal pain who had abdominal US performed as their primary imaging modality, the group was subdivided based upon diagnosis into infectious colitis, inflammatory colitis, ischemic colitis, and malignant tumors. When the US findings were compared, arterial flow in the wall of the involved segments of colon was absent in 43% of those with CI compared with 12% of those with other diagnoses (P<0.05) (¹³²). These US findings do not invalidate the reason why we do not recommend angiography in the evaluation of a first episode of CI, namely that by the time patients present with CI, colon blood flow has usually returned to normal, because they only measured signs of arterial flow in the bowel wall. This study was small and did not provide any criteria for the diagnosis of CI or information on the severity of the CI. Moreover, it is not known how many of the patients had infarcted bowel that, of course, would not have intramural flow. We would be hesitant to challenge this long-held belief until the results are confirmed in a robust manner. Experience with US in the setting of CI is very limited and it is believed that this technique lacks specificity for bowel wall thickening and has a high false negative rate (¹³³).

MRI has been formally studied in only a small number of patients with CI; findings are similar to those of CT, but without the radiation exposure and the need for iodinated contrast agents. In one prospective study that assessed seven patients with CI, all of whom had an initial diagnostic CT and a subsequent MRI, similar segmental involvement and wall thickness were shown on both modalities (¹³⁴). As with CT, such findings usually are not specific enough to make a definitive diagnosis. MRI may have a role when repeat imaging is required or patients have poor renal function.

COLONOSCOPY IN THE DIAGNOSIS OF CI

Recommendations

1. Early colonoscopy (within 48 h of presentation) should be performed in suspected CI cases to confirm the diagnosis (strong recommendation, low level of evidence) (¹⁷).
2. When performing colonoscopy on a patient with suspected CI, the colon should be insufflated minimally (conditional recommendation, very low level of evidence) (^{69, 135}).
3. In patients with severe CI, CT should be used to evaluate the distribution of disease. Limited colonoscopy is appropriate to confirm the nature of the CT abnormality. The endoscopic procedure should be stopped at the distal-most extent of the disease (strong recommendation, low level of evidence).
4. Biopsies of the colonic mucosa should be obtained except in cases of gangrene (strong recommendation, very low level of evidence).
5. Colonoscopy should not be performed in patients who have signs of acute peritonitis or evidence of irreversible ischemic damage (i.e., gangrene and pneumatosis) (strong recommendation, very low level of evidence).

Summary of evidence

Colonoscopy has become the primary technique to diagnose CI, usually after CT (see above) has revealed a segment of colon that is thickened with or without other more specific signs of ischemic damage, e.g., pneumatosis linearis and portal venous gas (see Figure 1). BE (see above) had been the main diagnostic mode in the 1960s but was gradually replaced by colonoscopy as the latter became increasingly available. Colonoscopy offers several advantages over radiologic imaging, including the ability to directly observe the abnormal tissue, to sometimes make a diagnosis just by appearance, and to obtain tissue samples. Nonspecific features of CI in addition to erythema, edema, and ulceration include aphthous ulcers, which may suggest Crohn’s disease, pseudomembranes, which are more often seen with C. difficile infection, and pseudopolyposis, which may develop with healing. Although the dusky, cyanotic hue of necrotic mucosa typifies gangrene and CI can be assumed in its presence, the validity of this assumption has never been tested. One retrospective study of 85 consecutive patients who had a colonoscopy to confirm diagnosis of CI showed that the most common findings in transient CI are edematous and fragile mucosa, segmental erythema, scattered erosions, longitudinal ulcerations, petechial hemorrhages interspersed with pale areas, purple hemorrhagic nodules, and a sharply defined segmental involvement (¹³⁶). This study provides a clear description of the colonoscopic findings seen in patients with mild, transient CI, but it suffers from the flaws of many of the currently published studies that do not provide frequencies of these findings in the context of disease severity. Montoro et al. (¹⁷) looked at 297 patients who underwent colonoscopy for all levels of CI severity: findings were erythema (83.7%), edema (69.9%), friability (42.6%), superficial ulcerations including the single stripe sign (57.4%), deep ulcerations (21.7%), luminal narrowing and stenosis (8.4%), intraluminal blood (8.4%), and blue-black nodules with dark-dusky backgrounds suggestive of gangrene (5.5%). Scattered erosions may also be seen both proximal and distal to continuous segments of involvement.

