Abdominal Imaging

, Volume 33, Issue 1, pp 87–93

Staging of peritoneal carcinomatosis: enhanced CT vs. PET/CT

Authors

    • Department of RadiologyInstitut Gustave-Roussy
  • Sophie Leboulleux
    • Department of Nuclear medicineInstitut Gustave-Roussy
  • Anne Auperin
    • Department of StatisticInstitut Gustave-Roussy
  • Diane Goere
    • Department of SurgeryInstitut Gustave-Roussy
  • David Malka
    • Department of MedicineInstitut Gustave-Roussy
  • Jean Lumbroso
    • Department of Nuclear medicineInstitut Gustave-Roussy
  • Martin Schumberger
    • Department of Nuclear medicineInstitut Gustave-Roussy
  • Robert Sigal
    • Department of RadiologyInstitut Gustave-Roussy
  • Dominique Elias
    • Department of SurgeryInstitut Gustave-Roussy
Article

DOI: 10.1007/s00261-007-9211-7

Cite this article as:
Dromain, C., Leboulleux, S., Auperin, A. et al. Abdom Imaging (2008) 33: 87. doi:10.1007/s00261-007-9211-7
  • 353 Views

Abstract

Purpose

To assess and compare the performance of CT and 18F-FDG-PET/CT in the evaluation of peritoneal carcinomatosis (PC).

Method and materials

Thirty consecutive patients with PC and scheduled for a surgery underwent a CT of the abdomen and pelvis and a whole-body 18F-FDG PET/CT. The extent of PC was assessed precisely using the peritoneal cancer index combining the distribution of tumor throughout 11 abdominopelvic regions with a lesion size score. CT and PET/CT imaging results were compared in all patients with intraoperative findings using an interclass correlation test.

Results

The presence of PC was correctly determined on CT and PET/CT in 23/28 and 16/28 patients, respectively. The extent of PC was understaged with CT and PET/CT in 27 patients and overstaged with CT and PET/CT in 1 and 2 patients, respectively. The interclass correlation was 0.53 (moderate) between CT and surgery and 0.12 (low) between PET/CT and surgery. The interclass correlation was higher for mucinous tumor (0.63) than for non-mucinous (0.16) on CT imaging whereas no difference was found in PET/CT.

Conclusion

The intraperitoneal assessment of the extent of carcinomatosis, necessary to assess prognosis and treatment planning, is not accurate enough with CT and PET/CT imaging.

Keywords

Peritoneal carcinomatosisFDG-PETComputed tomographyColorectal carcinomaPeritoneum CTMucinous

Peritoneal carcinomatosis from gastrointestinal cancer is usually seen at the terminal stage of the disease and is associated with poor overall survival at 3–6 months. Recent aggressive surgical treatment has been described with complete resection of the peritoneal carcinomatosis associated with intraperitoneal chemotherapy and hyperthermy (ICHP) [13]. The intraperitoneal chemotherapy yields a high local drug concentration and hyperthermia enhances the cytotoxicity of anticancer drugs and increases the pharmacokinetic advantage of the intraperitoneal route for chemotherapy. This new treatment yields a highly significant increase of overall survival rate of 53% at 3 years and 48.5% at 5 years [1]. This new aggressive surgical approach requires accurate intraperitoneal assessment of the presence or the absence of peritoneal carcinomatosis but also of the exact extent of carcinomatosis. Indeed, it might be important to detect location, extension, and size of peritoneal implants in view of treatment planning and to estimate whether adequate cytoreductive surgery is likely to be performed. The extent of disease could be reported using the Peritoneal Cancer Index proposed by Sugarbaker. This index has been proved to be of prognostic value for the probability of achieving optimal cytoreduction and for the final outcome [4]. The aim of our study was to assess and compare the performance of CT and 18F-FDG-PET/CT in the evaluation of the diagnosis and the extent of peritoneal carcinomatosis from gastrointestinal cancer with correlation with intraoperative findings.

