This study was approved by the institutional review board of Dana-Farber Cancer Institute, and the requirement for informed consent was waived. Twenty-three patients with melanoma undergoing ipilimumab treatment who visited an outpatient clinic or emergency department with a chief complaint of gastrointestinal symptoms or who required consultation for gastrointestinal symptoms were identified in the electronic medical record database. An additional search of the radiology database did not add patients. Of the 23 patients identified, six patients were excluded because they had non–immune-mediated causes of their symptoms, as follows: one patient had constipation, one patient had blood-tinged stool due to hemorrhoids, one patient had a positive Clostridium difficile stool test, one patient had extensive peritoneal metastases, one patient had right lower quadrant pain due to large pelvic metastases, and one patient had colon perforation caused by necrosis of a serosal metastasis of the colon. The last patient was excluded because the colon perforation was secondary to disease response to ipilimumab rather than an immune-mediated adverse inflamma-tory event. The remaining 17 patients were diagnosed as having ipilimumab-associated colitis on the basis of the complete workup, including clinical features and radiologic, pathologic, endoscopic, or laboratory findings. One of these 17 patients was excluded because of the lack of an available CT scan at the time of symptoms. Finally, 16 consecutive patients (mean age, 62.4 ± 11.6 years; 12 men and four women) who were diagnosed with ipilimumab-associated colitis and who underwent CT scans at the time of gastrointestinal symptoms were included in this study.

Ipilimumab was administered at a dose of 3 mg/kg every 3 weeks during the induction period (four doses) and then every 12 weeks during the maintenance phase. All patients underwent contrast-enhanced CT of the chest, abdomen, and pelvis at treatment initiation (baseline CT) and every 3–4 months during ipilimumab treatment (restaging CT scans). All patients in our series also underwent CT scans of the abdomen and pelvis at the time of symptoms.

CT scans of the abdomen and pelvis or chest were performed by using a 64-MDCT scanner (Aquilion 64, Toshiba America Medical Systems) with the following protocols: 64-MDCT scanner at 0.5-mm collimation, 120 kVp, tube current maximum of 500 mA using dose modulation with noise index of 12.5 HU, 0.5-second gantry rotation time, and a table speed of 26.5 mm per rotation. One hundred milliliters of iopromide (300 mg I/mL; Ultravist 300, Bayer HealthCare Pharmaceuticals) was injected IV with an automated injector (Stellant, Medrad) at a rate of 2–3 mL/s, with a scan delay of 60 seconds. Oral contrast agent (diatrizoate meglumine; Gastrografin, Bracco Diagnostics) was administrated before the CT scans. Axial images with 5-mm thickness and coronal images with 4-mm thickness were reconstructed using standard abdomen algorithms and were transferred to the PACS.

CT scans were evaluated on a PACS retrospectively and jointly by two radiologists with 12 and 7 years of experience. Disagreement was minor and was resolved through consensus. The CT scans were analyzed for the presence of bowel wall thickening, bowel mucosal enhancement, bowel distention, pneumatosis, perforation, pericolic fat stranding, and mesenteric vessel engorgement in the colon. The bowel wall thickness was measured at two representative locations on the PACS, and the mean value was used for grading. The degree of bowel wall thickness was graded as mild (4–8 mm), moderate (8–12 mm), and severe (> 12 mm), which is modified from the criteria described by Bharucha et al. [10]. Wall thickness of 4 mm or less was considered normal because the normal colonic wall thickness can reach 3–4 mm when the lumen is collapsed [11]. The extent of colon wall thickening were recorded as focal (a few centimeters), segmental (10–30 cm), multisegmental (separate sites of segmental bowel wall thickening), and diffuse (> 30 cm) [12]. The colonic mucosal enhancement was evaluated and compared with the enhancement of other segments of the gastrointestinal tract, such as the small bowel and stomach. The presence of mural stratification, such as a “target” or “halo” appearance of bowel, was recorded. Bowel distention was defined as large-bowel diameter greater than 6 cm or small-bowel diameter greater than 2.5 cm [13, 14]. Mesenteric vessel engorgement was defined as prominence or tortuosity of the vasa recta increased over the baseline CT.

CT scans performed at the time of gastrointestinal symptoms were compared with the baseline pretreatment CT scans to identify interval change or development of the aforementioned CT findings. Any available follow-up CT studies were also assessed for resolution of the acute colitis-related findings after cessation of ipilimumab.

Clinical data regarding ipilimumab treatment duration, symptoms, laboratory abnormalities, pathologic results of endoscopic biopsy specimens, and the treatment of the ipilimumab-associated colitis were recorded during review of the medical records. The baseline CT scan obtained before the start of ipilimumab was compared with all available follow-up CT scans. The radiologic response according to the immune-related response criteria were determined at the time of diagnosis of ipilimumab-associated colitis at the time of next CT follow-up (i.e., complete response, partial response, stable disease, and progressive disease) [3].

Results were statistically analyzed to identify the difference between patients with two distinct imaging patterns, which are described in the Results section. The Mann-Whitney U test was used to compare mean bowel wall thickness, and the Fisher exact test or chi-square test was used to compare the categoric data, using MedCalc software (MedCalc). All p values of less than 0.05 were considered to indicate a statistically significant difference.

