| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Memorial Sloan-Kettering Cancer Center, 1275 York
Ave., Rm. C276F, New York, NY 10021.
2 Department of Nuclear Medicine, Memorial Sloan-Kettering Cancer Center, New
York, NY 10021.
3 Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, NY
10021.
4 Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY
10021.
5 Department of Radiology, Weill Medical College, Cornell University, New York,
NY 10021.
Received August 19, 2005;
accepted after revision December 7, 2005.
Address correspondence to M. J. Gollub
(gollubm{at}mskcc.org).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Eighteen patients with suspected colorectal polyps enrolled in this prospective study. Before colonoscopy, 17 of the patients underwent a combination of FDG PET and CTC. CTC consisted of 4-MDCT merged with PET. PET of the abdomen and pelvis was performed after each CTC scan. One radiologist and one nuclear medicine physician in consensus analyzed PET and CTC fusion data. PET standard uptake value was correlated with the findings at histologic examination of polyps. Patient feasibility was defined as the ability to tolerate prolonged scanning with good colonic distention. Technical feasibility was determined by how closely anatomically matched polyps overlapped on fusion images.
RESULTS. Seventeen of 18 patients tolerated scanning. Eighty-five percent of colon segments were optimally distended. Twenty-three of 27 FDG-avid polyps measuring 10 mm or more had excellent overlap at fusion imaging. PET depicted 23 of 39 premalignant polyps and even showed increased tracer activity associated with four small tubular adenomas (4-6 mm). Sixteen benign polyps (10-25 mm) were not depicted on PET. All nine cases of cancer (tumors measuring 11-60 mm) were detected with both PET and CTC. The standard uptake value of malignant tumors ranged from 4 to 20 (mean, 9). However, six benign flat polyps did not exhibit FDG avidity.
CONCLUSION. The novel combination of CTC and PET was feasible in 17 of 18 patients and allowed excellent image correlation in 23 of 27 proven polyps measuring 10 mm or more on PET-CTC fusion. This technique shows promise in accurate anatomic correlation of both malignant and premalignant lesions evaluated with FDG PET.
Keywords: colon • CT colonography • PET
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Wider application and use of 18F-FDG PET whole-body scanning led to a number of investigations in which incidental focal intense colonic FDG uptake was proved to represent colonic polyps. In these small, mainly retrospective studies, size, location, and morphologic features appeared to limit polyp detection with FDG PET [4-11]. Results of these imaging investigations point to obstacles that must be overcome for accurate identification of colonic polyps with current imaging techniques. CTC shows only structural abnormalities, and PET is limited by occasional nonspecific intestinal uptake. Unlike morphologic techniques such as CT, however, PET is a functional and molecular imaging tool that, when performed with FDG, shows the accelerated hexokinase activity present in neoplasia. We hypothesized that combining the strengths of these individual imaging tests might provide the best overall strategy for detecting biologically important polyps in the colon.
Our goals were to determine whether combined CTC and thoracoabdominopelvic PET performed with a commercially available fusion scanner would be technically capable of producing high-quality images with accurate anatomic coordination (fusion) of polyps; to obtain preliminary data on the ability of PET to depict colorectal adenoma; and to estimate whether additional information obtained with PET can improve the sensitivity and specificity of CTC in the detection of biologically significant colorectal adenoma. To the best of our knowledge and according to our literature review, our study was the first investigation of the combination of CTC and PET.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Six women and 12 men enrolled in the study. Patients cleansed their colons using a standardized sodium phosphate preparation consisting of two doses of Phospho-Soda (90 mL), four bisacodyl tablets (Fleet Prepare Kit 2, Fleet Pharmaceuticals), several 8-ounce (235-mL) glasses of water, and dietary restriction. One woman chose not to participate because of claustrophobia, and the other 17 patients completed all aspects of the study. The mean age of patients who underwent imaging was 61 years (range, 17-82 years; median, 72 years). Patients were referred for surveillance after previous removal of colon cancer (n = 4) and because polyps or cancer had been detected at medical evaluations at institutions other than ours (n = 14). Patients received an average of 19.8 mCi FDG (range, 15.7-22 mCi) 60-90 minutes before scanning. The average time the patient was in the imaging room was 77 minutes (range, 67-115 minutes). The volume of rectally administered CO2 was recorded for 16 patients and averaged 33.5 L (range, 14-65 L). The pressure of CO2 used was in the range of 12-25 mm Hg.
Colonoscopy was completed in 15 of 17 patients. Incomplete colonoscopy was due to inadequate preparation (n = 1) (repeated successfully the next day) and obstructing mass (n = 1). At surgery, palpation of the proximal colon revealed no additional lesions, and these results were used as the reference standard for this patient.
PET and CTC
All patients fasted for at least 6 hours before imaging. All images were
obtained with a dedicated PET-CT system that operates in 2D mode, has an axial
field of view of 15.5 cm, and has an axial slice thickness (resolution) of
4.2-mm (full width at half maximum) intensity at the center of the field of
view (Discovery LS, GE Healthcare Technologies). Images were acquired after IV
injection of 15-20 mCi (555-740 MBq) FDG and a 90-minute uptake period and
were reconstructed with a 16-subset two-iterations algorithm, 256 x 256
matrix, and segmented attenuation correction.