The colon single-stripe sign (CSSS) is a highly specific sign of CI described by Zuckerman et al. (¹³⁷) in 2003. CSSS is defined as a single inflammatory band of erythema with erosion and/or ulceration along the longitudinal axis of the colon; all CSSSs in the study of Zuckerman et al. (¹³⁷) were >5 cm in length and 89% of CSSSs were found in the left colon. The 26 patients with CSSS had a better prognosis than did the 22 patients with circumferential CI and exhibited a reduced need for surgery (0% vs. 27%) and a lesser mortality (4% vs. 41%). Although of great interest, the study of Zuckerman et al. (¹³⁷) was retrospective, small, had limited follow-up information, and a cause-and-effect relationship between ischemia and the CSSS was not proven.

Colonoscopy, in addition to detailing specific findings, also enables accurate determination of the anatomic limits of involvement and, therefore, perhaps helps predict prognosis. In 2010, Brandt et al. (⁷) showed CI that was isolated to the right side of the colon (IRCI) had a worse prognosis than CI involving any other anatomic segment of colon. In a multicenter Spanish study led by Montoro et al. (¹⁷) of 364 patients with CI, 345 of whom had colonoscopy, timing of colonoscopy was found to be important for diagnosis: when colonoscopy was performed within 48 h of presentation, typical hemorrhagic nodules were found in 47.1% of cases, but were evident in 33.3% of those in whom the examination was performed 2–5 days after presentation, and in only 9.7% of patients when colonoscopy was done after day 5. In 2011, in a retrospective study of 106 cases of CI divided into two groups, one with erythema and erosions and another with linear and circumferential ulcers, Beppu et al. (²³) showed that CI was more severe with the latter pattern of disease; abdominal pain, C-reactive protein levels, and length of hospital stay were the only measures of severity given in this analysis that were significant. It is important to note that the correlation between endoscopic appearance and clinical severity is not absolute and even disease in patients with boggy, purple mucosa may resolve completely without complication. It is essential to assess the entire clinical picture before making any judgments about the need for surgical intervention. Finally, colonoscopy enables biopsy specimens to be obtained and, although uncommonly pathognomonic for CI, they often support the diagnosis, but rarely establish etiology. Risks of colonoscopy are small but must be considered in terms of the pathophysiology of CI. In 1969, Boley et al., showed that intraluminal pressures of the magnitude generated during colonoscopy and BE examinations, i.e., 30-60 mm Hg, reduced intestinal blood flow, especially to the mucosa, and caused shunting of blood from the mucosa to the serosa (⁴⁸). In 1980, Kozarek et al. (¹³⁵) showed that serosal splitting followed by mucosal tears, pneumatosis, and transmural perforation occurred in human cadaver colons at air pressures of 52–230 mm Hg, and intraluminal pressures that ruptured the cecum and sigmoid were 81 and 169 mm Hg respectively. They further showed during routine colonoscopy in three patients that intraluminal pressures ranged from 9 to 57 mm Hg when the tip of the colonoscope was free in the lumen of the bowel and 34–138 mm Hg when the tip was impacted against the colon wall (¹³⁵). Carbon dioxide has been recommended for colonoscopy (and virtual colonography) because it is 10 times more rapidly absorbed from the bowel than is room air, and therefore its use results in a shorter period of distention and a more comfortable examination (^{138, 139, 140}). In addition, Brandt et al. (⁶⁹) showed that at all intracolonic pressures studied, CO₂ increased colonic blood flow in contrast to room air that, at similar pressures, decreased blood flow, thus making CI less likely a complication of colonoscopy when CO₂ rather than room air is used. However, CO₂ has never been formally studied in humans with CI and all evidence presented is indirect. Despite this, CO₂ appears to be the best insufflating agent to use in patients with CI and should be used whenever available.

There are currently no formal studies looking at perforation rates in patients undergoing colonoscopic evaluation of CI. The Brandt et al. (⁷) study cohort from 2010 included 251 patients who underwent a colonoscopic examination to the cecum using room air for insufflation; none of these colonoscopies were complicated by perforation (unpublished data). Of the 424 hospitalizations for CI observed by Longstreth and Yao (⁹), 408 colonoscopic evaluations were conducted using room air for insufflation, and again there were no documented perforations (unpublished data). In contrast to the assumed increased risks of endoscopic evaluation of CI, there does not seem to be a higher perforation rate in patients who undergo colonoscopic evaluation.

