Original Research
Gastrointestinal Imaging
February 2010

PET/CT Pattern Analysis for Surgical Staple Line Recurrence in Patients With Colorectal Cancer

Abstract

OBJECTIVE. The purpose of our study was to determine whether 18F-FDG PET/CT interpretation with metabolic–anatomic pattern analysis can be used to accurately assess for surgical staple line recurrence after colorectal cancer resection.
MATERIALS AND METHODS. Seventy-nine consecutive patients with previous surgical resection of colorectal cancer were studied retrospectively. The surgical anastomotic or Hartmann's pouch staple lines were evaluated for presence or absence of tumor recurrence with FDG PET/CT metabolic–anatomic pattern analysis. Focal, eccentric, or perianastomotic CT masses with any associated PET pattern were regarded as positive for staple line recurrence. If the perianastomotic CT abnormality was presacral in location, then FDG uptake at least as intense as normal liver was required for positive interpretation. Eccentric or perianastomotic PET patterns matched with normal or diffuse thickening CT patterns were regarded as indeterminate. Presence or absence of recurrent tumor was confirmed by pathology, surgery, colonoscopy, imaging follow-up of at least 3 months, or clinical follow-up of at least 1 year.
RESULTS. Nine patients (11.4%) had staple line recurrence and 70 (88.6%) did not. FDG PET/CT interpretation yielded sensitivity, specificity, positive predictive value, negative predictive value, and accuracy results of 100% (9/9), 97.1% (68/70), 81.8% (9/11), 100% (68/68), and 97.5% (77/79), respectively. All nine patients with staple line recurrence showed perianastomotic or eccentric masses on CT, eight with matching perianastomotic or eccentric FDG uptake patterns. Background, diffuse, curvilinear, or focal FDG uptake patterns, regardless of FDG uptake intensity, paired with normal findings or diffuse mural thickening on CT were seen only in patients without staple line recurrence.
CONCLUSION. FDG PET/CT pattern analysis enables accurate assessment for staple line recurrence in patients with previous resection of colorectal cancer. The most reliable PET/CT pattern predicting staple line recurrence is an eccentric or perianastomotic mass on CT with corresponding eccentric or perianastomotic FDG uptake on PET. Background, diffuse (on one or both sides of the staple line), curvilinear, and focal patterns of FDG uptake do not correlate with recurrence in the absence of a mass on CT.

Introduction

Anastomotic recurrences after colon cancer resection, excluding rectal cancer, occur in 2–4% of patients [1]. Anastomotic recurrences after resection of rectal cancer are historically up to 10 times more common than after colon cancer resection [1]. Improvements in surgical technique and the trend toward total mesorectal excision now result in anastomotic recurrence rates of less than 10% for rectal cancer [1]. Local recurrences, including those at anastomotic and Hartmann's pouch staple lines, have been associated with 5-year mortality rates of 80–90% and are more common in patients with poorly differentiated or advanced-stage tumors [2, 3]. Studies indicate improved survival rates attributable to improvements in surgical management, chemotherapy, and radiation therapy [1, 3, 4].
Imaging surveillance after surgery may afford earlier detection of staple line recurrence than serum tumor markers or colonoscopy, potentially facilitating salvage surgery or other local therapies [58]. Early detection and successful resection of staple line recurrence have the potential to improve patient survival.
PET and PET/CT with 18F-FDG are now used to restage colorectal cancer after surgical resection [9]. A meta-analysis of 32 studies reported lesion-based sensitivity and specificity for extrahepatic recurrences of 92% and 95%, respectively, for FDG PET compared with 61% and 91% for CT [10]. PET led to changes in clinical management in 32% of patients. The added value of PET/CT relative to PET alone was addressed in another study that documented a 55% decrease in uncertain lesion localizations for PET/CT compared with PET alone. This improved localization of FDG uptake led to a 50% decrease in equivocal or probable diagnoses [11].
Studies of FDG PET and PET/CT in the assessment of local recurrence of colorectal cancer have not focused on anastomotic or Hartmann's pouch staple line involvement, even though the altered physiology and anatomy of surgical staple lines have the potential to complicate PET/CT interpretation in these areas. PET/CT was used successfully in one study to detect 24 pelvic recurrences, including six anastomotic recurrences in 62 patients with previous rectal cancer resection [12]. Specific PET and CT findings, however, were not addressed with respect to the staple line regions. The purpose of this study was to determine if FDG PET/CT interpretation with metabolic–anatomic pattern analysis can be used to accurately assess for staple line recurrence after colorectal cancer resection.

