Peritoneal Carcinomatosis: Role of ¹⁸F-FDG PET

MATERIALS AND METHODS

Imaging

All ¹⁸F-FDG PET scans were performed on a high-resolution dedicated whole-body PET camera, GE ADVANCE (General Electric Medical Systems), consisting of 336 bismuth germanate detector units arranged to form 18 rings. The average axial resolution (full width at half maximum) is 4.0 mm at r = 0 cm, 5.5 mm at r = 10 cm, and 6.6 mm at r = 20 cm. Total system sensitivity is 6 kcps/Bq/mL (223 kcps/μCi/mL) with septa in and 32 kcps/Bq/mL (1,200 kcps/μCi/mL) with septa out (11). Scanning was generally performed from the neck to the proximal thighs. In 1 patient, scanning included the chest and upper abdomen only. Emission studies were commenced approximately 45 min after the intravenous administration of 370 MBq ¹⁸F-FDG followed by a transmission study covering the same area. Studies were reconstructed using filtered backprojection alone or with iterative reconstruction and segmented attenuation correction. All patients had CT scans of the abdomen and pelvis, which were correlated with the PET scan.

Peritoneal tumor was suspected on ¹⁸F-FDG PET when certain focal or diffuse metabolic abnormalities were identified in the abdomen or pelvis. Focal abnormalities were considered positive for peritoneal tumor if discrete foci of increased ¹⁸F-FDG metabolism were located randomly and anteriorly within the abdomen or dependently within the pelvis, unrelated to solid viscera or nodal stations (Fig. 1A). The anterior location of these lesions within the abdomen was often best appreciated on the sagittal image (Fig. 1B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

An 80-y-old woman with stage IIIC ovarian carcinoma after 6 cycles of neoadjuvant chemotherapy. Preoperative ¹⁸F-FDG PET scan shows focal pattern of peritoneal carcinomatosis. (A) Coronal section demonstrates hypermetabolic foci distributed randomly throughout abdomen and pelvis, unrelated to solid viscera or nodal stations. (B) Sagittal section demonstrates their typical anterior location. After debulking procedure, including total abdominal hysterectomy and bilateral salpingo-oophorectomy and partial omentectomy, peritoneal implants progressed on subsequent PET scans.

Foci of ¹⁸F-FDG uptake confined only to the posterior abdomen or pelvis, clustered in the region of the celiac, paraaortic, or iliac nodal stations, suggested nodal metastases and were not included in the study.

The pattern of diffuse, low-grade glucose hypermetabolism spreading uniformly throughout the abdomen and pelvis obscuring visceral outlines—particularly the normal serpiginous pattern of the large and small bowel and physiologic hepatic and splenic uptake—was also suggestive of peritoneal tumor (Fig. 2).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

A 47-y-old woman with poorly differentiated adenocarcinoma of stomach (linitis plastica type), 6 mo after sub-total gastrectomy, with persistent nausea and vomiting, referred for exclusion of disease recurrence. ¹⁸F-FDG PET scan, coronal section, demonstrates metabolic pattern characteristic of diffuse peritoneal carcinomatosis with uniform low-grade ¹⁸F-FDG uptake throughout entire abdomen and pelvis obscuring normal, discrete, visceral outlines, particularly bowel, liver, and spleen. Her demise was precipitated by gastric outlet obstruction and progressive cachexia.

CT scans were suggestive of peritoneal tumor if they demonstrated infiltration, thickening, or studding of the peritoneum, bowel wall, mesentery, or omentum with or without ascites. Neither ascites alone nor the presence of abdominal or pelvic lymphadenopathy was considered representative of peritoneal carcinomatosis.

A PET scan was true-positive if findings suggestive of peritoneal tumor, as described above, were confirmed by a positive tissue diagnosis or clinical or colocated radiographic abnormalities on subsequent follow-up; a PET scan was true-negative if neither the PET study, histology, radiographic, nor clinical follow-up demonstrated peritoneal carcinomatosis. A PET scan was false-positive if it demonstrated findings suggestive of peritoneal carcinomatosis, with negative findings on pathology and follow-up; a PET scan was false-negative if either histology or follow-up demonstrated evidence of peritoneal tumor, whereas the PET study remained negative.

A CT scan was true-positive if it demonstrated findings of peritoneal disease that were confirmed by peritoneal pathology; other radiographic findings, including PET; or clinical follow-up. A false-positive CT scan was characterized by positive findings of peritoneal tumor that were not supported by other evidence of peritoneal disease. A true-negative CT scan occurred when other correlative investigations failed to identify peritoneal disease. If a CT scan was negative for peritoneal disease, whereas other investigations such as peritoneal pathology, radiographic, PET or clinical follow-up suggested otherwise, this constituted a false-negative CT scan.

