As new and more effective treatments for peritoneal carcinomatosis emerge, patients are beginning to live longer with the disease. Thus, more patients may present to the ED with complications of their underlying cancer or from their treatment. Indeed, in our initial database of 524 cases with a carcinomatosis-like appearance on CT, 350 patients (67%) had a known cancer (Fig. 1). When imaging for presurgical planning and follow-up, structured reporting utilizing the peritoneal carcinomatosis index (PCI) (which divides the peritoneum into 13 locations including a three-by-three grid over the abdomen and pelvis, upper and lower jejunum, and upper and lower ileum) and necessitating mention of the largest implant sizes and presence of vascular and ureteral involvement has been advocated [5, 6]. A comprehensive description such as this is generally not possible in the ED setting.

The goal of our study was to evaluate new cases of peritoneal carcinomatosis diagnosed in the ED, in patients without a preexisting malignancy. This diagnosis is typically made on abdominal CT scan, the workhorse of ED abdominal imaging. Other modalities can be used to diagnose or characterize peritoneal carcinomatosis [7], but these are infrequently performed in the ED setting.

In our study, the most common cause of new peritoneal carcinomatosis in women was of GYN origin (66%), and in men was of GI tract origin (proximal GI plus colon, 57%) (Table 1). Our study sample did not find any cases of peritoneal mesothelioma, a rare primary peritoneal tumor [8]. Among all cases of new carcinomatosis-like appearance on CT which received further workup, the positive predictive value (PPV) of pathologically confirmed new peritoneal malignancy was 82%. Many mimics of peritoneal carcinomatosis have been described in the literature, including peritoneal tuberculosis (TB) [9], fibroids [10], peritoneal sarcoid [11], endometriosis [12], actinomycosis [14], and splenosis [13], some of which were found to result in false positive CT findings in our study. The PPV of these mimics will vary based on their respective prevalences in the population. In our study, only one in 131 patients (0.8%) with a new carcinomatosis-like appearance who received further workup was diagnosed with peritoneal TB, which is not unexpected given the low prevalence of TB in the United States [14]. CT findings suggestive of peritoneal TB rather than carcinomatosis include loculated ascites, splenomegaly, and lymph node conglomeration, while focal hepatic and nodular omental lesions favor carcinomatosis [11]. A multivariate model [15] and texture analysis [16] have also been used to differentiate these two entities.

Identifying the primary lesion on CT scan once cancer has spread to the peritoneum is challenging but is beneficial in the ED setting in order to direct the trajectory of care to the correct medical service. The use of IV contrast may assist in this identification. For example, IV contrast may help delineate a hypo- or hyper-enhancing parenchymal primary lesion which would be obscured on a non-contrast study, particularly in the setting of ascites and fluid infiltration of abdominal fat (which is commonly present in peritoneal carcinomatosis). In our study, the primary lesion was identified in 40% of cases without IV contrast compared with 58% of cases with IV contrast, suggesting the utility of IV contrast; however, this difference was not statistically significant. Ascites is believed to develop in peritoneal carcinomatosis due to lymphatic obstruction preventing the resorption of peritoneal fluid, as well as due to secretion of vascular permeability factor by the cancer cells [1]. We found a statistically significant association of higher volume ascites with GYN tumors and adenocarcinomas of unknown primary compared to other primary tumors, although overlap was present (Table 3).

Our study had several limitations. Due to the retrospective design and reliance on a report database keyword search, there may have been positive cases which were not included in our cohort because the interpreting radiologist did not appreciate the presence of peritoneal carcinomatosis. This could have been a problem particularly for CT scans without intravenous contrast, as the presence of ascites can easily obscure peritoneal nodularity on non-contrast exams. We also did not analyze differences in accuracy of reporting amongst radiologists of different subspecialties (e.g. body radiologists, ED radiologists, and radiologists on-call reading outside their subspecialty). Because all radiologists in our department routinely interpret body CT while on-call, we believe that this reflects a typical ED practice setting where there is a baseline level of competence amongst all the radiologists, with some having varying levels of advanced expertise. Finally, our finding of intravenous contrast improving prediction accuracy was not statistically significant. However, intravenous contrast has been shown to improve lesion detection in other contexts [17], and it is therefore likely to also improve diagnostic performance in the context of peritoneal carcinomatosis. A larger cohort would probably be necessary to achieve statistical significance.

In conclusion, most cases of carcinomatosis-like appearance encountered on CT scan in the ED setting will be in patients with known malignancy, but of the new cases, there is a high PPV for carcinomatosis-like appearance to represent new peritoneal carcinomatosis rather than other non-cancer diagnoses. GYN malignancies were the most common primary lesion in women and GI primaries were the most common in men. The presence of high-volume ascites suggests a GYN primary or adenocarcinoma of unknown primary. In our cohort, usage of intravenous contrast improved accurate prediction of the primary lesion from 40 to 59%; however, this was not statistically significant. Future studies with a larger cohort size may be helpful in determining the role of intravenous contrast in accurately predicting the location of the primary lesion.