Materials and Methods

Study Population

This retrospective study was performed with institutional review board approval and was in compliance with HIPAA standards. Informed consent was waived. All patients with colorectal cancer undergoing restaging PET/CT between September 2003 and November 2007 were identified. Exclusion criteria included unknown date of surgical resection, inability to visualize the tumor resection staple line on CT, or lack of a reference standard as described later. Seventy-nine patients (50 men, 29 women; age range, 33–87 years; mean, 59 years) were enrolled. The study group was not part of a routine surveillance program for early detection of anastomotic recurrences. For each patient, the initial PET/CT performed at our institution after surgical resection was reviewed.
Fig. 1A Spectrum of PET/CT patterns observed at staple line region: gray lines indicate CT anatomy and red areas indicate increased 18F-FDG uptake on PET. Normal CT pattern bisected by staple line and with background FDG uptake.
Fig. 1B Spectrum of PET/CT patterns observed at staple line region: gray lines indicate CT anatomy and red areas indicate increased 18F-FDG uptake on PET. Normal CT pattern with diffuse PET pattern of FDG uptake extending above and below staple line.
Fig. 1C Spectrum of PET/CT patterns observed at staple line region: gray lines indicate CT anatomy and red areas indicate increased 18F-FDG uptake on PET. Normal CT pattern with diffuse PET pattern showing abrupt transition in intensity at staple line.
Fig. 1D Spectrum of PET/CT patterns observed at staple line region: gray lines indicate CT anatomy and red areas indicate increased 18F-FDG uptake on PET. Normal CT pattern with curvilinear PET pattern closely following staple line.
Fig. 1E Spectrum of PET/CT patterns observed at staple line region: gray lines indicate CT anatomy and red areas indicate increased 18F-FDG uptake on PET. Normal CT pattern with focal PET pattern at staple line.
Fig. 1F Spectrum of PET/CT patterns observed at staple line region: gray lines indicate CT anatomy and red areas indicate increased 18F-FDG uptake on PET. Eccentric thickening or mass on CT with corresponding eccentric PET pattern of FDG uptake.
Fig. 1G Spectrum of PET/CT patterns observed at staple line region: gray lines indicate CT anatomy and red areas indicate increased 18F-FDG uptake on PET. Perianastomotic mass on CT with corresponding perianastomotic PET pattern.

Reference Standards

The PET/CT interpretation reference standards for the nine patients with staple line recurrence included pathology in seven (CT-guided percutaneous biopsy in two, colonoscopic biopsy in two, and surgical pathology in three). In the remaining two patients, confirmation of recurrence was by imaging; one showed tumor progression on PET/CT performed 11 months after the initial positive PET/CT, and the other showed tumor regression after chemotherapy on PET/CT performed 5 months after the initial positive PET/CT along with a drop in serum carcinoembryonic antigen from 6.6 to 1.5 ng/mL (normal = 0–2.5 ng/mL). The reference standards for the 70 patients who did not develop staple line recurrence included pathology in nine (surgical pathology in one patient, colonoscopy with biopsy in seven, and CT-guided percutaneous needle biopsy in one). Surgery without pathology provided confirmation in three patients, colonoscopy without biopsy in five, and imaging-only follow-up in 51 (35 PET/CT, 15 CT, one MRI; mean follow-up interval, 19.4 months; range, 3–49 months; 46 with at least 6 months of imaging follow-up). Clinical-only follow-up was available in two patients (12 months and 36 months, respectively, without symptoms referable to abdominal recurrence). Reference standards for all other local and distant recurrences not involving the staple line included positive imaging findings, colonoscopy, surgery, or pathology.

PET/CT Acquisition

PET/CT scans were performed on a Discovery ST or Discovery VCT (GE Healthcare). Patient preparation included no oral intake other than water for 4–6 hours. FDG (555–1,036 MBq [mean, 755 MBq, 20.4 mCi]), was administered IV 60 minutes before PET/CT. Blood glucose levels were less than 150 mg/dL for all patients. CT and PET acquisitions were obtained during quiet breathing from the skull base to the proximal thighs. Unenhanced CT (no oral or IV contrast material) was performed with 140 kVp and 40–120 mAs; the tube current was based on weight. PET was performed in the 2D mode with 4-minute acquisitions per bed position. An ordered subset expectation maximization iterative reconstruction algorithm was used for CT attenuation correction of the PET images. PET/CT images were reconstructed with a 3.75-mm slice thickness and 3.27-mm interval.