Limited colonoscopic evaluation is a safe and useful technique to evaluate patients with CI or suspected of having CI. CT should be used to characterize the stage of CI and the distribution of disease, but when colonoscopy is performed, the colonoscope should be passed only to the distal-most aspect of the affected region. Biopsies should be taken in all cases, except when there is evidence of gangrene.

Histopathologic specimens obtained colonoscopically are rarely diagnostic for CI. Pathognomonic features include infarction and ghost cells, i.e., preserved individual cellular outlines without cell content, and are infrequently seen (¹³⁸). It is more common to see biopsy specimens showing mucosal and submucosal hemorrhage and edema and capillary fibrin thrombi with neutrophilic infiltration that are consistent with CI which, when taken in the context of clinical presentation, will support the diagnosis. Most commonly, pathologic features are nonspecific (¹³⁸).

The diagnosis of CI, however, cannot be made conclusively by colonoscopy on just one examination unless mucosal gangrene is observed or infarction or ghost cells are seen at histopathology. Montoro et al. (¹⁷) found that of their 297 patients who underwent colonoscopy with biopsy, ghost cells and mucosal infarction were each seen in 7.7% of patients; of the 25 patients who had surgical resections, ghost cells and infarction were seen in 20% and 63%, respectively; thus, these histopathologic findings are infrequent and of limited value in diagnosing CI.

SEVERITY AND TREATMENT OF CI

Recommendations

1. Most cases of CI resolve spontaneously and do not require specific therapy (strong recommendation, low quality of evidence) (^{107, 108, 139}).
2. Surgical intervention should be considered in the presence of CI accompanied by hypotension, tachycardia, and abdominal pain without rectal bleeding; for IRCI and pancolonic CI; and in the presence of gangrene (strong recommendation, moderate level of evidence) (^{17, 107, 108}).
3. Antimicrobial therapy should be considered for patients with moderate or severe disease (strong recommendation, very low level of evidence) (^{107, 108, 140}).

Summary statement

1. When considering the mortality risk for patients undergoing surgical intervention for acute CI, Ischemic Colitis Mortality Risk (ICMR) factors should be utilized (^{141, 142}).

Summary of evidence

Surgical intervention and mortality are the most commonly used end points in studies that assess factors associated with poor outcome in CI and are used to define severe or complicated (poor outcome) disease in the data discussed below (Table 7). The strength of diagnosis from these studies is based upon the scheme presented in Table 8. Longstreth and Yao (⁹), Montoro et al. (¹⁷), and Brandt et al. (⁷) assessed large cohorts of patients looking at predictors of poor outcome, whereas Añón et al. (¹⁰⁸), Huguier et al. (⁹⁴), Lee et al. (²⁴), O’Neill et al. (¹⁴³), and Mosele et al. (¹⁰⁷) each assessed smaller cohorts. These studies all conducted multivariate analyses and found various significant predictors of outcome including epidemiologic factors, clinical presentation of disease, vital signs, serologic values, and disease distribution. Epidemiologic factors that were associated with poor outcome included antimicrobial therapy for CI, hepatitis C positivity, history of cancer, male gender, and warfarin use at the time of diagnosis (^{9, 94, 107}). The use of antimicrobial therapy is likely an indicator of the clinician’s response to disease severity rather than the antibiotic therapy per se contributing to poor outcome. Presentations of disease, e.g., abdominal pain without rectal bleeding, nonbloody diarrhea, peritoneal signs on physical examination and symptom onset after admission, are all associated with poor outcome (^{9, 17, 94, 107, 108, 143}). A heart rate of >100 beats per min and systolic blood pressure of <90 mm Hg at the time of diagnosis are associated with the need for surgical intervention and/or mortality as are Hgb <12 mg/dl, hyponatremia (Na <136 mEq/l [mmol/l]), LDH >450 U/l, and blood urea nitrogen (>28 mg/dl [mmol/l]) (^{9, 17, 107, 108}). As discussed previously, a pancolonic distribution of disease and IRCI also portend a poor outcome (^{7, 9, 107, 143}). These studies were heterogeneous in terms of patient characteristics, physical examination findings, medical comorbidities, serology, and thresholds of various factors considered, but several risk factors appeared to be significantly associated with poor outcome in several of these studies: hypotension, tachycardia, abdominal pain without rectal bleeding, IRCI, and pancolonic CI.

