ORIGINAL ARTICLE: Clinical Endoscopy

Detection of peritoneal carcinomatosis by EUS fine-needle aspiration: impact on staging and resectability (with videos) Michael J. Levy, MD,1 Barham K. Abu Dayyeh, MD, MPH,1 Larissa L. Fujii, MD,1 Amy C. Clayton, MD,2 Jordan P. Reynolds, MD,2 Tercio L. Lopes, MD,1 Archana S. Rao, MD,1 Jonathan E. Clain, MD,1 Ferga C. Gleeson, MD,1 Prasad G. Iyer, MD,1 Michael L. Kendrick, MD,3 Elizabeth Rajan, MD,1 Mark D. Topazian, MD,1 Kenneth K. Wang, MD,1 Maurits J. Wiersema, MD,4 Suresh T. Chari, MD1 Rochester, Minnesota; Fort Wayne, Indiana, USA

Background: Peritoneal carcinomatosis (PC) greatly affects cancer staging and resectability. Objective: To compare the PC detection rate by using EUS and noninvasive imaging and to determine the impact on staging and resectability. Design: Retrospective study. Setting: Single tertiary-care referral center. Patients: A prospectively maintained EUS database was reviewed to identify patients who underwent EUS-guided FNA (EUS-FNA) of a peritoneal anomaly. Findings were compared with a strict criterion standard that incorporated cytohistologic, radiologic, and clinical data. Intervention: EUS-FNA of a peritoneal anomaly. Main Outcome Measurements: Safety and diagnostic yield. Results: Of 106 patients, a criterion standard was available in 98 (39 female patients; median age, 65 years). The sensitivity, specificity, and accuracy of EUS-FNA versus CT/magnetic resonance imaging (MRI) was 91% versus 28%, 100% versus 85%, and 94% versus 47%, respectively. In newly diagnosed cancer patients, peritoneal FNA upstaged 17 patients (23.6%). Of 32 patients deemed resectable by pre-EUS CT/MRI, 15 (46.9%) were deemed unresectable based solely on peritoneal FNA. The odds of FNA changing the resectability status remained highly significant after adjustment for cancer type, time between CT/MRI and EUS-FNA, and the quality of CT/MRI. The malignant appearance of the peritoneal anomaly but not the presence of ascites on EUS predicted a positive FNA finding (odds ratio 2.56; 95% confidence interval, 1.23-5.4 and odds ratio 0.83; 95% confidence interval, 0.4-1.8, respectively). There were 3 adverse events among 4 patients. Two of the patients developed abdominal pain and one each hypertensive urgency and pancreatitis. Limitations: Retrospective design, single-center, bias toward EUS as a diagnostic test. Conclusion: Peritoneal EUS-FNA appears to safely detect radiographically occult PC and improve cancer staging and patient care. (Gastrointest Endosc 2015;81:1215-24.)

Abbreviations: CI, confidence interval; EUS-FNA, EUS-guided FNA; MRI, magnetic resonance imaging; OR, odds ratio; PC, peritoneal carcinomatosis.

Copyright ª 2015 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2014.10.028

DISCLOSURE: All authors disclosed no financial relationships relevant to this article.

Received July 7, 2014. Accepted October 23, 2014.

Use your mobile device to scan this QR code and watch the author interview. Download a free QR code scanner by searching “QR Scanner” in your mobile device’s app store. www.giejournal.org

Current affiliations: Division of Gastroenterology and Hepatology (1), Departments of Laboratory Medicine and Pathology (2) and Surgery (3), College of Medicine, Mayo Clinic, Rochester, Minnesota, Lutheran Medical Group, Fort Wayne, Indiana (4), USA. Reprint requests: Michael J. Levy, MD, Mayo Clinic, Division of Gastroenterology and Hepatology, 200 First St. SW, Rochester, MN 55905.