PET/CT Interpretation

PET/CT scans were reinterpreted for this study at an Xeleris workstation (GE Healthcare) by one radiology fellow and one attending radiologist reading jointly. The radiology fellow had previously completed a 3-month dedicated PET/CT rotation during the fellowship. The attending radiologist had 7 years of experience independently interpreting PET or PET/CT. The scans were read by consensus with both readers blinded to clinical, imaging, and pathology data except for the history of previous colorectal cancer resection. Disagreements between readers were uncommon and were resolved by further review, discussion, and mutual agreement. Multiplanar CT, attenuation-corrected PET, and coregistered PET/CT displays were reviewed at the workstation, along with non-attenuation-corrected PET and rotating maximum-intensity-projection images. CT images from the PET/CT scans were also reviewed on a Centricity PACS workstation (GE Healthcare). PET and CT data sets were not reviewed independently but rather were read simultaneously to assess a combined PET/CT pattern-interpretation approach.
FDG uptake patterns at the staple line region on PET were classified as follows: background, uptake similar to or indistinguishable from adjacent normal structures; diffuse, linear uptake greater than background activity along the bowel segments on one or both sides of the staple line; curvilinear, uptake greater than background in a curvilinear pattern closely following the staple line; focal, uptake greater than background without curvilinear pattern localized to the staple line; eccentric, uptake greater than background centered in a portion of the staple line wall; and perianastomotic, uptake greater than background centered outside the bowel wall but contiguous with the staple line (Fig. 1A, 1B, 1C, 1D, 1E, 1F, 1G). The distinction between curvilinear and focal uptake was related to the size or length of the staple line; larger diameter or longer staple lines allowed delineation of a curvilinear pattern, whereas smaller or shorter staple lines appeared focal without curvilinear definition. PET pattern assignments were dependent on anatomic localization using the coregistered CT images.
CT patterns at the staple line region were classified as follows: normal, no abnormality of the bowel segment at the staple line and no contiguous abnormality other than soft-tissue stranding without nodule or mass; diffuse thickening, uniform circumferential thickening along the bowel segments on one or both sides of the staple line; focal, nodule or mass centered within the bowel lumen at the staple line; eccentric, thickening or mass centered in a portion of the staple line wall; and perianastomotic, nodule or mass centered outside the bowel wall and contiguous with the staple line (Fig. 1A, 1B, 1C, 1D, 1E, 1F, 1G). Presacral masses were not considered staple line abnormalities unless contiguous with the staple line; local recurrences not involving the surgical anastomosis or Hartmann's pouch staple line were not included in this study. The PET and CT pattern definitions outlined were based on prior reader experience.
Interpretation of PET/CT findings at the staple line region was based on metabolic–anatomic pattern assignments as follows: background, diffuse, curvilinear, or focal PET patterns matched with normal or diffuse thickening CT patterns were considered negative for recurrence (Table 1). Focal, eccentric, or perianastomotic CT patterns matched with any of the PET patterns were considered positive for recurrence. Eccentric or perianastomotic PET patterns matched with normal or diffuse thickening CT patterns were regarded as indeterminate. A perianastomotic mass in a presacral location was a special circumstance and was considered positive for recurrence when associated with FDG uptake greater than or equal to normal liver and negative for recurrence with FDG uptake less than normal liver. Subjective assessment of FDG uptake intensity was used for PET/CT interpretation; however, semiquantitative measurements of FDG uptake were recorded for separate analysis using standardized uptake value (SUV) measurements normalized to body weight [13]. The maximum SUV within a region of interest at the staple line including any contiguous perianastomotic abnormality (SUVmax), the average SUV within a region of interest over normal liver (SUVavg), and the ratio of anastomotic SUVmax to normal liver SUVavg (SUVratio) were recorded for each PET/CT scan.
TABLE 1: Potential Pattern Combinations of PET and CT Findings With Associated Interpretation
PET PatternCT PatternInterpretation
AnyFocal, eccentric, or perianastomoticPositive
Background, diffuse, curvilinear, or focalNormal or diffuse thickeningNegative
Eccentric or perianastomoticNormal or diffuse thickeningIndeterminate
Presacral perianastomotic (SUV ≥ normal liver)
Presacral perianastomotic
Positive
Note—PET/CT patterns are defined in the text. No focal CT pattern and no indeterminate pattern combinations were encountered in this study. SUV = standardized uptake value.

Statistical Methods

The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of FDG PET/CT pattern analysis in detecting staple line recurrence were calculated with 95% CIs. The same parameters were then calculated for SUVmax assessment alone and for SUVratio assessment alone. The McNemar test was used to compare the performance of PET/CT pattern analysis, SUVmax alone, and SUVratio alone.