Chung et al. (¹⁴⁴) assessed 152 patients with clinically confirmed CI (74.5% confirmed pathologically) to analyze significant risk factors for poor outcome (e.g., improvement delayed by >2 weeks, 30-day colectomy, and 30-day mortality) and used these factors to devise a prognostic scoring system. The strongest predictors in decreasing order of significance were ulceration on colonoscopy, shock within 24 h of admission, and tachycardia at the time of diagnosis (¹⁴⁴). If a patient had all three factors, their risk index for poor outcome was 74.5, whereas if ulceration and shock were present, but no tachycardia, the risk index decreased to 39.4; the risk index was 1.00 when none of these factors were present (¹⁴⁴). This simple system is easy to use but requires further validation before being broadly applied, given their inclusion of patients with nonpathologically confirmed CI and including a nontraditional end point for severe disease (i.e., improvement delayed by >2 weeks). Ulceration seen on colonoscopy also has been associated with poor outcome in a study by Matsumoto et al. (¹³⁹). This small study using clinical diagnostic criteria showed that patients with ulceration on colonoscopy required longer periods of fasting and had a longer duration of admission before clinical improvement compared with those without ulceration. This study was very small but is a reinforcement of the importance of mucosal ulceration as a predictor of outcome in patients with CI.

In view of the above data and previously described diagnostic options, this guideline proposes redefining disease severity to guide treatment (Table 6 and Figure 1). This is currently a nonvalidated guide to rating disease severity but incorporates the most up-to-date knowledge of risk factors for disease severity. Patients with mild disease are those who have typical symptoms of CI with a segmental colitis not isolated to the right colon and with none of the commonly associated risk factors for poorer outcome that are seen in moderate disease. Moderate disease includes any patient with up to three of the following factors: male gender, hypotension (systolic blood pressure <90 mm Hg), tachycardia (heart rate >100 beats/min), abdominal pain without rectal bleeding, blood urea nitrogen >20 mg/dl, Hgb <12 g/dl, LDH >350 U/l, serum sodium <136 mEq/l (mmol/l), WBC >15 × 10⁹/l, or colonoscopically identified mucosal ulceration. Severe disease is defined by more than three of the previously listed criteria or any of the following: peritoneal signs on physical examination, pneumatosis on CT, gangrene on colonoscopy, and a pancolonic distribution or IRCI on CT or colonoscopy.

Treatment of CI varies with the severity of the disease and its presentation. In general, many patients have a benign, self-limited episode of CI that is neither diagnosed nor treated or is managed in the outpatient setting. Most cases of CI resolve spontaneously and do not require specific therapy; such patients have reversible ischemic colopathy or transient ischemic colitis (⁹¹). Patients with more significant symptoms or findings require hospitalization to enable observation for complications or signs of irreversible disease. Initial medical management most commonly includes general supportive measures, bowel rest, intravenous hydration, and correction of possible precipitating conditions. Parenteral nutrition may be indicated if a protracted course is anticipated. No therapeutic modality has ever been tested in a rigorous clinical trial and the mostly benign nature of CI has led to a shortage of trials assessing various therapies. It is also important to consider that the currently available literature for the management of CI focuses on hospitalized patients who have complicated disease with less emphasis on the milder cases that are not hospitalized. The recommendations presented in this guideline are mostly based on small retrospective studies that lack control groups. Most guideline statements are based upon extrapolation from the available literature and expert opinion.

One systematic review of the management of CI included 10 retrospective studies comprising 841 patients (¹⁴⁵). Within these studies there was no standard protocol for management; however, nonsurgical treatment usually included bowel rest, intravenous fluid, electrolyte repletion, and antibiotic usage with occasional administration of total parenteral nutrition. All articles reviewed did agree that surgery was indicated for patients with peritonitis or those who were hemodynamically unstable (¹⁴⁵). This systematic review had significant limitations including methodologic anomalies such as excluding some well-structured publications and also including three studies with patients who had CI as a complication of vascular procedures despite methodology stating this as a criterion for exclusion. In addition, the study included some patients without pathologically confirmed disease. This lack of uniformity characterizes the CI management literature, but the abstract’s conclusion reads “There is very little evidence base for the management of this condition” (¹⁴⁵). We believe this statement is accurate.