RESULTS

Peritoneal tumor was suspected in 24 of the 88 patients on the basis of the ¹⁸F-FDG PET or CT scan. A final diagnosis of peritoneal tumor was made in 23 patients based on peritoneal biopsy or cytology of ascitic fluid or on radiographic or clinical follow-up. Seventeen of the 23 patients had histologic confirmation of peritoneal disease. A further 6 patients (2 with a negative biopsy, 4 without concurrent biopsy) were diagnosed with peritoneal disease on the basis of radiographic or clinical follow-up. Of these 6 patients, 3 had progressive peritoneal disease demonstrated on serial CT scanning alone, 2 patients had progressive CT scans in addition to worsening abdominal symptomatology and positive aspiration cytology performed at 13 and 20 mo after the PET scan, and 1 patient had a positive ¹¹¹In-satumomab pendetide (OncoScint CR/OV; Cytogen Corp.) scan.

¹⁸F-FDG PET was positive in 13 of the 23 patients who were ultimately diagnosed with peritoneal disease: confirmed in 8 patients by biopsy, by radiographic follow-up in 4 patients, and in 1 patient by combined clinical and radiographic follow-up. In 1 patient, although peritoneal tumor was suspected on ¹⁸F-FDG PET, other investigations did not confirm this suspicion. The other 10 of the 23 patients who were diagnosed with peritoneal disease demonstrated a negative PET scan. These findings yielded 13 true-positive, 10 false-negative, and 1 false-positive result on PET imaging.

CT was positive for peritoneal carcinomatosis in 10 of the 23 patients diagnosed with peritoneal disease, this result being confirmed by biopsy and by radiographic and combined clinical and radiographic follow-up in 7, 2, and 1 patient, respectively. CT was negative for peritoneal carcinomatosis in the remaining 13 of the 23 patients, yielding 10 true-positive, 1 true-negative, 13 false-negative, and no false-positive results on CT imaging.

When PET was evaluated together with CT, there were 18 true-positive, 5 false-negative, and 1 false-positive result.

Sensitivities for PET, CT, and both modalities together were 57%, 43%, and 78%, respectively. The positive predictive values for ¹⁸F-FDG PET and CT in detecting peritoneal carcinomatosis in our patient group were 93% and 100%. When PET and CT scans were evaluated together, not only was sensitivity increased but also a high positive predictive value of 95% was maintained (Table 1).

A total of 10 patients had congruent PET and CT findings, 5 positive and 5 negative for peritoneal disease. The remaining 14 patients had incongruent findings on PET and CT. Of this group, PET was positive in 9 patients who had a negative CT scan (only 1 of these PET scans was false-positive), and CT was true-positive in 5 patients with a negative PET scan. Had either the PET scan or the CT scan been evaluated alone, 13 patients may have been misdiagnosed on the basis of their imaging findings.

The peritoneum was not surgically explored in 64 of the initial 88 patients because of the retrospective nature of this study. As such, biopsy confirmation of peritoneal normality was not available, and neither a true-negative nor a false-positive status in this group could be accurately established.

DISCUSSION

Peritoneal tumor deposits are often low in volume, are individually small, and may be few in number and variable in their gross morphology, developing either as discrete masses or smaller nodules, studs, flat plaques or large sheets of tumor only several cells thick (12). Early disease may be microscopic and limited only to the ascitic fluid. More advanced disease may extend along the parietal peritoneum, stud visceral surfaces, or cause omental or mesenteric caking (13). Peritoneal tumor seeding may be a result of hematogenous, lymphatic, or direct local spread (13). Such variability in tumor morphology, site, and mode of spread within the abdomen may help to explain why the diagnosis of peritoneal carcinomatosis remains a challenge.

Open laparotomy with peritoneal biopsy is the gold standard for diagnosis of peritoneal carcinomatosis, because the peritoneum may be assessed visually and examined carefully by hand. Although a greater number of diagnostic biopsies may be performed and ascitic fluid easily sampled, this technique is invasive. Some authors have described metastases along the laparoscopic tracts in some patients, suggesting seeding at the time of surgery (14). Laparascopic staging is commonly used in staging of patients at diagnosis (4), but its invasive nature suggests a limited role in regular patient follow-up.

Conventional CT scanning, currently the preoperative imaging procedure of choice for diagnosis of peritoneal metastases, has varying sensitivity with study results ranging widely from 17% (5) to 54% (6), but usually falling in the lower range (7,13). More reliable detection of tumor with CT may depend on site, size, and tumor morphology as well as other features. Gryspeerdt et al. reported an overall sensitivity of 38% for CT in the diagnosis of peritoneal metastases with higher sensitivities for tumors located paravertebrally or in the right paracolic gutter. Nodular lesions were more easily detected than flat lesions (100% vs. 21% detection) and were less commonly detected if associated with omental metastases, ascites, or history of prior abdominal surgery (7). Adachi et al. reported (15) that peritoneal and nodal metastases could be overlooked on CT examinations in patients with advanced gastric cancer. Double-contrast MRI (6), CT peritoneography (6), and video-laparoscopic staging (16) have yielded higher sensitivity, specificity, and accuracy than conventional CT. However, CT remains widely available at most centers, is less expensive than other imaging modalities, and is less invasive than surgical procedures.