Volume 81, No. 5 : 2015 GASTROINTESTINAL ENDOSCOPY 1215

Detection of peritoneal carcinomatosis

Accurate cancer staging is necessary to enhance prognostic determination, provide stage-appropriate therapy, and improve patient outcome. The presence of malignant cells within the peritoneal cavity is referred to as peritoneal carcinomatosis (PC), which represents an important site of distant metastasis for most tumor types.1,2 This process is variably manifest by the presence of thickening or solid malignant tumor masses or deposits on the peritoneal surface and/or within the omentum. Ascites is present in only approximately one-third of patients with PC.3 These changes may be identified by cross-sectional imaging or gross examination during surgery or at autopsy. The peritoneum and omentum may be involved by metastasis from a variety of cancers (eg, GI and ovarian), from primary tumors (eg, mesothelioma), or altered by benign processes including infectious (eg, tuberculosis) and inflammatory (eg, acute pancreatitis) conditions.4-13 For most cancers including pancreatic carcinoma, the finding of PC indicates a site of distant metastasis, a noncurative status, and poor prognosis with shortened survival.13 Tumor staging and resectability status are routinely assessed by CT or magnetic resonance imaging (MRI). However, the accuracy of these imaging modalities for detecting PC and determining the cause is limited.14 Therefore, tissue confirmation of PC is often pursued when there is no other confirmed evidence of unresectability. There is also a need for biopsy confirmation of PC with evolving treatment options and improved outcomes for some patients with gastric, colorectal, appendiceal, or ovarian cancer.15-17 A tissue diagnosis is also advocated given that 10% of patients with a known primary cancer have a second different type of cancer that involves the peritoneum or omentum.18 In addition, patients with a tumor of unknown origin may benefit from peritoneal or omental biopsy to help establish a tissue diagnosis and to identify the site of the primary malignancy.19 Furthermore, distinguishing between PC and peritoneal disease (eg, tuberculosis, lymphomatosis, pseudomyxoma peritonei, and sarcoidosis) is important and difficult to do based on imaging alone.20 Given that imaging cannot reliably distinguish a mass or thickening that involves the peritoneum alone, the omentum alone, or both structures, we use the term peritoneal anomaly to refer to potential pathology of these sites. EUS can detect primary tumors, liver metastases, lymphadenopathy, and other pathologies that are sometimes undetected with CT. We have similarly noted that EUS can also identify radiographically occult PC. Our aim in this study was to compare the detection rate of PC by EUS-guided FNA (EUS-FNA) with that of noninvasive pre-EUS cross-sectional imaging in a select cohort of patients who underwent EUS-FNA of a peritoneal anomaly and to determine its impact on staging and resectability. 1216 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 5 : 2015

Levy et al

PATIENTS AND METHODS The study was approved by the Mayo Clinic Institutional Review Board. We retrospectively reviewed a prospectively maintained EUS database to identify all patients who underwent EUS-FNA of a peritoneal anomaly for indications listed in Table 1. After informed consent was obtained, EUS was performed by using a curvilinear echoendoscope (Olympus GF-UC30P, GF-UC140P-AL5, GF-UCT180 or GF-UC160P-AT8; Olympus Medical Systems, Center Valley, Pa) and FNA was performed with a 22-gauge needle (Quick-Core; Wilson-Cook Medical Inc, Winston-Salem, NC). Biopsy was performed for peritoneal anomalies only when the FNA needle trajectory did not require traversal of a luminal cancer and for sites located remote from an extraluminal tumor (eg, pancreatic carcinoma). An on-site cytotechnologist was present for a cellular assessment of FNA specimen adequacy for all patients with simultaneous telecytology review with a cytopathologist at the discretion of the cytologist or endosonographer. Findings were compared with a strict diagnostic criterion standard. A patient was considered to have a malignant process involving the peritoneum or omentum (ie, PC) if within 60 days of the index EUS-FNA there was either (1) cytologic and/or histologic diagnosis of malignancy based on material obtained via EUS-FNA, percutaneous biopsy, surgical exploration, or autopsy or (2) a new radiographic (CT or MRI) abnormality of the peritoneum or omentum that the radiologist interpreted as clearly indicative of PC. Radiographic findings on CT/MRI that the radiologist deemed indicative of PC were variably phrased, but included nodular peritoneal implants, thickened omentum in the presence of ascites, malignant-appearing omental or peritoneal deposits, metastatic peritoneal or omental deposits, or simply the notation of finding typical or classic PC. Only cytology and histology interpretations of positive for malignancy were deemed malignant and all other (eg, suspicious, aypical, negative, inadequate cellularity) were regarded as benign. Each follow-up cross-sectional imaging study was compared with the one performed before EUS, typically by a radiologist subspecialized in GI imaging. Less decisive radiographic interpretations with diagnostic uncertainty were not included as part of the criterion standard. Although cross-sectional imaging does not accurately identify all patients with peritoneal carcinomatosis, it was used in the criterion standard diagnosis because CT and/or MRI are commonly performed in patients with a suspected intra-abdominal malignancy. A patient was considered to have a benign process involving the peritoneum or omentum if more than 60 days after the index EUS-FNA there was either (1) benign interpretation of subsequent cytology and/or histology samples obtained via EUS-FNA, percutaneous biopsy, surgical exploration, or autopsy or (2) lack of progression on imaging (CT or MRI). When the findings did not satisfy the www.giejournal.org

Levy et al

Detection of peritoneal carcinomatosis

TABLE 1. EUS indications, sex, and age No. of patient

Female, no. (%)

Median age, y

Age, y range

Primary pancreatic carcinoma

Indication

50

18 (36.0)

70

37-87

Other primary cancers

22

9 (40.9)

60

36-78

Cholangiocarcinoma

12

Pancreatic neuroendocrine

3

Pancreatic lymphoma

2

Gallbladder adenocarcinoma

2

Gastric adenocarcinoma

2

Duodenal adenocarcinoma

1 7 (43.8)

54

33-74

5 (50)

62

27-76

39 (39.8)

65

27-87

Cancer, miscellaneous

16

Recurrent pancreatic adenocarcinoma

4

Recurrent gastric adenocarcinoma

4

Recurrent esophageal adenocarcinoma

2

Recurrent cholangiocarcinoma

1

Recurrent rectal adenocarcinoma

1

Recurrent ovarian serous

1

Pancreas Met, NSCLC primary

1

Pancreas Met, colon adenocarcinoma primary

1

Cancer of unknown origin

1

Benign disorders

10

Pancreatitis (nonspecific)