O’Neill et al. (¹⁴⁸) conducted a subsequent systematic review of the management of CI, attempting to improve on the methodologies from Diaz-Nieto et al. (¹⁴⁵). This review provided a more rigorous screening evaluation of the studies by including only biopsy-proven disease in each patient and excluding all studies with CI resulting from vascular surgery or associated with colonic adenocarcinoma (¹⁴⁸). Eight retrospective case-series and three case-control studies comprising 1049 patients were included. Overall, 80.3% of cases were medically managed without surgical intervention and had a 6.2% mortality rate; 19.6% were managed surgically with a mortality rate of 39.3% (¹⁴⁸). These data offer the best overall insight into medical vs. surgical management of CI and associated outcomes. This analysis is also consistent with the presumption that those requiring surgery are the sickest and at highest risk for poor outcome. Medical management most commonly included fasting, intravenous fluids, parenteral nutrition, antibiotics, and heparin prophylaxis (¹⁴⁸). Requirement for pathologic confirmation of disease is a strength of this study, and is also a marker of disease because colonoscopy or surgical intervention isolates those with more advanced disease or when the diagnosis might be unclear. There was, however, also significant heterogeneity among the algorithms used in various studies for medical management of disease, with some including and others excluding, antimicrobial therapy, and none detailing the frequency, duration, or type of antibiotics used.

Indication and timing for antimicrobial use in CI remains untested. There is no clinical evidence to show beneficial effects of such therapy and it is unlikely that a randomized controlled trial will be forthcoming; because the prognosis of CI is excellent in most cases, the number of patients in such a trial needed to show benefit (or to disprove a lack of benefit) would have to be immense. Previous recommendations for antibiotic usage were based on experimental studies that showed reduction in the severity and extent of bowel damage when antibiotics were given before or during an ischemic event, with the majority of these trials being several decades old (^{140, 146, 147, 149, 150}). Yoshiya et al. (¹⁴⁰) pretreated mice with ampicillin, vancomycin, neomycin, and metronidazole, and subsequently induced intestinal ischemia with reperfusion by ligating the SMA for 30 min. These mice were then compared with control mice that also underwent intestinal ischemia but did not receive antimicrobial therapy. In the mice pretreated with antibiotic therapy, the expression of Toll-like receptors 2 and 4 was decreased as was expression of inflammatory markers (e.g., tumor necrosis factor, interleukin-6, and cyclo-oxygenase-2). As Toll-like receptors normally bind Gram-positive and Gram-negative bacteria to stimulate inflammatory responses, the study supports the theory that antimicrobial therapy decreases the overall inflammatory response to ischemic injury. Intestinal injury resulting from ischemia and reperfusion is also believed to be stimulated by ischemic cells presenting neoantigens that bind natural immunoglobulins to activate the complement pathway. Yoshiya et al. (¹⁴⁰) also showed decreased complement (C3), IgM, and IgA in ischemic tissue of mice pretreated with antimicrobials compared with the ischemic tissues of mice not exposed to antimicrobial therapy.

Antimicrobial therapy is also believed to prevent poor outcome through decreased bacterial translocation in the setting of acute ischemia and reperfusion injury. The loss of mucosal integrity as a result of vigorous inflammation has been shown in mouse models to facilitate bacterial translocation and resulting bacteremia (^{151, 152, 153}). Luo et al. (¹⁵³) studied mice that had ischemia induced by SMA occlusion for 30, 60, or 90 min followed by reperfusion for 30 min. These mice were given varying oral doses of bacteria after the ischemic episode and had blood cultures drawn immediately and 15 min after reperfusion. A control group of mice was fed the same bacteria but did not have any ischemic episode. Control mice showed no translocation of bacteria, whereas the mice that were exposed to even 30 min of ischemia and a low bacterial inoculum developed bacteremia (¹⁵³). The studies of Yoshiya et al. (¹⁴⁰) and Luo et al. (¹⁵³) reinforce the theories behind antimicrobial therapy in the setting of CI in mouse models. Antibiotics are believed to improve outcome in CI by reducing inflammatory responses stimulated by the normal fecal microbiome, reducing the antigens that prompt innate immune response and minimizing bacterial translocation through compromised colonic mucosa.