The clinical features and imaging findings of all patients are described in Table 1. The common CT findings of ipilimumab-associated colitis were mesenteric vessel engorgement (13/16 [81.3%]) followed by bowel wall thickening (12/16 [75%]). Mesenteric vessel engorgement was very common in the CT scans at the time of colitis symptoms, but resolved at the immediate follow-up CT scans in all patients with this finding. In four patients, only mesenteric vessel engorgement was noted, without bowel wall thickening, which may suggest that it is an early finding of ipilimumab-associated colitis. Fluid-filled distended colon suggestive of diarrhea was noted in four (25%) patients. Increased mucosal enhancement was noted in three (18.8%) patients. Small-bowel involvement was not present except in two patients showing mild wall thickening of terminal ileum. Colon perforation was noted in one patient. None of the patients with ipilimumab-associated colitis had pneumatosis or halo or target signs suggestive of bowel wall edema or bowel obstruction.

Two distinct patterns of ipilimumab-associated colitis were observed, as illustrated in Figure 1 and Table 2. The first pattern is characterized by mesenteric vessel engorgement with mild diffuse bowel wall thickening or fluid-filled distended colon. These are nonspecific findings frequently seen in any kind of colitis (Fig. 2). The second pattern is characterized by segmental moderate wall thickening and associated pericolic fat stranding in a segment of preexisting diverticulosis (Fig. 3). In this study, we labeled the first pattern as diffuse colitis and the second pattern as segmental colitis associated with diverticulosis (SCAD). The distribution of the bowel wall thickening varied in the patients with the diffuse colitis pattern, as follows: pancolitis (n = 3), terminal ileum or cecum and sigmoid colon (n = 1), ascending and transverse colon (n = 1), descending and sigmoid colon (n = 1), and rectosigmoid colon (n = 2). In contrast, the SCAD pattern in all four affected patients involved the sigmoid colon, where underlying diffuse diverticulosis was present. In addition, baseline CT in those four patients showed mild (n = 3) or moderate (n = 1) bowel wall thickening preexisting in the sigmoid colon, which may suggest the presence of subclinical chronic inflammation [15]. The SCAD pattern resembles typical diverticulitis seen in other settings, with similar appearance of segmental wall thickening, pericolic fat stranding, pelvic fascial thickening, mesenteric vessel engorgement, and a few inflamed diverticula.

Between the two distinct CT patterns, the bowel wall thickness was significantly greater in the SCAD pattern than that in the diffuse colitis pattern (13.8 ± 7.6 mm vs 6.2 ± 3.5 mm; p = 0.014), and pericolic fat stranding was significantly more common in the SCAD pattern than that in the diffuse colitis pattern (75% vs 8.3%; p = 0.027). The fluid-filled colonic distention was found only in the diffuse colitis pattern (Table 2).

The most common clinical symptoms in our series were watery diarrhea and increased number of bowel movements. The median number of treatment cycles and the time from beginning ipilimumab therapy to CT at the time of symptoms was 3 cycles and 8 weeks, respectively (range, 1–4 cycles and 2–14 weeks, respectively). In all patients, colitis symptoms started within 1 or 2 weeks after the last ipilimumab administration.

Differences in the clinical features between patients with the diffuse colitis pattern and patients with the SCAD pattern were observed (Table 2). In the patients with the diffuse colitis pattern, profuse watery stool was predominant, whereas the mixed watery and bloody diarrhea with frequently associated cramping pain predominated in the patients with the SCAD pattern. However, systemic symptoms such as fever and chills were not frequent in either groups. Examinations for stool leukocytes, stool cultures, and a C. difficile titer were within normal limits in all patients. Immune-related adverse events outside the gastrointestinal tracts were observed in five patients (skin rash, n = 3; hepatitis, n = 1; hypophysitis, n = 1).

Among patients with the diffuse colitis pattern, three were treated with oral steroids and nine were treated initially with IV steroids followed by oral steroids. Two patients were refractory to steroid therapy. In one patient, symptoms resolved after administration of infliximab, a monoclonal antibody against tumor necrosis factor–α used to treat autoimmune disease. In the other refractory patient, pancolitis with colonic distention progressed on serial imaging, and multifocal bowel perforation developed (seven perforation sites were confirmed on the surgical specimen). In all but two patients with the diffuse colitis pattern (one underwent total colectomy and one had no available follow-up CT), the abnormal CT findings at the time of colitis completely resolved on the follow-up restaging CT scans (median, 12 weeks; range, 8–17 weeks).

The patients with the SCAD pattern were treated with steroids and antibiotics. Symptoms resolved within 2–6 weeks after treatment in all patients. One patient showed marked improvement of the bowel wall thickening at the first follow-up CT performed 11 weeks after the colitis; the other three patients showed relatively slow improvement of bowel wall thickening, which decreased to baseline at the second follow-up CT scans (median, 22 weeks; range, 16–24 weeks).

Regarding treatment response at the time of development of the gastrointestinal symptoms (median interval from the beginning of the ipilimumab, 9 weeks), 14 of 16 patients (87.5%) achieved partial response (n = 4) or stable disease (n = 10), and only two patients (12.5%) had progressive disease. However, at the time of follow-up CT scans (median, 12 weeks; range, 8–23 weeks, after cessation of ipilimumab treatment), eight patients (50%) had stable disease and six patients (37.5%) had progressive disease. For two patients, no follow-up data were available. After cessation of ipilimumab treatment, treatment response changed from stable disease or partial response to progressive disease in six patients (37.5%).