5

Autoimmune pancreatitis

1

Immunoglobulin G4dassociated cholangiopathy

1

Pancreatic cyst

1

Gastric polyp

1

Postthrombotic syndrome

1

Total

98

Met, metastasis; NSCLC, non-small cell lung cancer.

criteria for PC or a benign process, the results were deemed indeterminate, and the patient was excluded from analysis. As previously indicated, when using follow-up CT/MRI findings as part of the criterion standard, we mandated clear CT or MRI evidence of PC. We did so to ensure the quality of the criterion standard. However, when comparing the performance characteristics of the initial CT/MRI with those of EUS-FNA, the benefit of the doubt was given to the radiographic findings and any mention of peritoneal or omental changes was considered diagnostic of PC. We did so to avoid any bias in favor of EUSFNA. Furthermore, for patients who underwent both CT and MRI, the mention of a peritoneal anomaly on either study was deemed diagnostic of PC. www.giejournal.org

Given the implications of diagnosing PC, we attempted to verify whether the EUS-FNA was obtained from a peritoneal anomaly or an unintended target. To do so, archived cytology slides and cell blocks were retrieved and reviewed by 2 expert cytopathologists for the first 50 patients. The slides were reviewed with the cytopathologists blinded to the diagnosis, and each pass was examined for evidence of FNA sampling of a structure other than the peritoneum or omentum, namely, parenchyma representing inadvertent sampling of a lymph node or surrounding structure including the pancreas, liver, kidney, spleen, or liver. In addition, although percutaneous peritoneal or omental biopsy often fails to yield cells that clearly confirm the target site, the slides were examined quantitatively and qualitatively Volume 81, No. 5 : 2015 GASTROINTESTINAL ENDOSCOPY 1217

Detection of peritoneal carcinomatosis

Levy et al

Figure 1. An 84-year-old woman presented with fatigue, nausea, and abdominal discomfort. CT revealed a locally advanced pancreatic tail cancer and peritoneal carcinomatosis (PC) with tumor implants on the liver surface (A, circle) and omentum with ascites (B, circle and arrow). C, EUS confirmed the presence of a hypoechoic 1-cm tumor deposit thought to be located within the greater omentum (arrow and circle). D, FNA confirmed peritoneal carcinomatosis (Diff-Quik; orig. mag. 600). Reprinted with permission from the Mayo Clinic.

for the presence of mesothelial cells, mucin, inflammatory changes, and fat necrosis, features that may be seen with peritoneal or omental biopsy. Finally, surgical pathology results were reviewed in those patients who underwent surgical biopsy or resection.

Statistical analysis

RESULTS Demographic information Between June 16, 2006, and November 15, 2013, 106 patients underwent EUS-FNA of a peritoneal anomaly. A criterion standard was available in 98 patients (39 female), with a median age of 65 (range 27-87) years. The final diagnoses are presented in Table 1 and include newly diagnosed pancreatic carcinoma (n Z 50), cholangiocarcinoma (n Z 12), pancreatic neuroendocrine tumor (n Z 3), pancreatic lymphoma (n Z 2), gallbladder carcinoma (n Z 2), gastric carcinoma (n Z 2), and duodenal carcinoma (n Z 1). EUS was also performed in 16 patients for cancer restaging and in 10 patients with a benign disorder.

Continuous variables are expressed as mean (SD) or median (range or interquartile range) and compared by using the Student t test for variables that are normally distributed or Mann-Whitney U test for nonparametric variables. Categorical variables were reported as frequency (%) and were compared by either a 2-tailed Fisher exact test or a Pearson c2 test, when appropriate. Multivariate logistic regression analysis was used to adjust for potential confounders including cancer type, time between CT/MRI and EUS-FNA (%30 vs O30 days), and the quality of CT/MRI (high quality based on %3 mm cuts and contrast administration vs low quality based on O3 mm cuts and/or intravenous contrast not administered). Furthermore, we used logistic regression analysis with backward stepwise variable selection to identify predictors of PC by EUS-FNA. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported. For all analyses, a significant 2-sided P value was set at !.05. SAS version 9.2 software (SAS Institute Inc, Cary, NC) and JMP Version 9 (SAS Institute Inc) were used for statistical analysis and modeling.

EUS findings were compared with noninvasive imaging that included CT alone (n Z 80), MRI alone (n Z 5), and both CT and MRI (n Z 13). Seventy-seven patients (78.6%) had a high-quality imaging study; the remainder of the patients underwent imaging, often at an outside institution, for a different indication in which the reason for the EUS was incidentally identified. The median time between the most recent CT or MRI and EUS-FNA was 6 days (range 1-133 days). Ascites was observed in 43 patients (43.9%) on CT/MRI. Of the 23 patients with a peritoneal anomaly identified by CT/MRI, 18 were true positive and 5 were false positive compared with the criterion