Available human studies of antimicrobial treatment in CI are few in number, heterogeneous in design, and do not directly address antimicrobial use for a specific therapeutic benefit. Three studies have presented data regarding broad-spectrum antimicrobial therapy of CI and associated outcomes (^{107, 108, 139}). Añón et al. (¹⁰⁸) looked at 85 consecutive patients, of whom 69 had “mild disease” that was treated with fluid and electrolyte correction and “wide-spectrum antibiotics”; all but 1 patient survived (98.6%) and another 16 patients had severe disease that was initially treated medically, but subsequently required surgery. Mosele et al. (¹⁰⁷) retrospectively assessed 46 patients who were at least 65 years of age, of whom 67.4% clinically improved with only medical therapy consisting of intestinal rest, total parenteral nutrition, and broad-spectrum antibiotics; 6 of the original cohort received medical therapy and subsequently died (13.0%), whereas another 9 would ultimately require surgical intervention (19.6%). Matsumoto et al. (¹³⁹) retrospectively assessed 41 patients with CI, with 31.7% of patients receiving a “combination of antibiotics” along with fasting, bowel rest, and fluid replacement; none of these patients required surgery or died. These studies are similar as they were all small, not designed to assess the effect of antimicrobial therapy on outcome, mentioned antimicrobial therapy as part of the medical treatment regimen but mostly did not detail the number of patients within the “medical therapy” cohort who received antibiotics, and did not delineate the specific antibiotics used or the duration of therapy. Perhaps most importantly, there were no comparator groups of patients who did not receive antimicrobials.

By performing multivariate analysis on 401 consecutive patients, Longstreth and Yao (⁹) found that antimicrobial therapy was associated with a 3.94 (confidence interval: 1.23–164, P<0.05)-fold increased risk of severe disease as defined by need for surgery or mortality. This finding speaks to one of the major weaknesses of all available treatment data, namely that this statistical association is likely the result of a selection bias by which patients with the most severe disease are those receiving antimicrobial therapy, most of whom are most at risk for poor outcomes regardless of therapy.

Because there is a shortage of clinical trials of antimicrobial therapy in CI and there is experimental evidence that these medications should be beneficial, this guideline recommends antimicrobial therapy for CI patients who have either “moderate” or “severe” disease as newly classified in a previous section of this guideline (see also Figure 1). The criteria for classification are based upon risk factors for poor outcome in CI that is believed to be mediated through a vigorous inflammatory response, bacterial translocation, and extensive damage to the large bowel. Mouse models have shown antimicrobial use moderates these processes and we believe these will improve outcome in the appropriate clinical setting. This recommendation is based upon expert opinion considering the murine models, retrospective human studies, and personal experience. As previously discussed, it is highly unlikely that a properly designed randomized control trial of antibiotics in humans stratified for mild, moderate, and severe CI will ever be performed.

The question of which antimicrobial agents are most efficacious in patients with CI is also unanswered. Plonka et al. (¹⁵⁴) transected the vascular supply to the colon in mice and then treated them with saline, gentamicin alone, metronidazole alone, or gentamicin plus metronidazole. Mice receiving metronidazole alone or gentamicin plus metronidazole showed improved survival. This study supports the role of anaerobic bacteria as a contributing factor to poor outcome. Because of the risks of bacterial translocation across compromised colonic mucosa, with inflammatory and innate immunologic reactions that are stimulated by the colonic microbiota, we recommend a broad antimicrobial regimen to “cover” these organisms. Possible antimicrobial regimens include an anti-anaerobic agent plus a fluoroquinolone or an aminoglycoside or a third-generation cephalosporin.

The optimal duration of antimicrobial therapy is unclear and studies to date have not addressed this issue. Once therapy has been initiated, this guideline recommends that antimicrobials be continued for at least 72 h, at which time the patient’s clinical status should be reevaluated. If the patient has not clinically improved, one should consider consultation with an infectious disease expert to help define the antimicrobial regimen. If the patient is symptomatically improved after 72 h, a 7-day course of therapy should be considered.

Glucocorticoids are not recommended for CI, except when it is a complication of a vasculitis. In such cases, the steroids are used to treat the vasculitis and not the CI. The only potential role for these agents or others used in the treatment of IBD might be in patients with chronic CI (see above), but there is no published experience to support the use of local or systemic glucocorticoids, sulfasalazine, aminosalicylates, or fatty acid enemas to treat CI. Indeed, there are numerous reports implicating these drugs as a cause of CI and, in one experimental study using Fischer rats, azathioprine and methylprednisolone augmented the damage of intestinal ischemia (¹⁵⁵). Nonsteroidal anti-inflammatory drugs are generally accepted to be a cause of colonic ulcerations, but one large trial of patients with CI showed that a lack of nonsteroidal anti-inflammatory drug use at the time of diagnosis was independently associated with severe disease (e.g., surgery and/or death) (⁹). This potentially protective effect of nonsteroidal anti-inflammatory drugs was an indirect observation with no further trials to date (⁹). Anecdotal evidence has shown that fatty acid enemas have helped to heal the necrotic mucosa of two patients with otherwise unresponsive CI, although there are no formalized publications supporting this intervention (LJ Brandt and SJ Boley, unpublished).