Results

Nine patients (11.4%) had staple line recurrence and 70 (88.6%) did not. The mean time interval from surgical resection of the primary tumor to initial PET/CT in patients with recurrence was 40 months (range 5 weeks–139 months). If the patient with a 139-month interval is excluded, the mean interval was 28 months. The patient with a 5-week interval from right hemicolectomy to positive PET/CT was thought to be free of all visible tumor at surgical resection; however, the surgical pathology report indicated extensively invasive adenocarcinoma including lymphatic invasion, venous invasion, and eight of 13 positive lymph nodes. Surgical margins were reported to be histologically negative. The rapid development of an anastomotic recurrence in this patient is most likely attributable to a substantial microscopic residual tumor burden adjacent to the staple line. The next shortest time interval to staple line recurrence on PET/CT was 10 months. Two staple line recurrences occurred in women, seven in men. Eight had anastomotic recurrences (two ileocolic, two left colocolonic, three colorectal, and one coloanal anastomoses) and one had a Hartmann's pouch staple line recurrence. Staple line recurrences ranged in size from 1.2 to 4.6 cm in diameter (mean, 2.9 cm).
For the 70 patients without staple line recurrence, the mean time interval from surgical resection of the primary tumor to initial PET/CT was 21 months (range, 1–156 months). Forty-three of these patients were men and 27 were women. Sixty-three had bowel anastomoses (26 ileocolic, nine colocolonic, 22 colorectal, and six coloanal anastomoses), and seven had Hartmann's pouch staple lines. Six (66.7%) of nine patients with staple line recurrence and 51 (72.9%) of 70 patients without staple line recurrence had other local or distant sites of recurrence at some point during their postoperative course.
PET/CT pattern analysis was true-positive in all patients with staple line recurrence (Table 2). There were no false-negative findings. All nine true-positive interpretations showed a perianastomotic or eccentric mass or nodule on CT, eight with a concordant perianastomotic or eccentric PET uptake pattern (Fig. 2A, 2B, 2C). One of these recurrences included both a perianastomotic mass and eccentric thickening at the anastomosis on CT with associated perianastomotic FDG uptake on PET. One recurrence was characterized by an eccentric mass on CT with a corresponding eccentric PET pattern of uptake (Fig. 3A, 3B, 3C). Another patient had a background PET pattern (perianastomotic FDG uptake indistinguishable from adjacent normal bladder activity) and a CT pattern of perianastomotic tumor growing into the left ureterovesical junction (Fig. 4A, 4B, 4C).
TABLE 2: PET/CT Patterns Observed in Patients With and Without Staple Line Recurrence After Resection of Colorectal Cancer
PET PatternCT PatternPatients With Staple Line Recurrence (n = 9)Patients Without Staple Line Recurrence (n = 70)
BackgroundNormal 18 True-negative
DiffuseNormal 38 True-negative
DiffuseDiffuse thickening 5 True-negative
CurvilinearNormal 5 True-negative
FocalNormal 2 True-negative
BackgroundPerianastomotic1 True-positive 
EccentricEccentric1 True-positive 
PerianastomoticPerianastomotic and eccentric1 True-positive1 False-positive
Perianastomotic
Perianastomotic
6 True-positive
1 False-positive
Note—PET/CT patterns are defined in the text. The number of patients with true-positive, true-negative, and false-positive PET/CT interpretations are listed for each metabolic—anatomic pattern combination observed in this study. There were no false-negative interpretations.
Among 70 patients without staple line recurrence, PET/CT interpretation was true-negative in 68 patients and false-positive in two (Table 2). Background, diffuse, curvilinear, or focal FDG uptake patterns matched with either normal CT or diffuse thickening of the bowel segment on CT were seen in all true-negative interpretations. One of the false-positive interpretations combined perianastomotic FDG uptake with a perianastomotic and eccentric CT mass. The CT abnormality consisted of a presacral mass, contiguous with a Hartmann's pouch staple line and containing a cystic focus; however, percutaneous needle biopsy 8 days later revealed only inflammation and scar tissue. PET/CT follow-up at 15 months revealed no change. The other false-positive interpretation showed apparent perianastomotic FDG uptake within a perianastomotic presacral scar seen on CT (Fig. 5A, 5B, 5C). Absence of recurrence in this case was confirmed by colonoscopy with biopsy and PET/CT follow-up at 7 months without interval change. One of the true-negative patients paired an ill-defined presacral soft-tissue opacity, contiguous with the staple line but without mass or mass effect that was thought to represent scar tissue (normal CT pattern) with a background PET pattern.
Fig. 2A 58-year-old woman with anastomotic recurrence after previous rectal cancer resection. Axial CT image shows perianastomotic mass (arrow) anterior to anastomosis and posterior to uterus.