1218 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 5 : 2015

www.giejournal.org

Pre-EUS CT and MRI

Levy et al

Detection of peritoneal carcinomatosis

Figure 2. A 37-year-old man with primary sclerosing cholangitis (PSC) presented with new-onset jaundice and normal CA19-9. A, Magnetic resonance imaging revealed no evidence of a mass. B, Endoscopic retrograde cholangiography demonstrated PSC and a dominant stricture involving the intrapancreatic portion of the bile duct. Brush cytology and biopsy findings were negative. C, EUS revealed a 5-mm mass surrounding the dominant stricture and a 7  6-mm hypoechoic malignant-appearing lesion located deep to the duodenal wall in the region of the hepatoduodenal ligament that was thought to represent peritoneal carcinomatosis (PC). FNA confirmed adenocarcinoma involving the bile duct and PC. Surgical biopsy confirmed PC at the site indicated by EUS as well as several other locations. Reprinted with permission from the Mayo Clinic.

standard. The 5 false positives occurred in 2 patients undergoing initial cancer staging, and 3 were reported in patients with benign disorders. Ascites was noted in 16 of the 23 patients with a peritoneal anomaly, among whom 11 and 5 patients had a malignant and benign peritoneal anomaly, respectively. For the 7 of 23 patients without ascites, the peritoneal anomaly was malignant versus benign in 1 versus 6 patients, respectively. There was no difference in technical difficulty in performing FNA of the peritoneal anomaly in patients with and without ascites.

EUS examination Of the 98 patients with a peritoneal anomaly, ascites was observed by EUS in 55 patients (56.1%) (Figs. 1-4). The www.giejournal.org

median number of FNA passes obtained from the peritoneal anomaly was 3 (range 1-17) revealing positive cytology in 59 patients. The EUS appearance was thought to be indicative of PC in 67 patients, benign in 24 patients, and was either indeterminate or not stated in the remaining 7 patients. Given the retrospective nature of the report, detailed and exhaustive descriptions of the peritoneal anomalies were not always provided. However, most malignant-appearing peritoneal anomalies were described as being solid, hypoechoic masses, thickening, or nodularity involving the peritoneum and/or omentum (Video 1, available online at www.giejournal.org). In contrast, most peritoneal anomalies interpreted by EUS as likely benign were described by the absence of a discrete mass with Volume 81, No. 5 : 2015 GASTROINTESTINAL ENDOSCOPY 1219

Detection of peritoneal carcinomatosis

Levy et al

Figure 3. A 27-year old man presented with abdominal pain, anorexia, and 10-lb weight loss. CT revealed an indeterminate 12.5  12.3-cm low-density mass (A, asterisk) appearing to arise from the pancreatic body and a suspected inflammatory process involving the mesentery (B, arrow). The serum CA19-9 measured 352 U/mL (normal !55 U/mL). EUS revealed a 12  10-cm cyst abutting, but not appearing to arise from, the pancreas, spleen, or kidney. C, Ascites was seen along with benign-appearing greater omental folds that were isoechoic to the surrounding structures (arrows). Cyst FNA revealed viscous and milky-appearing fluid with a carcinoembryonic antigen level of 6218 ng/mL, a CA 19-9 level of 4,457,000 U/mL, cholesterol level of 99 mg/dL, and triglyceride level of 43 mg/dL. FNA results of the cyst and mesentery were both negative for malignancy. At surgery, the patient was found to have a duplication cyst with a surrounding inflammatory component involving the greater omentum. Reprinted with permission from the Mayo Clinic.

only minimal thickening or nodularity that was nearly isoechoic to surrounding structures (Video 2, available online at www.giejournal.org). On multivariate logistic regression, the malignant appearance on EUS was predictive of positive FNA and PC (OR 2.56; 95% CI, 1.23-5.4). The presence or absence of ascites did not predict a positive FNA (OR 0.834; 95% CI, 0.4-1.8). Among the 39 patients with a negative EUS-FNA cytology result, 33 (84.6%) represented a true-negative result versus 6 (15.4%) that were considered to be false negative.

In all 98 patients, the sensitivity, specificity, and accuracy of EUS-FNA versus CT/MRI in detecting PC was 91% versus 28%, 100% versus 85%, and 94% versus 47%, respectively. These percentages are similar for the 50 patients

with newly diagnosed pancreatic carcinoma (Table 2). Among the 72 newly diagnosed cancer patients, EUS upstaged 25 patients (34.7%), with 17 patients (23.6%) upstaged based solely on EUS-FNA detection of PC (Table 3). Among the 32 patients deemed resectable by CT/MRI, EUS converted 21 (65.6%) to unresectable and 15 of those 21 patients (46.9%) were based solely on EUS-FNA of the peritoneal anomaly. In the remaining 6 of the 21 patients, EUS-FNA cytology verified other evidence of unresectability including liver metastases (n Z 4) and celiac artery FNA (n Z 2). The effects of EUS-FNA of the peritoneal anomaly on the resectability status of the cancer was not affected by age (OR 1.014; 95% CI, 0.97-1.06), cancer type (OR 1.27; 95% CI, 0.11-14), time in days between CT/MRI and EUS-FNA (OR 1.009; 95% CI, 0.98–1.03), or the quality of the CT scan (OR 0.405; 95% CI, 0.13-1.25).