Surgical intervention for patients with CI is required in its most severe presentations. The O’Neill et al. (¹⁴⁸) review of 11 studies comprising 1,049 patients showed that 19.6% of pooled patients required surgical intervention and had a mortality rate of 39.3%. Published data have wide ranges for the need for surgery and vary with the study population (e.g., surgical service vs. general population; age >65 vs. <65 years); 8–19.8% of large, broad populations with biopsy-proven CI have been estimated to require surgical intervention, whereas the need for surgery in patients admitted to surgical services is as high as 45.2% (^{7, 9, 94}). The requirement of biopsy-proven disease in some studies mandates endoscopic evaluation, thereby perhaps excluding patients not sick enough to warrant colonoscopy, biasing the study population to patients who are more acutely ill, and overestimating the true need for surgical intervention. Inclusion of patients on a surgical service is not representational of a general population and cannot be applied to all patients with CI. Given the mostly benign course of CI, and the fact that many patients with CI do not present to medical attention, estimates of the frequency of surgery in hospitalized patients with CI is likely a significant overestimation. We recommend considering surgical consultation if patients fulfill the risk stratification for moderate or severe disease as these factors are associated with need for surgery and/or mortality.

There are several indications for surgery in acute CI, but the most common is the presence of signs of necrotic bowel (Table 9). These might include peritoneal signs on physical examination, pneumatosis or portal venous gas on radiologic imaging, or gangrene on colonoscopic examination. Without surgical intervention, mortality from necrotic bowel approaches 100%. Following the new classification scheme proposed here, patients with “severe disease” would require emergent surgical consultation. The current literature does not clarify surgical indications or optimal timing for surgical intervention. Huguier et al. (⁹⁴) assessed 33 patients who underwent surgery for CI, 13 of whom had immediate surgery (i.e., <12 h after admission) for “abdominal tenderness,” with 5 having peritonitis upon surgical inspection and 8 expiring; 20 patients had delayed surgical treatment (i.e., days 2–7 of hospitalization) for clinical deterioration after admission, with 6 subsequently dying from complications of CI. Patients who develop a segmental colitis pattern during the evolution of their disease and whose symptoms persist for more than 2–3 weeks or who have a continuing protein-losing colopathy for several weeks usually are best treated by segmental colectomy. A less-recognized indication for surgery is the development of recurrent sepsis in a patient who has symptomatically recovered from an acute episode of CI. Such patients usually have a short segment of unhealed bowel that, via bacterial translocation, is the source of the sepsis; resection of this segment is curative.

The surgical procedure patients undergo for CI depends upon the affected segment of colon, but most commonly include total or subtotal colectomy, right hemicolectomy, or segmental colectomy with either a primary anastomosis or diverting stoma (Table 10). Antolovic et al. (¹⁵⁶) prospectively collected data on patients requiring surgical intervention at the University of Heidelberg between 2001 and 2004. Of the 85 consecutive patients undergoing surgery, 56 (66.7%) required procedures within 24 h of presentation to the surgical service. Left hemicolectomy was performed in 8%, right hemicolectomy in 26%, sigmoid resection in 5%, and total colectomy in 49%, with the remainder having other segmental resections; 42% had a Hartmann procedure and another 38% had a diverting stoma created in addition to bowel resection (¹⁵⁶). Castleberry et al. (¹⁴¹) retrospectively assessed 115 consecutive patients undergoing surgery for acute CI: segmental colon resection was most common (53%), with right hemicolectomy (49%) being the predominant segmental resection performed; 26% had subtotal or total colectomy, 10% had ileocecectomy, and 10% had a Hartmann procedure; 87% of patients also required an ileostomy or colostomy. Reissfelder et al. (¹⁴²) prospectively assessed 177 consecutive patients undergoing surgical management for acute CI between 2002 and 2008. Subtotal colectomy was performed in 54%, right hemicolectomy in 38%, and left hemicolectomy in 8%; 61% of their patients had a stoma placed (¹⁴²). These studies provide a cross-sectional view of surgical treatment patterns for CI, but of course do not assess which surgical procedure is most efficacious for which presentation. The choice of surgical procedure is influenced by a number of factors including the training of the surgeon and his or her practice patterns.