Material and method

Patients

From September 2004 to September 2005, 30 consecutive patients with peritoneal carcinomatosis from gastrointestinal cancer and scheduled for a surgery were enrolled in this prospective study. The study was approved by our Institutional Review Board. There were 17 males and 13 females with an average age of 49 years, range between 20 and 63 years. All patients have been operated previously for their primary tumor. The location of the primary tumor was the colon in 15 patients (50%), the rectum in 3 patients (10%), the appendix in 10 patients (33%), the stomach in 1 patient (3.5%) and the small bowel in 1 patient (3.5%). All patients underwent a CT and a FDG-PET examination made less than 1 month before the surgery. CT and PET-CT findings were correlated with per-operative surgical findings and histopathological findings in all patients.

Imaging techniques

CT examination

In 24 out of 30 patients, abdominal and pelvic CT examinations were performed in outer centers using different imaging units. All these examinations were performed after oral and intravenous contrast medium administration. In 6 patients the abdominal and pelvic CT were performed in our institution using a 16-detector row CT (GE Lightspeed 16; GE Healthcare Technologies, Waukesha, WI, USA). The imaging protocol included an oral administration of 500 mL of water and an intravenous bolus of 120 mL of Omnipaque 300® at a rate of 3 mL/s administered using a power injector. A portal phase abdominal and pelvic acquisition with a 70 s delay after the initiation of the injection were acquired in all patients with a 1.5 mm collimation, 120 kV, mean tube current-time product 160 mAs, pitch 1.00.

18-FDG PET examination

All imaging and data acquisitions were performed on an integrated PET-CT Biograph LSO system (Siemens Medical Solutions, CTI) using a single table serving for both the attenuation correction CT and PET elements. PET-CT scanning was performed after the intravenous injection of 555 MBq of 18-FDG, followed by a 50–70 min uptake phase. All patients had fasted for 6 h and capillary glycemia was normal in all patients. During the image acquisition, patients maintained their arms above their head and no specific breathing instructions were given. The PET elements of the system are based on a full ring tomograph (ECAT ACCEL, CTI). 3D mode was used for PET image acquisition. PET data were reconstructed on a 128 × 128 matrix, using an iterative algorithm (FORE and AWOSEM) with two iterations, eight subsets and a 5 mm FWHM gaussian postfilter. Reconstruction data were acquired with a single slice spiral CT (Somatom Emotion, Siemens Medical Solutions). CT parameters were set to 80 mAs and 110 kV, slice thickness of 5 mm and pitch 1.5.

Images analysis

Images of each modality were analyzed by two readers, (nuclear physician and radiologist), blindly and independently. Radiologists and nuclear physicians were unaware of clinical and biological findings or of any imaging studies concerning the patients. Peritoneal tumor was suspected on PET scans if they demonstrated an intense peritoneal focal uptake, an irregularly distributed spotty uptake in the abdomen, nodular or curvilinear uptake along the liver or nodular uptake in the left subphrenic space. Peritoneal spread was suspected on CT scans if they demonstrated a parietal peritoneal thickening or enhancement either smooth or nodular, a small bowel involvement with wall thickening and bowel distortion, an involvement of the omentum such as soft tissue permeation of fat, enhancing nodules or omental cake.

Surgeons and reviewers precisely assessed the extent of PC using the peritoneal cancer index. The presence of tumor deposit was noted in 11 abdomino-pelvic areas defined as follow: 1: right upper, 2: epigastrium, 3: left upper, 4: right flank, 5: central, 6: left flank, 7: right lower, 8: pelvis, 9: left lower, 10: upper small bowel, 11: lower small bowel (Fig. 1). The size of the largest tumor deposit was also scored on CT examination and during surgery. Zero indicates the absence of cancer within a particular abdomino-pelvic region. Lesion score of 1 indicates tumor nodules less than 5 mm in diameter, lesion score of 2 indicates tumor 0.5–5 cm in diameter, lesion score of 3 indicates tumor nodules greater than 5 cm in diameter. If an organ is coated by a mat of tumor (confluent disease), this region or site is scored as lesion size 3. These scores were summated for all abdomino-pelvic areas ranging from 0 to 33. The presence of ascitis was also recorded. Intraoperative findings were regarded as the gold standard. Comparison of CT, PET-CT and surgical findings were made using an inter-class correlation coefficient (Figs. 2, 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs00261-007-9211-7/MediaObjects/261_2007_9211_Fig1_HTML.jpg
Fig. 1.