1220 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 5 : 2015

www.giejournal.org

Performance characteristics of EUS-FNA and CT/MRI and impact on staging and resectability

Levy et al

Detection of peritoneal carcinomatosis

confirmed PC in these 3 patients, and no false-positive EUS-FNA findings were reported. Among the remaining 16 patients with negative EUS-FNA cytology, surgical pathology confirmed the absence of PC in 14 patients and revealed a false-negative EUS-FNA result in 2 patients. In the latter 2 patients, the EUS appearance was considered typical for PC; however, the finding of scant cellularity within the FNA specimen in both patients may have accounted for the false-negative result.

Clinical follow-up Routine clinical follow-up was available in 91 patients (92.9%) within 5 days of EUS. Patients continued to be followed for 6.9 (interquartile range 1.0-16.6) months (Table 4). Four adverse events (4.1%) were reported including abdominal pain and vomiting (n Z 2) (1 patient was hospitalized for 1 day), intraprocedural hypertensive urgency (n Z 1), and mild pancreatitis in a patient who underwent both EUS and ERCP with sphincterotomy and stent placement for pancreatic carcinoma (n Z 1) that resulted in a 2-day hospitalization. Nineteen patients (19.4%) received antibiotics during the examination. No infectious adverse events were reported.

DISCUSSION Figure 4. A 33-year-old woman with autoimmune hepatitis, cirrhosis, decompensated liver disease, and portal vein and superior mesenteric vein thrombosis presented with recurrent gastric variceal bleeding. CT demonstrated a coarse and nodular liver, a large amount of ascites (A, asterisk), chronic portal vein thrombosis; numerous gastric, periesophageal, and retroperitoneal varices; and a recanalized umbilical vein with ventral abdominal hernia containing both omentum and mesenteric vessels (arrow). B, On EUS, the presence of ascites (asterisk) facilitated visualization of the omentum, which was otherwise normal appearing (arrow). Reprinted with permission from the Mayo Clinic.

Verification of peritoneal anomaly as the target site For the first 50 patients in whom the archived cytology slides and cell blocks were re-reviewed, 37 (74%) patients were found to have 1 or more potential markers to indicate a peritoneal or omental biopsy showing benign or reactive changes, including the presence of mesothelial cells, mucin, inflammatory changes, and fat necrosis. Only 1 patient (2%) had evidence of FNA sampling of a structure other than the peritoneum or omentum, namely, the presence of benign pancreatic parenchyma. That patient had a negative FNA result, and therefore the incorrectly targeted FNA did not negatively affect care. In the remaining 12 patients, there was neither confirmation of peritoneal/omental FNA nor evidence of sampling of an unintended site. Among the entire cohort, 19 patients underwent surgical biopsy or resection, including 3 patients despite EUS-FNA documenting the presence of PC. The surgical findings www.giejournal.org

The peritoneum is a serous membrane that lines the wall of the abdominal cavity and pelvis (parietal peritoneum) and intraperitoneal organs (visceral peritoneum).21,22 It is composed of a single layer of flat mesothelial cells that are supported by underlying submesothelial connective tissue consisting of blood vessels, lymphatics, fat cells, and inflammatory cells. Regions of folding or double layers of the peritoneum are referred to as either omentum or mesentery, which serves to support the abdominal viscera. The term true mesentery is applied to regions of the mesentery that directly connect to the posterior peritoneal wall and include the (1) small-bowel mesentery, (2) transverse mesocolon, and (3) sigmoid mesentery. In contrast, the term specialized mesentery refers to mesenteries that do not directly connect to the posterior peritoneal wall. The specialized mesenteries include the (1) greater omentum that connects the stomach to the colon, (2) lesser omentum that connects the stomach to the liver, and (3) mesoappendix that connects the appendix to the ileum (Fig. 5). Normally, in the absence of disease, there is a small amount of peritoneal fluid that continuously circulates throughout the abdominal cavity. The peritoneum and omentum serve to contain disease (eg, appendicitis), but also provide a conduit for disease spread (eg, PC). PC most often results from cancers of the stomach, colon, rectum, pancreas, ovary, uterus, cervix, and lymphoproliferative malignancies.4-13 For pancreatic carcinoma, PC represents the second most common site of distant metastasis, after the liver, Volume 81, No. 5 : 2015 GASTROINTESTINAL ENDOSCOPY 1221

Detection of peritoneal carcinomatosis

Levy et al

TABLE 2. Comparison of CT/MRI and EUS-FNA in detecting peritoneal carcinomatosis Sensitivity

Specificity

Accuracy

CT/MRI (N Z 98), no. (%)

18/65 (28)

28/33 (85)

46/98 (47)

EUS-FNA (N Z 98), no. (%)

59/65 (91)

33/33 (100)

92/98 (94)

CT/MRI (pancreatic carcinoma, n Z 50), no. (%)

11/39 (28)

10/11 (91)

21/50 (42)

EUS-FNA (pancreatic carcinoma, n Z 50), no. (%)

35/39 (90)

11/11 (100)

46/50 (92)

EUS-FNA, EUS-guided FNA.