Fig. 2B 58-year-old woman with anastomotic recurrence after previous rectal cancer resection. Fused PET/CT image confirms 18F-FDG uptake localized to perianastomotic mass.
Fig. 2C 58-year-old woman with anastomotic recurrence after previous rectal cancer resection. Intensity of tumor FDG uptake (arrow) relative to surrounding normal structures is better appreciated on unfused PET image.
Of the 43 patients with a diffuse pattern of FDG uptake and without staple line recurrence, seven had diffuse uptake characterized by an abrupt transition in intensity at the anastomotic staple line. FDG uptake was diffusely more intense in the segment either above or below the staple line (Fig. 5A, 5B, 5C). The SUVmax on the more intense side of the anastomosis ranged from 1.8 to 10.3. Three of these patients had ileocolic anastomoses, three had colorectal anastomoses, and one had a coloanal anastomosis. Reversal of a diffuse transition pattern, such that greater uptake was seen in the bowel segment on the side of the anastomosis that previously showed lesser uptake, was observed in one patient with ileocolic anastomosis and one patient with coloanal anastomosis on follow-up PET/CT. Three patients with a diffuse uptake pattern showed intraluminal FDG accumulation confirmed by air–fluid levels on CT with matching air–FDG levels on PET.
Semiquantitative measurements of FDG uptake at the staple line region, including SUVmax and SUVratio, tended to be higher in patients with staple line recurrence than in those without recurrence; however, there was considerable overlap in the ranges of these parameters in the two populations (Fig. 6). The mean SUVmax in patients with recurrence was 6.98 (range, 2.3–10.1) and in those without recurrence 2.98 (0.9–10.4). The mean SUVratio in patients with recurrence was 3.0 (0.96–5.6) and in those without recurrence 1.18 (0.41–5.15).
Fig. 3A 33-year-old woman with recurrence of colorectal cancer at descending colonic anastomosis. Axial CT image reveals eccentric mass at anastomosis (arrow).
Fig. 3B 33-year-old woman with recurrence of colorectal cancer at descending colonic anastomosis. PET/CT image localizes eccentric 18F-FDG uptake to portion of staple line circumference.
Fig. 3C 33-year-old woman with recurrence of colorectal cancer at descending colonic anastomosis. PET image shows intense FDG uptake in mass (arrow).
Fig. 4A 56-year-old man with anastomotic recurrence of rectal cancer involving left ureterovesical junction region and detectable only with CT. Axial CT image shows perianastomotic tumor (arrow) invading left ureterovesical junction.
Fig. 4B 56-year-old man with anastomotic recurrence of rectal cancer involving left ureterovesical junction region and detectable only with CT. On PET/CT image, tumor that is visible on CT alone is obscured by PET overlay.
Fig. 4C 56-year-old man with anastomotic recurrence of rectal cancer involving left ureterovesical junction region and detectable only with CT. On PET image in perianastomotic focus (arrow), 18F-FDG uptake is indistinguishable from adjacent physiologic bladder activity, background pattern.
The accuracy of PET/CT pattern analysis in assessing for staple line recurrence was significantly greater than either SUVmax alone (p < 0.01) or SUVratio alone (p < 0.01) (Table 3). The specificity of PET/CT pattern analysis was also significantly greater than either SUVmax (p < 0.05) or SUVratio (p < 0.05). No statistically significant differences were found in sensitivity of PET/CT pattern analysis versus SUVmax (p = 1) or SUVratio (p = 0.5). No statistically significant difference was found in the sensitivity, specificity, or accuracy of SUVmax versus SUVratio (p = 1) as independent variables in the assessment of staple line recurrence.
TABLE 3: Performance of PET/CT Pattern Analysis Compared With Semiquantitative Standardized Uptake Value (SUV) Measurements in Assessing for Staple Line Recurrence
Method of Assessing Staple LineSensitivity (%)Specificity (%)Positive Predictive Value (%)Negative Predictive Value (%)Accuracy (%)
PET/CT pattern analysis100 (9/9) [70.0-100]97.1 (68/70) [90.2-99.2]81.8 (9/11) [52.3-94.7]100 (68/68) [94.7-100]97.5 (77/79) [91.2-99.3]
Staple line (SUVmax)88.9 (8/9) [56.5-98.0]87.1 (61/70) [77.3-93.1]47.1 (8/17) [26.2-69.0]98.4 (61/62) [91.4-99.7]87.3 (69/79) [78.2-92.9]
Staple line to normal liver (SUVratio)
77.8 (7/9) [45.3-93.7]
88.6 (62/70) [79.0-94.1]
46.7 (7/15) [24.8-69.9]
96.9 (62/64) [89.3-99.1]
87.3 (69/79) [78.2-92.9]
Note—Values are expressed as percentages with fractions in parentheses and 95% CIs in brackets. SUVmax = maximum standardized uptake value at the staple line region. SUVratio = ratio of staple line SUVmax to normal liver average SUV. The cutoff value for SUVmax was 4.45 and for SUVratio was 1.96. These values were selected to maximize area under the receiver operating characteristic curve. Lower cutoff values would have improved sensitivity at the expense of specificity.