Abdominopelvic regions 1–11. 1: right upper, 2: epigastrium, 3: left upper, 4: right flank, 5: central, 6: left flank, 7: right lower, 8: pelvis, 9: left lower, 10: upper small bowel (arrow), 11: lower small bowel (arrow).

https://static-content.springer.com/image/art%3A10.1007%2Fs00261-007-9211-7/MediaObjects/261_2007_9211_Fig2_HTML.jpg
Fig. 2.

A 52-year-old man with a peritoneal carcinomatosis from colorectal cancer. A Transverse enhanced CT scan reveals a mass in the right lower and in the pelvis (arrows) corresponding to peritoneal carcinomatosis. B 18F-FDG-PET scan, coronal section, demonstrates focal uptake in the right lower and pelvis (arrows) associated with uptake scattered through the colon (arrow heads) corresponding to an extensive peritoneal carcinomatosis.

https://static-content.springer.com/image/art%3A10.1007%2Fs00261-007-9211-7/MediaObjects/261_2007_9211_Fig3_HTML.jpg
Fig. 3.

A 47-year-old women with peritoneal carcinomatosis from colorectal cancer at surgery. A Transverse enhanced CT scan reveals a ceacal mass corresponding to the primary tumor (arrow). B Transverse enhanced CT scan reveals a small bowel wall thickening (arrowheads) and tumoral involvement of the omentum (arrows) corresponding to peritoneal carcinomatosis. C 18F-FDG-PET scan, coronal section, demonstrates an intense tracer uptake in the right lower corresponding to the ceacal primary tumor but without significant peritoneal uptake.

Results

Intraoperatively, 2 of the 30 patients had no peritoneal carcinomatosis. The sensitivity for the diagnosis of peritoneal carcinomatosis was 82% for CT and 57% for PET-CT examination. The numbers of abdomino-pelvic areas involved by peritoneal carcinomatosis detected by surgery, CT and PET-CT examinations are summarized in Table 1. The mean index of peritoneal carcinomatosis score was 10.2 for surgery and 5.4 for CT-scan. This score was not available for PET-scan because no size lesion could be assessed using PET-scan. Peritoneal implants with adhesion with the small bowel were detected in 25 patients (83%) during surgery, 8 patients (26%) using CT-scan and 0 patient using PET-CT examination. The extent of the peritoneal dissemination was overestimated in 1 patient using CT and 2 patients using PET-CT-scan. The extent of the peritoneal dissemination was underestimated in 21 patients (70%) using CT-scan and 24 patients (80%) using PET-CT. The mean differences of the number of involved areas between CT-scan and surgery and PET-scan and surgery are summarized in Table 2.
Table 1.

Number of abdomino-pelvic areas involved by peritoneal carcinomatosis detected during surgery, on CT and PET examinations

No. of involved areas

Surgery no. of patient (%)

CT no. of patient (%)

PET no. of patient (%)

0

2 (6.67)

6 (20)

12 (40)

1

4 (13.33)

7 (23.33)

6 (20)

2

1 (3.33)

5 (16.67)

4 (13.33)

3

1 (3.33)

4 (13.33)

3 (10)

4

6 (20)

1 (3.33)

1 (3.33)

5

1 (3.33)

0

0

6

3 (10)

2 (6.67)

1 (3.33)

7

3 (10)

1 (3.33)

0

8

1 (3.33)

0

1 (3.33)

9

8 (26.67)

4 (13.33)

2 (6.67)

On histological analysis of surgical specimen, the tumor was classified non-mucinous adenocarcinoma in 21 patients and mucinous adenocarcinoma in 9 patients. The mean index of peritoneal carcinomatosis score for surgery and CT-scan were 6.75 and 2.95, respectively, for non mucinous tumors and 17.88 and 12.57, respectively, for mucinous tumors (Table 2).
Table 2.