TABLE 3. Impact of EUS-FNA detection of peritoneal carcinomatosis on the CT/MRI stage

Any EUS finding resulting in upstage

EUS-FNA diagnosis of PC exclusively up-stage

Any EUS finding convert to unresectable

EUS-FNA diagnosis of PC exclusively convert to unresectable

Pancreatic carcinoma (n Z 50), no. (%)

19/50 (38)

12/50 (24)

16/22 (73)

11/22 (50)

Other primary cancers (n Z 22), no. (%)

6/22 (27)

5/22 (23)

5/10 (50)

4/10 (40)

25 (35)

17 (24)

21 (66)

15 (47)

Indication

Total, no. (%)

and indicates stage IV disease, a noncurative status, poor prognosis, and shortened survival.13,23 PC should be distinguished from positive peritoneal lavage, which represents fluid cytology that is positive for malignancy in the absence of even minute peritoneal or omental tumor deposits. The distinction may have more than just anatomic importance. Although the presence of either PC or positive peritoneal lavage is indicative of stage IV disease,24 the prognostic significance and impact on survival may differ. PC is universally regarded as an indicator of incurable disease with shortened survival; however, the impact of cytologically or molecularly positive peritoneal lavage is debated. Some report worse prognosis and shortened survival similar to PC,25-27 whereas others note minimal or no impact on survival.26,28-30 Efforts in pancreatic carcinoma patients to resect the primary tumor and debulk tumor deposits associated with PC have increased the risk of adverse events without providing a survival advantage, thereby leading surgeons to abandon this practice.31,32 Although abbreviated for both, the survival in patients with isolated PC is significantly shorter compared with patients having isolated liver metastases,13 with even shorter survival for patients with simultaneous PC and hepatic metastasis.33 Although most pancreatic carcinoma patients with PC have coexisting hepatic metastases, PC represents an isolated site of distant disease in approximately 30% to 40% of patients.13,27 The limited published data indicate that PC is present in approximately 20% to 50% of pancreatic carcinoma patients at the time of death,3,23 approximately 20% at staging laparoscopy, and 10% at the time of initial diagnosis.13,34 The latter study demonstrated an increased incidence of

PC over the 15-year study period with rates of 6.8% and 11.8% for 1995 to 1999 versus 2005 to 2009, respectively. The discrepant rate likely reflects the increased use and higher quality of imaging over time. However, the rate of PC is likely much greater owing to the difficulty in detecting small tumor implants. Peritoneal and omental pathology are often difficult to detect and accurately diagnose by using CT and MRI because of the nonspecific and overlapping imaging features. The peritoneum and omentum are not clearly discernible on EUS in the absence of disease because they are mostly isoechoic to surrounding structures and because they are folded upon themselves, further blending with surrounding structures. However, peritoneal or omental thickening or tumor deposits may be seen, especially when accompanied by ascites, which provides a contrasting echodensity and echopattern and allows the omentum to unfold and become easily discernible. Case reports and small case series have in total described fewer than 10 patients who have undergone EUS-FNA to diagnose PC resulting from cancer of the pancreas, stomach, or an unknown primary cancer or for diagnosis of tuberculous disease.9,10,12,35-37 Our findings further demonstrate that among patients with an array of tumor types and benign disease processes, EUS may provide an advantage over noninvasive imaging given the ability to identify small peritoneal and omental implants, often in the absence of ascites. Although the malignant appearance of the peritoneal anomaly on EUS was significantly predictive of a positive cytology, we advocate FNA to confirm PC rather than relying on the appearance alone given the impact on prognosis and patient care when diagnosing PC. EUS has proven utility as a complementary tool for noninvasive imaging for enhancing

1222 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 5 : 2015

www.giejournal.org

Levy et al

Detection of peritoneal carcinomatosis

TABLE 4. Follow-up duration and subsequent surgery Follow-up, mo, median (IQR)

Subsequent surgery

Pancreatic carcinoma (n Z 50)

Indication

5.4 (0.5-11.6)

8 (16.0)

Other primary cancers (n Z 22)

6.7 (0.6-12.0)

8 (36.4)

Cancer, miscellaneous (n Z 16)

8.5 (1.0-21.1)

1 (6.3)

Benign disorder (n Z 10)

12.1 (9.1-36.6)

4 (40.0)

Total

6.9 (1.0-16.6)

21 (21.4)

IQR, Interquartile range.

Figure 5. Artist rendering of the normal anatomy as demonstrated on sagittal (A) and coronal planes (B, C). Reprinted with permission from the Mayo Clinic.