The role of MRI in peritoneal carcinomatosis is still under evaluation. Recent reports (17,18) suggest that reasonable accuracy can be achieved, however, often at a cost of significant patient preparation and use of specific imaging sequences. As such, it has so far been unable to replace CT as the gold standard for preoperative evaluation of peritoneal carcinomatosis. Low et al. (6) report a sensitivity of peritoneal metastasis detection of 85% with delayed, gadolinium-enhanced, breath-hold, fast, mutiplanar sequence with fat saturation with gradient-recalled acquisition in the steady state compared with a sensitivity of 33% for standard T1-weighted sequences. Others (8) have reported sensitivities of 91% when artifact reduction and peritoneal contrast enhancement are performed. One study reports comparable MRI and CT results in the detection of peritoneal spread in pancreatic (10) and ovarian carcinoma with accuracy for both of approximately 70% (19). Others report complete nonvisualization of peritoneal implants in ovarian cancer with CT or MRI, either in the mesentery or in the small bowel wall (20).

Although the role of ¹⁸F-FDG PET in the evaluation of the peritoneal tumors is not yet established, experience is increasing, largely in metastatic tumors. Tanaka et al. (21) have demonstrated sensitivities far superior to CT in the evaluation of colorectal tumor peritoneal recurrence in a retrospective study. The results in the evaluation of other tumors (18,22) appear similarly promising.

Our study demonstrates a limited sensitivity for detection of peritoneal carcinomatosis with ¹⁸F-FDG PET or CT scanning alone (57% and 43%, respectively). Adding PET to CT increased sensitivity further (78.0%) than could have been achieved with either modality, while maintaining a high positive predictive value (95%).

This result strongly suggests that, in an appropriate clinical setting, an abnormal metabolic pattern in the peritoneum can rule in a diagnosis of peritoneal disease, even though little can be said about a negative study.

One of our patients had a false-positive ¹⁸F-FDG PET scan, with sampling error most likely responsible for this result. In this patient, the PET scan was strongly positive for nodular disease at multiple sites in the abdomen, including the peritoneum. Only a peritoneal tap had been performed without biopsy. Accurate CT-guided localization and biopsy may have confirmed the true-positive status of this patient and, hence, improved sensitivity as well as increasing the positive predictive value of PET to 100%.

Familiarity with the patient’s history as well as the metastatic characteristics of tumor—whether the peritoneum is a preferred site—and histologic staging of the primary tumor at diagnosis may all help to determine the pre-test likelihood of peritoneal metastasis.

Knowledge of the diverse morphologies of peritoneal tumor also is critical and, in our patients, contributed to the recognition of 2 distinct patterns of abdominal ¹⁸F-FDG uptake, typical of nodular and diffuse peritoneal disease on pathologic examination. We believe that familiarization with these 2 patterns of glucose metabolism is important in increasing detectability and diagnostic accuracy of peritoneal carcinomatosis with PET and, particularly, to distinguish tumor from physiologic activity of other abdominal viscera.

The study had some limitations. Because of the retrospective nature of this study, we were unable to obtain histologic data on all patients. In fact, 64 of the initial 88 patients were excluded on the basis of missing peritoneal biopsy or ascitic tap, with follow-up also inadequate in many. Although most patients in this group did not have a PET scan with suspicious finding(s), CT abnormalities did occur and they could not be corroborated with a suitable gold standard. Without an accurate true-negative value, neither specificity nor negative predictive value could be determined. To resolve these issues, a large, prospective study is a goal of future research in this area, including other tumors where peritoneal implants are more common, such as colorectal carcinoma. Indeed, up to 50% of all patients undergoing further surgery for recurrent colorectal carcinoma have been reported with peritoneal seeding (23). This study would also aim to include larger patient numbers; accurate, contemporaneous, and targeted biopsies in all patients; and CT scans timed to the PET images or acquired simultaneously with PET/CT hybrid cameras to provide more accurate information and, therefore, greater statistical power.

CONCLUSION

Evaluation of peritoneal carcinomatosis with ¹⁸F-FDG PET is possible and rewarding. Our study demonstrates that ¹⁸F-FDG PET adds substantially to CT scanning in the detection of this disease. It is a useful diagnostic tool when peritoneal biopsy is either unavailable or inappropriate. We have identified 2 distinct patterns of glucose metabolism that appear to predict either nodular or diffuse peritoneal pathology and should alert the clinician to the possibility of peritoneal carcinomatosis in the appropriate clinical setting. We believe that awareness of these metabolic patterns is crucial for accurate interpretation of both abnormal and physiologic metabolic findings in the abdomen and pelvis when cancer patients are staged with ¹⁸F-FDG PET.