Discussion

Early detection of local colorectal cancer recurrence, including staple line recurrence, has the potential to increase the rate of successful salvage surgery and improve the efficacy of other local therapies that may improve patient survival. The increasing use of FDG PET/CT in the restaging of colorectal cancer raises the question of how accurately anastomotic or Hartmann's pouch staple line recurrences can be diagnosed with PET/CT. The study design incorporated a correlative, metabolic–anatomic pattern approach to the assessment for recurrences superimposed on the altered physiology and anatomy of surgical staple lines.
PET/CT interpretation with pattern analysis was 97.5% accurate in assessing for staple line recurrence in this study. PET/CT pattern analysis was true-positive in all nine recurrences and false-positive in two. CT was abnormal in all nine patients with recurrence. PET alone would have missed one of the recurrences due to inability to distinguish tumor FDG uptake from adjacent urinary bladder activity. Absence of a mass on the CT portion of the scan was very helpful in excluding staple line recurrences, given the broad overlap of SUVmax and SUVratio measurements at the staple line regions in patients with and without recurrence.
Anastomotic recurrences after surgery for colorectal cancer may potentially develop along the mucosal margin of the staple line, in the intramural portion of the staple line, or in the perianastomotic tissues outside the bowel wall. However, the preponderance of anastomotic recurrences are reported in the literature to originate in the perianastomotic soft tissues with subsequent ingrowth into the staple line [3, 14]. In eight of the nine recurrences in our study, the tumor was centered in the perianastomotic tissues, confirming the tendency for anastomotic recurrences to originate outside the bowel wall. Anastomotic recurrence manifesting as an endoluminal mass on CT (focal pattern) was not seen in this study but has been described, suggesting that this may represent an uncommon pattern of staple line recurrence [15]. The staple line recurrences observed in our patients were relatively large, with only three measuring less than 2.0 cm in diameter and six measuring up to 4.6 cm in diameter. The size of staple line recurrences in our population along with their frequent perianastomotic location may explain the excellent sensitivity of the nonoptimized CT technique used in our PET/CT protocol.
Fig. 5A 52-year-old man with previous rectal cancer resection and false-positive PET/CT. Sagittal CT image shows diffuse thickening of rectum, including anastomosis (arrow), as well as contiguous presacral scar. Patient had previous radiation therapy.
Fig. 5B 52-year-old man with previous rectal cancer resection and false-positive PET/CT. Diffuse intense 18F-FDG uptake on PET/CT image with transition at staple line was initially thought to have perianastomotic uptake component because of lack of defined soft-tissue planes between rectum and presacral scar on CT. This led to false-positive interpretation; however, more careful review confirmed uptake confined to rectal wall.
Fig. 5C 52-year-old man with previous rectal cancer resection and false-positive PET/CT. PET image reveals intense diffuse FDG uptake (arrow) extending up to staple line, then transitioning to mild uptake above staple line. Colonoscopy with needle biopsy revealed only fibrosis. PET/CT follow-up at 7 months was unchanged.
Fig. 6 Box plots of standardized uptake values (SUVs) (maximum [SUVmax] or ratio [SUVratio]) for patients with and without staple line recurrence. Normal liver average SUV (SUVavg) is also provided. Boxes delimit 25th through 75th percentiles, and whiskers indicate range. Horizontal bar is 50th percentile, or median. Mean values are reported in text. Although SUVmax and SUVratio are often higher in patients with recurrence than in those without, broad overlap in SUV range limits specificity.
Several PET patterns were consistently observed in patients without staple line recurrences. Background or diffuse FDG uptake patterns, with or without associated diffuse thickening on CT and regardless of SUV value, were only seen in patients without staple line recurrence. Similarly, a diffuse FDG uptake pattern with transition at the staple line was only seen in patients without staple line recurrence and may reflect physiologic or inflammatory differences in metabolic activity of bowel segments on either side of the anastomosis. Diffuse FDG uptake may, as was seen in three of our patients, reflect intraluminal accumulation of FDG. Intraluminal FDG may result from swallowed salivary activity, mucosal secretion or transport of FDG into the bowel lumen. Curvilinear and focal PET patterns at the staple line were only seen in patients without staple line recurrence in this study and may represent postoperative chronic or granulomatous inflammation. These patients had normal CT findings. Background, diffuse, curvilinear, and focal PET patterns in the absence of a mass on CT, therefore, are not predictive of staple line recurrence.