primary tumor (T), nodal (N), and metastatic (M) staging. We believe that our findings highlight the potential of EUS to further enhance cancer staging accuracy through the detection of radiographically occult PC. The increased staging accuracy may help refine treatment strategies, improve selection of surgical candidates or patients for downstaging strategies, guide the use of palliative interventions, and aid the interpretation of clinical trials. No falsepositive results were identified in our patient cohort, but false-negative results occurred in 12.5% (2 of 16 patients), highlighting the importance that a negative EUS-FNA of a peritoneal anomaly does not necessarily exclude malignancy, and additional biopsy specimens should be obtained in patients with a high clinical suspicion of PC. There are several study limitations. We were able to detect and confirm PC in 41 patients with EUS-FNA that was not identified with CT/MRI. However, our study included a select patient population in whom study inclusion was based on EUS detection of a peritoneal or omental anomaly. It is likely that many additional patients seen at our institution with typical cross-sectional imaging features of PC were never referred for EUS evaluation during the study time period, biasing our results in favor of EUS. Furthermore, for the 18 patients found to have PC www.giejournal.org

on CT/MRI, the initial imaging provided a clue to the endosonographer, thereby aiding EUS detection. An accurate comparison of the performance characteristics of EUSFNA to CT/MRI requires prospective evaluation within a consecutive patient cohort that includes all comers. Likewise, the number of patients who were upstaged and converted from resectable to unresectable was substantial. These rates do not reflect the results anticipated among an unselected cohort. Finally, given the impact of PC on prognosis and patient care, it is critical to ensure that the biopsy specimens are obtained from the intended site. It is impossible to have surgical confirmation in all patients given the lack of a surgical indication for the majority of patients with advanced malignant or benign disease. However, our use of a strict criterion standard, the exclusion of indeterminate cases, the findings of the cytopathology rereview, and results for those patients who did go to surgery all suggest accurate targeting of the peritoneal anomaly. In summary, PC is a relatively common and important site of metastasis in cancer patients. Our findings support the contention that EUS-guided biopsy of peritoneal and omental masses or thickening is a safe and accurate method for diagnosing radiographically occult PC, particularly in patients without ascites. Biopsy confirmation of PC allows Volume 81, No. 5 : 2015 GASTROINTESTINAL ENDOSCOPY 1223

Detection of peritoneal carcinomatosis

Levy et al

1. Ceelen WP, Bracke ME. Peritoneal minimal residual disease in colorectal cancer: mechanisms, prevention, and treatment. Lancet Oncology 2009;10:72-9. 2. Lemmens VE, Klaver YL, Verwaal VJ, et al. Predictors and survival of synchronous peritoneal carcinomatosis of colorectal origin: a population-based study. Int J Cancer 2011;128:2717-25. 3. Blastik M, Plavecz E, Zalatnai A. Pancreatic carcinomas in a 60-year, institute-based autopsy material with special emphasis of metastatic pattern. Pancreas 2011;40:478-80. 4. Levitt RG, Sagel SS, Stanley RJ. Detection of neoplastic involvement of the mesentery and omentum by computed tomography. AJR Am J Roentgenol 1978;131:835-8. 5. Cooper C, Jeffrey RB, Silverman PM, et al. Computed tomography of omental pathology. J Comput Assist Tomogr 1986;10:62-6. 6. Sistrom CL, Abbitt PL, Feldman PS. Ultrasound guidance for biopsy of omental abnormalities. J Clin Ultrasound 1992;20:27-36. 7. Rioux M, Michaud C. Sonographic detection of peritoneal carcinomatosis: a prospective study of 37 cases. Abdom Imaging 1995;20: 47-51; discussion 56-7. 8. Spencer JA, Swift SE, Wilkinson N, et al. Peritoneal carcinomatosis: image-guided peritoneal core biopsy for tumor type and patient care. Radiology 2001;221:173-7. 9. Rana SS, Bhasin DK, Srinivasan R, et al. Endoscopic ultrasound-guided fine needle aspiration of peritoneal nodules in patients with ascites of unknown cause. Endoscopy 2011;43:1010-3. 10. Rana SS, Bhasin DK, Srinivisan R, et al. Endoscopic ultrasound fine-needle aspiration of peritoneal deposits for diagnosis of tubercular peritonitis in a cirrhotic patient with ascites. Endoscopy 2010;42(Suppl 2):E306-7. 11. Layfield LJ, Gopez EV. Percutaneous image-guided fine-needle aspiration of peritoneal lesions. Diagn Cytopathol 2003;28:6-12. 12. Rana SS, Bhasin DK, Gupta N. Endoscopic ultrasound-guided fine-needle aspiration of peritoneal deposit in otherwise resectable pancreatic cancer. Clin Gastroenterol Hepatol 2012;10:e60. 13. Thomassen I, Lemmens VEPP, Nienhuijs SW, et al. Incidence, prognosis, and possible treatment strategies of peritoneal carcinomatosis of pancreatic origin: a population-based study. Pancreas 2013;42:72-5. 14. Gonzalez-Moreno S, Gonzalez-Bayon L, Ortega-Perez G, et al. Imaging of peritoneal carcinomatosis. Cancer J 2009;15:184-9.