Mean differences of the number of involved area between CT and surgery and between PET/CT and surgery

Mean difference

CT vs. surgery (%)

PET vs. surgery (%)

−2

0

7.1

−1

3.3

0

0

26.6

14.3

1

16.6

14.3

2

6.6

7.1

3

13.3

10.7

4

16.6

17.8

5

10

7.1

6

0

3.5

7

6.6

7.1

8

0

3.5

9

0

7.1

The interclass correlation was 0.53 (moderate) between CT and surgery and 0.12 (low) between PET/CT and surgery. The interclass correlation was higher for mucinous tumor (0.63) than for non-mucinous (0.16) on CT imaging whereas no difference was found in PET/CT (0.11 vs. 0.14) (Table 3).
Table 3.

Interclass correlation coefficient between TDM and PET examinations and surgery for all tumors, non-mucinous tumors and mucinous tumors

 

All tumors

Non-mucinous tumors

Mucinous tumors

TDM vs. surgery

0.53

0.157

0.632

PET vs. surgery

0.123

0.11

0.14

Discussion

The combination of a maximal cytoreductive surgery with intraperitoneal chemohyperthermia results in cure in a significant number of patients. Complete resection of peritoneal carcinomatosis is the determining factor for the efficacy of this treatment [5]. Other clinical features showing prognostic significance are appendiceal vs. colorectal cancer, grade 1 vs. grade 2 and 3 histopathology, lymph node-negative vs. lymph node-positive primary tumors and volume of peritoneal carcinomatosis present preoperatively for colon cancer [5]. Thus in patients with colorectal cancer it is important to diagnose carcinomatosis as well as to detect location and size of peritoneal tumor dissemination in view of treatment planning. Imaging plays an important role in the assessment of the volume of peritoneal carcinomatosis. The peritoneal cancer index, used by surgeons to evaluate peritoneal carcinomatosis, is also helpful to evaluate carcinomatosis using imaging. The peritoneal cancer index (PCI) combines the distribution assessment and the lesion size assessment to yield a composite score. Patients who have a PCI determination greater than 12 are not recommended for elective cytoreductive surgery with perioperative intraperitoneal chemotherapy [6]. Other important predictive factors of incomplete resectability of colo-rectal cancer described in previous studies are the persistence of ascites and any progression of the disease under chemotherapy [7].

Our study showed that CT detection of peritoneal carcinomatosis for colorectal cancer is good but detection of individual peritoneal tumor deposits is poor. Our results are in accordance with previous studies showing an overall sensitivity of CT in the detection of PC ranging from 60% to 79% [8, 9]. This sensitivity is depending on radiologists, tumor size and the location of tumor deposits. De Bree et al. [8] decribed a poor detection of individual peritoneal implants varying from 9.1% to 24.3% for tumor size < 1 cm and 59.3%–66.7% for tumor size > 5 cm and an overall sensitivity for tumor involvement per area of 24.5%. Similarly, Jacquet et al. [9] found a sensitivity of only 28% when tumors nodules were less than 0.5 cm in thickness and sensitivity lower in the pelvis (60%) and greater in the right and left flanks and right and left lower abdominal regions (80%). The good overall sensitivity of CT scan for the detection of peritoneal carcinomatosis, observed in this study could be explained by the fact that all CT images were analyzed prospectively by an experienced gastrointestinal radiologist that was specially focused on PC. Indeed, the main weakness of CT-imaging in the diagnosis of the PC is its poor interobserver agreement. Warde et al. [10] investigated the assessment of the presence of PC by 3 radiologists in 50 patients. The sensitivities varied from 0.30 to 0.65 among the radiologists and positive predictive value from 0.50 to 0.73. Coakley et al. [11] assessed the inter-reader agreement evaluated with the kappa statistic of three independent readers in the depiction of peritoneal carcinomatosis. The k varied from 0.35 to 0.50 among readers for the assessment of peritoneal thickening and from 0.12 to 0.25 for the assessment of a small bowel involvement (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs00261-007-9211-7/MediaObjects/261_2007_9211_Fig4_HTML.jpg
Fig. 4.