CT findings in patients without staple line recurrence were almost always normal or diffuse thickening patterns. Diffuse bowel thickening on CT may be caused by ischemic, infectious, or other inflammatory causes including previous radiation therapy [16]. Two patients without staple line recurrence showed a deformity or outpouching at the anastomotic staple line on CT that could mimic a small perianastomotic recurrence. Observing the anastomosis morphology on multiplanar CT reconstructions and noting extension of the staple line into the outpouching allowed distinction from perianastomotic tumor. Perianastomotic, presacral masses resulted in the two false-positive PET/CT scans in our study (Fig. 5A, 5B, 5C). Differentiation of abscess from necrotic tumor remains a problem for PET/CT, particularly in the presacral region because both conditions may produce cystic or cavitary masses on CT with peripheral intense FDG uptake. Presacral scar tissue, contiguous with the staple line and resulting from previous surgery or radiation therapy, can also simulate recurrence on CT. Inflammatory FDG uptake contributes to the difficulty in characterizing these presacral scars on PET/CT but tends to decrease with time. PET/CT performed 6 months or longer after radiation therapy allows FDG uptake to diminish and is more accurate in characterizing presacral scars than PET/CT performed earlier [17].
Focality and intensity of FDG uptake are considered key features in distinguishing physiologic from potentially malignant colonic PET findings; however, only 64–80% of focal FDG-avid colonic findings are reported to be malignant or premalignant [18]. With respect to intensity of FDG uptake, one author reported a mean SUV of 5.5 in pelvic foci of recurrent rectal carcinoma and 4.9 in benign pelvic foci, values that were not statistically significantly different [12]. In our study as well, there was considerable overlap in the range of staple line SUVmax values in patients with and without staple line recurrence, which limited their utility (Fig. 6). Even in patients without recurrence but with focal or curvilinear uptake patterns at the staple line, the SUVmax was variable and occasionally quite high even months to years after surgery (Fig. 7). The absence of a mass on CT was important in avoiding false-positive PET/CT interpretations in these cases. SUVratio did not perform better than SUVmax as an independent predictor of recurrence. In addition to potential limitations of specificity, the sensitivity of PET may be reduced in the settings of predominant mucinous tumor histology, hyperglycemia, or recent chemotherapy [1921]. Because FDG uptake may be low or inseparable from physiologic sites of activity in some staple line recurrences, any suspicious CT finding, even without abnormal FDG uptake, should be investigated further.
Fig. 7 Graph shows results of 15 PET/CT scans on 11 patients without staple line recurrence but with persistent localized 18F-FDG uptake at staple line in curvilinear or focal pattern. Maximum staple line FDG standardized uptake value (SUVmax) is compared with number of years since surgery. In some cases, localized intense FDG uptake may persist at surgical staple line for years in absence of tumor recurrence.
Limitations of this study include pathologic confirmation in only seven of nine patients with staple line recurrence; however, follow-up PET/CT findings were conclusive in the other two cases. Pathology, surgery, or colonoscopy were available to confirm negative results in only 17 patients; in the remaining patients without staple line recurrence, invasive procedures were not clinically justified. Imaging follow-up with a mean interval of 19.4 months provided confirmation in all but two of the remaining patients. A future prospective study would ideally incorporate surveillance colonoscopy or sigmoidoscopy in all patients for more definitive confirmation of absence of staple line recurrence. CT pattern analysis alone appeared to perform as well as combined PET/CT pattern analysis in this study, suggesting no incremental benefit of PET metabolic data in the specific assessment for recurrence at the surgical staple line. It should be emphasized, however, that this study was not designed to compare the performance of PET alone versus CT alone. Furthermore, the pattern approach used in this study could be applied to CT alone but would not be easily applied to PET alone, given the dependence of PET pattern assignments on anatomic correlation with coregistered CT. This study also was not designed to assess the contribution to reader confidence of the individual PET and CT components of the scans. For example, in eight of the nine patients with confirmed staple line recurrence, eccentric or perianastomotic FDG uptake patterns corresponding to eccentric or perianastomotic CT patterns may have increased reader confidence of a positive diagnosis. In short, the performance of integrated PET/CT pattern analysis, an approach that more closely mirrors the actual clinical interpretation of PET/CT scans, was studied. No patients in this study showed the anticipated indeterminate PET/CT findings of an eccentric or perianastomotic PET pattern matched with a normal or diffuse thickening CT pattern. A larger study may have identified such examples, and until more data become available with respect to these indeterminate pattern combinations, caution should be used in their interpretation. That no such case was observed in this study likely reflects the tendency of anastomotic recurrences to originate in the perianastomotic tissues where they are often apparent on CT.
In conclusion, FDG PET/CT pattern analysis can be used accurately to assess for staple line recurrence after colorectal cancer resection. Perianastomotic or eccentric masses on CT with corresponding perianastomotic or eccentric FDG uptake on PET are highly predictive of staple line recurrence. Background, diffuse, curvilinear, and focal patterns of FDG uptake, regardless of intensity of FDG uptake, do not correlate with recurrence in the absence of a mass on CT. SUV measurements alone have low specificity in assessing for staple line recurrence because of physiologic and postsurgical factors that may increase FDG uptake at or including the staple line. Presacral abscesses and inflammatory scar tissue are potential sources of false-positive PET/CT interpretations.