15. Dovern E, de Hingh IHJT, Verwaal VJ, et al. Hyperthermic intraperitoneal chemotherapy added to the treatment of ovarian cancer. A review of achieved results and complications. Eur J Gynaecol Oncol 2010;31:256-61. 16. Gill RS, Al-Adra DP, Nagendran J, et al. Treatment of gastric cancer with peritoneal carcinomatosis by cytoreductive surgery and HIPEC: a systematic review of survival, mortality, and morbidity. J Surg Oncol 2011;104:692-8. 17. Munoz-Casares FC, Rufian S, Arjona-Sanchez A, et al. Neoadjuvant intraperitoneal chemotherapy with paclitaxel for the radical surgical treatment of peritoneal carcinomatosis in ovarian cancer: a prospective pilot study. Cancer Chemother Pharmacol 2011;68:267-74. 18. Souza FF, Mortele KJ, Cibas ES, et al. Predictive value of percutaneous imaging-guided biopsy of peritoneal and omental masses: results in 111 patients. AJR Am J Roentgenol 2009;192:131-6. 19. Peter S, Eltoum I, Eloubeidi MA. EUS-guided FNA of peritoneal carcinomatosis in patients with unknown primary malignancy. Gastrointest Endosc 2009;70:1266-70. 20. Diop AD, Fontarensky M, Montoriol PF, et al. CT imaging of peritoneal carcinomatosis and its mimics. Diagn Interv Imaging 2014;95:861-72. 21. Yoo E, Kim JH, Kim M-J, et al. Greater and lesser omenta: normal anatomy and pathologic processes. Radiographics 2007;27:707-20. 22. Desai G, Filly RA. Sonographic anatomy of the gastrohepatic ligament. J Ultrasound Med 2010;29:87-93. 23. del Castillo CF, Warshaw L. Peritoneal metastases in pancreatic carcinoma. Hepatogastroenterology 1993;40:430-2. 24. Edge SB, Byrd DR, Compton CC, et al. AJCC Cancer Staging Manual. New York (NY): Springer; 2010. 25. Makary MA, Warshaw AL, Centeno BA, et al. Implications of peritoneal cytology for pancreatic cancer management. Arch Surg 1998;133:361-5. 26. Ferrone CR, Haas B, Tang L, et al. The influence of positive peritoneal cytology on survival in patients with pancreatic adenocarcinoma. J Gastrointest Surg 2006;10:1347-53. 27. Clark CJ, Traverso LW. Positive peritoneal lavage cytology is a predictor of worse survival in locally advanced pancreatic cancer. Am J Surg 2010;199:657-62. 28. Meszoely IM, Lee JS, Watson JC, et al. Peritoneal cytology in patients with potentially resectable adenocarcinoma of the pancreas. Am Surg 2004;70:208-13; discussion 213-4. 29. Yamada S, Takeda S, Fujii T, et al. Clinical implications of peritoneal cytology in potentially resectable pancreatic cancer: positive peritoneal cytology may not confer an adverse prognosis. Ann Surg 2007;246: 254-8. 30. Yoshioka R, Saiura A, Koga R, et al. The implications of positive peritoneal lavage cytology in potentially resectable pancreatic cancer. World J Surg 2012;36:2187-91. 31. Shrikhande SV, Kleeff J, Reiser C, et al. Pancreatic resection for M1 pancreatic ductal adenocarcinoma. Ann Surg Oncol 2007;14:118-27. 32. Seelig SK, Burkert B, Chromik AM, et al. Pancreatic resections for advanced M1-pancreatic carcinoma: the value of synchronous metastasectomy. HPB Surg 2010;2010:579672. 33. Morizane C, Okusaka T, Morita S, et al. Construction and validation of a prognostic index for patients with metastatic pancreatic adenocarcinoma. Pancreas 2011;40:415-21. 34. White R, Winston C, Gonen M, et al. Current utility of staging laparoscopy for pancreatic and peripancreatic neoplasms. J Am Coll Surg 2008;206:445-50. 35. Lee YT, Ng EKW, Hung LCT, et al. Accuracy of endoscopic ultrasonography in diagnosing ascites and predicting peritoneal metastases in gastric cancer patients. Gut 2005;54:1541-5. 36. Schmulewitz N, Singh P, Safa M, et al. Diagnosis of peritoneal carcinomatosis by EUS-guided FNA. Gastrointest Endosc 2007;66:825-6; discussion 826. 37. Matsushita M, Uchida K, Okazaki K. Diagnosis of peritoneal carcinomatosis: transgastric versus transrectal EUS-guided FNA or percutaneous paracentesis. Gastrointest Endosc 2008;67:1211-2; author reply 1212.

1224 GASTROINTESTINAL ENDOSCOPY Volume 81, No. 5 : 2015

www.giejournal.org

patients with known primary cancers to be treated in a stageappropriate manner. In addition, the finding may reveal a second primary malignancy in patients who, without a biopsy, would be inappropriately managed. Also, in patients in whom there is uncertainty as to the presence of malignancy, EUS-FNA of a peritoneal anomaly may confirm malignancy or suggest a benign process. However, because most omental and peritoneal masses are malignant, a negative FNA result does not exclude PC, thereby mandating close clinical and imaging follow-up. Diagnosis and selection of therapy are aided by having a knowledge of the peritoneal and omental anatomy, by understanding the disease spectrum that may involve the omentum, and by identifying characteristic EUS features. Additional work is needed to more fully characterize the EUS appearance of PC. Further study is also needed in an unselected cohort of patients with an array of tumor types and benign diseases to determine the accuracy of EUS versus conventional imaging for detecting PC and the resulting impact on staging, resectability, patient care, and outcome. REFERENCES