Mortality after surgical intervention for CI is high, ranging from 37 to 47% (^{141, 142, 156}). Risk factors for postoperative mortality are more definable than most factors associated with CI, given the fact that the diagnosis is known after surgery, the disease is acute, and the various clinical factors can be quantified. Antolovic et al. (¹⁵⁶) assessed 85 consecutive patients with acute CI undergoing surgery and found a mortality rate of 47%. American Society of Anesthesiologists Status (ASA) >3, emergency surgery, and blood loss >300 ml were each independent and strong predictors for mortality. Reissfelder et al. (¹⁴²) conducted a retrospective analysis of 177 consecutive patients and identified low-output heart failure (e.g., cardiac ejection fraction <20% on echocardiogram), acute kidney injury, lactate >2.5 mmol/l, subtotal colectomy, and pre- and intraoperative catecholamine administration as being the risk factors associated with high perioperative mortality rate. This study then showed that if one point was given for each of the listed risk factors, mortality increased from 10.5 to 100% as points increased from 0 to 5 respectively (Table 11). This classification scheme was termed the Ischemic Colitis Mortality Risk Score (ICMR) (¹⁴²). Castleberry et al. (¹⁴¹) retrospectively assessed 115 patients who underwent surgical treatment of their CI to assess for factors associated with postoperative mortality and to validate the ICMR. 37% of the patients died in the postsurgical hospitalization period and multivariate predictors of mortality were as follows: ASA class >4 (odds ratio (OR): 6.91, confidence interval (CI): 2.17–30.98, P<0.001), peak preoperative lactate level (OR: 1.26, CI: 1.11–1.47, P<0.001), postoperative dialysis requirement (OR: 5.11, CI: 1.72–16.88, P<0.001), intraoperative adrenergic vasopressor requirement (OR: 2.07, CI: 1.47–8.82, P<0.01), total ICMR score (OR: 2.11, CI: 1.44–3.22, P<0.001), and blood loss >500 ml (OR: 2.63, CI: 1.00–7.21, P<0.05) (¹⁴¹). Despite the ICMR score being predictive of mortality when its individual components were considered, only elevated preoperative serum lactate (P<0.01), renal failure requiring hemodialysis (P<0.01), and intraoperative vasopressor requirement (P<0.01) were significantly associated with poor outcome, whereas low cardiac output and whether subtotal or total colectomy was performed were not associated (¹⁴¹); risk stratification using ICMR was validated in this study. There are some limitations of the ICMR including the challenges with differentiating occlusive from nonocclusive CI in the acute setting and that mortality associated with colectomy might result from other factors and not the surgery per se (e.g., the extent of colonic injury, significant blood loss, or overall hemodynamic instability). This guideline recommends that ICMR should be considered as part of the risk assessment of patients with acute CI who need surgical intervention.

Colon stricture after an episode of CI may be asymptomatic or even resolve over months to years. Surgery is indicated only when an ischemic stricture produces symptoms; in such cases, segmental resection is adequate. Transendoscopic dilation of an ischemic stricture is an alternative to surgery, although an unproven one. Chronic segmental CI is a more controversial indication for surgery and, as with other colitides, the decision to abandon medical therapy is a complex one that must be individualized for each case. Recurrent CI is uncommon and resection of the involved segment of colon, while usually curative, does not necessarily protect against recurrent CI in other areas of the colon. There are no published data on the frequency of such recurrence following surgery for recurrence.

Treatment of CI depends upon disease severity at presentation (see Figure 1). The overwhelming majority of patients will require simple conservative measures to manage their disease including fasting, intravenous fluids, and correction of underlying conditions. If patients have moderate or severe disease, broad-spectrum antimicrobial coverage should be instituted along with surgical evaluation. Surgery should be consulted promptly for patients with severe disease or colon necrosis.

ACKNOWLEDGMENTS

This guideline was produced in collaboration with the Practice Parameters Committee of the American College of Gastroenterology. The Committee gives special thanks to Fouad J. Moawad, MD, FACG, who served as guideline monitor for this document.