Footnote

Address correspondence to P. B. Shyn ([email protected]).

References

1.
Rex DK, Kahi CJ, Levin B, et al. Guidelines for colonoscopy surveillance after cancer resection: a consensus update by the American Cancer Society and the U.S. Multi-Society Task Force on Colorectal Cancer. Gastroenterology 2006; 130:1865 –1871
2.
Thomson WH, Foy CJ, Longman RJ. The nature of local recurrence after colorectal cancer resection. Colorectal Dis 2007; 10:69 –74
3.
Abulafi AM, Williams NS. Local recurrence of rectal cancer: the problem, mechanisms, management and adjuvant therapy. Br J Surg 1994; 81:7 –19
4.
Guillem JG, Chessin DB, Cohen AM, et al. Long-term oncologic outcome following preoperative combined modality therapy and total mesorectal excision of locally advanced rectal cancer. Ann Surg 2005; 241:829 –838
5.
Tjandra JJ, Chan MK. Follow-up after curative resection of colorectal cancer: a meta-analysis. Dis Colon Rectum 2007; 50:1783 –1799
6.
Arriola E, Navarro M, Pares D, et al. Imaging techniques contribute to increased surgical rescue of relapse in the follow-up of colorectal cancer. Dis Colon Rectum 2006; 49:478–484
7.
Kobayashi H, Mochizuki H, Sugihara K, et al. Characteristics of recurrence and surveillance tools after curative resection for colorectal cancer: a multicenter study. Surgery 2007; 141:67 –75
8.
Kobayashi A, Sugito M, Ito M, Saito N. Predictors of successful salvage surgery for local pelvic recurrence of rectosigmoid colon and rectal cancers. Surg Today 2007; 37:853–859
9.
de Geus-Oei LF, Ruers TJ, Punt CJ, Leer JW, Corstens FH, Oyen WJ. FDG-PET in colorectal cancer. Cancer Imaging 2006; 6:S71 –S81
10.
Wiering B, Krabbe PF, Jager GJ, Oyen WJ, Ruers TJ. The impact of fluor-18-deoxyglucose-positron emission tomography in the management of colorectal liver metastases: a systematic review and metaanalysis. Cancer 2005; 104:2658 –2670
11.
Cohade C, Osman M, Leal J, Wahl RL. Direct comparison of 18F-FDG PET and PET/CT in patients with colorectal carcinoma. J Nucl Med 2003; 44:1797 –1803
12.
Even-Sapir E, Parag Y, Lerman H, et al. Detection of recurrence in patients with rectal cancer: PET/CT after abdominoperineal or anterior resection. Radiology 2004; 232:815–822
13.
Zasadny KR, Wahl RL. Standardized uptake values of normal tissues at PET with 2-[fluorine-18]-fluoro-2-deoxy-D-glucose: variations with body weight and a method for correction. Radiology 1993; 189:847 –850
14.
Grabbe E, Winkler R. Local recurrence after sphincter-saving resection for rectal and rectosigmoid carcinoma: value of various diagnostic methods. Radiology 1985; 155:305–310
15.
Fletcher JG, Johnson CD, Frueger WR, et al. Contrast-enhanced CT colonography in recurrent colorectal carcinoma: feasibility of simultaneous evaluation for metastatic disease, local recurrence, and metachronous neoplasia in colorectal carcinoma. AJR 2002; 178:283 –290
16.
Thoeni RF, Cello JP. CT imaging of colitis. Radiology 2006; 240:623–638
17.
Moore HG, Akhurst T, Larson SM, Minsky BD, Mazumder M, Guillem JG. A case-controlled study of 18-fluorodeoxyglucose positron emission tomography in the detection of pelvic recurrence in previously irradiated rectal cancer patients. J Am Coll Surg 2003; 197:22–28
18.
Israel O, Yefremov N, Bar-Shalom R, et al. PET/CT detection of unexpected gastrointestinal foci of 18F-FDG uptake: incidence, localization patterns, and clinical significance. J Nucl Med 2005; 46:758 –762
19.
Berger KL, Nicholson SA, Dehdashti F, Siegel BA. FDG PET evaluation of mucinous neoplasms: correlation of FDG uptake with histopathologic features. AJR 2000; 174:1005 –1008
20.
Lindholm P, Minn H, Leskinen-Kallio S, Bergman J, Ruotsalainen U, Joensuu H. Influence of the blood glucose concentration on FDG uptake in cancer: a PET study. J Nucl Med 1993; 34:1–6
21.
Sharma RI, Smith TA. Colorectal tumor cells treated with 5-FU, oxaliplatin, irinotecan, and cetuximab exhibit changes in 18-FDG incorporation corresponding to hexokinase activity and glucose transport. J Nucl Med 2008; 49:1386 –1394

Information & Authors

Information

Published In

American Journal of Roentgenology
Pages: 414 - 421
PubMed: 20093604

History

Submitted: April 11, 2009
Accepted: July 21, 2009

Keywords

  1. anastomosis
  2. colorectal cancer
  3. PET/CT
  4. recurrent colorectal cancer
  5. staple line

Authors

Affiliations

Paul B. Shyn
Division of Abdominal Imaging and Intervention, Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, 75 Francis St., Boston, MA 02115.
Rachna Madan
Division of Abdominal Imaging and Intervention, Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, 75 Francis St., Boston, MA 02115.
Christopher Wu
Division of Abdominal Imaging and Intervention, Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, 75 Francis St., Boston, MA 02115.
Ş. Mehmet Ertu̇rk
Department of Radiology, Sisli Etfal Training and Research Hospital, Istanbul, Turkey.
Stuart G. Silverman
Division of Abdominal Imaging and Intervention, Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, 75 Francis St., Boston, MA 02115.

Metrics & Citations

Metrics

Citations

Export Citations

To download the citation to this article, select your reference manager software.

Articles citing this article

Media

Figures

Other

Tables

Share

Share

Copy the content Link

Share on social media