A 52-year-old man with a recurrent rectal carcinoma. A, B 18F-FDG-PET scans, axial fusion sections, demonstrate a focal uptake forward the sacrum and in the pelvis (arrows) corresponding to peritoneal carcinomatosis. C, D Tranverse enhanced CT scan retrospectively depict corresponding lesions (arrows) that were initially missed.

Our study shows a low sensitivity of 18F-FDG in the detection of peritoneal carcinomatosis and an important underestimation of the extent of the lesion (Fig. 5). Most previously published studies have evaluated the value of FDG-PET in the diagnosis of PC without consideration of the extent of the peritoneal involvement [1214]. At present, the major weakness of PET imaging in the detection of PC is its low spatial resolution because most of peritoneal tumor implants are small nodules or microscopic infiltration. Moreover, some abnormalities suggestive of peritoneal tumor on PET-scan imaging, such as diffuse low-grade glucose hypermetabolism spreading uniformly throughout the abdomen and pelvis obscuring outlines, are faint and associated with a low confidence of presence [12]. Other limitations inducing false negative results are the lack of substantial uptake of FDG by a malignant lesion, the underestimation of uptake due to physiologic movements, mislocalization of foci and recent completion of chemotherapy [15]. There are also some false-positive results, due to the variable physiologic tracer uptake of FDG. The normal stomach, colon, especially in the rectosigmoïd region, and the cecum, or small intestine, can show FDG uptake that is sometimes difficult to distinguish from that of peritoneal carcinomatosis even with anatomic correlation [16].
https://static-content.springer.com/image/art%3A10.1007%2Fs00261-007-9211-7/MediaObjects/261_2007_9211_Fig5_HTML.jpg
Fig. 5.

A 65-year-old man with a peritoneal carcinomatosis from colorectal cancer. A, B Transverse enhanced CT scans reveal soft tissue permeation of the omentum fat (arrows) corresponding to peritoneal carcinomatosis. C 18F-FDG-PET scan, coronal section, does not show the corresponding tracer uptake.

Unexpectedly, we found no significant difference in the value of PET-scan for mucinous carcinoma compared to non-mucinous carcinoma. Indeed previous studies have reported a limited sensitivity of FDG-PET in the detection of mucinous neoplasms [17, 18]. It was felt that the lack of detection by FDG-PET was attributable to the metabolic inactivity of the mucin and the relative hypocellularity of these tumors [17]. At the opposite we found a greater sensitivity of CT imaging for mucinous than for non-mucinous tumor. Indeed peritoneal mucinous carcinomatosis is most often associated with large aggregates of extracellular mucin well depicted on CT imaging.

In a previous study comparing pre-operative CT parameters between patients with complete and incomplete surgical resection, CT findings significantly different were the tumor volume in small bowel mesentery, in proximal jejunum, in distal jejunum, in proximal ileum, mesentery configuration and obstruction of bowel segments by tumor [19]. Although the extent of the tumor on the small bowel implant is considered as a major parameter to predict the probability of a complete resection, our study showed that neither CT nor PET is a reliable diagnostic test to assess the involvement of the small bowel.

In conclusion, accurate detection of peritoneal carcinomatosis from colorectal carcinoma remains a diagnostic challenge. Although the sensitivity of CT scan is good, superior to that of PET scan in the detection of peritoneal carcinomatosis, neither CT nor PET examination is reliable imaging method in the preoperative assessment of the extent of peritoneal involvement, in particular to predict small bowel involvement.

Copyright information

© Springer Science+Business Media, LLC 2007