Intramuscular glucagon (2 mg) was injected several minutes before automated
rectal CO2 insufflation (ProtoCO2l, EZ-EM). With a
conventional enema tip supplied by the vendor and inserted by a physician
(including use of the retention balloon), CO2 was administered at
12-25 mm Hg pressure (manufacturer's recommendation, 25 mm Hg) according to
patient comfort throughout the examination. Before and after all PET-CT
acquisitions, scout imaging was performed to ensure adequacy of colonic
distention. If the colon was not distended, scanning was delayed until
adequate distention was confirmed. Prone CTC was performed during one
breath-hold (
25 seconds depending on body length) with the 4-MDCT
packaged into the PET-CT scanner at 0.8-second scan rotation, 5-mm section
thickness, 1.5 beam pitch, 80-200 mA, and 120-140 kV. Prone PET of the abdomen
and pelvis was performed with the patients' arms above their heads. Three or
four bed positions were used to cover the entire colon at 3-5 minutes per bed
position. Supine CTC was performed from the dome of the diaphragm to the pubic
symphysis during one breath-hold with the same arm position. Supine PET
covering the identical landmarks was performed. Finally, a routine chest PET
scan was obtained for completion of thoracoabdominopelvic PET. Patients were
instructed to breathe quietly and remain still. An abdominal binder attached
to the table was fastened around the patient to reduce motion from respiratory
excursion. Oral and IV contrast agents were withheld.
Image Processing
CT data were reconstructed according to two protocols. For CTC,
2.5-mm-thick slices (3.2-mm effective thickness) at 1.9-mm overlap were sent
to an advanced imaging processing workstation (Advantage Windows Workstation,
versions 4.0-4.2, GE Healthcare Technologies) and analyzed with Navigator
software. For anatomic coordination with 5.0-mm PET slices (index, 4.2 mm),
5-mm-thick slices at 5-mm intervals were sent with PET data to an advanced
workstation (Centricity AW suite 1.0 GE Healthcare).
Image Analysis
CT and PET images were interpreted first separately and then with fusion
techniques in a combined session to facilitate polyp matching between
techniques. One radiologist with experience in interpreting CTC images (
400 cases) used a primary 2D strategy to review all cases on a PACS
(Centricity 2.1, GE Healthcare). Prone and supine data sets were reviewed at
lung (window, 1,500 H; level, -500 H) and abdominal (window, 400 H; level, 30
H) settings for the presence of polyps and masses in eight colonic segments
(cecum, ascending colon, hepatic flexure, transverse colon, splenic flexure,
descending colon, sigmoid colon, and rectum). Location within the segments
(proximal, middle, distal) was also recorded subjectively. Abnormalities were
characterized by greatest diameter (in millimeters measured on axial 2D images
with electronic calipers), relation to the interhaustral fold (on or not on),
appearance (sessile, pedunculated, or flat), and in the case of masses,
annularity or semiannularity.
All PET scans were interpreted by fellowship-trained nuclear medicine physicians. For each of the colonic segments, if an abnormality was found, standardized uptake value (SUV) was recorded. Focal FDG uptake in the colon that was visually more intense than background was considered abnormal. Diffuse curvilinear FDG uptake was assumed to represent normal or nonmalignant bowel activity. No threshold SUV was used for identifying these abnormalities. SUV was calculated semiquantitatively for each FDG uptake focus with automated region-of-interest analysis and was calculated as maximum activity concentration detected in the lesion divided by injected activity and corrected for body weight as indicated by the formula SUVbw = Q x W / Qinj, where SUVbw is SUV normalized to body weight, Q is activity in the lesion measured in megabecquerels per liter, Qinj is injected dose measured in megabecquerels, and W is the patient's body weight in kilograms.
Colonoscopy
All colonoscopic procedures were performed within 3 hours of PET-CTC.
Sedation was used according to institutional standards. The segmental
unblinding method was used whereby after withdrawal of the endoscope from a
segment of colon, a research assistant revealed the PET-CTC findings to the
colonoscopist. If colonoscopy did not reveal an abnormality for a particular
segment but PET-CTC had, the colonoscopist reinserted the endoscope and
reevaluated the segment. Polyps and masses were measured by visual inspection.
Diagnostic or excisional biopsy was performed as clinically indicated.
Pathologic Examination
Polyps were fixed in formalin, stained with H and E, and microscopically
characterized according to polyp type and grade of dysplasia or malignancy if
present.
Feasibility Assessment
At scanning, patient feasibility was subjectively assessed as the patient's
ability to tolerate continuous CO2 administration into the colon
throughout the long examination; ability to tolerate prone and supine body
positions in the arms-above-head position; and ability to remain inside the
scanning bore for the duration of two CT and two PET scans. Technical
feasibility was present if gaseous distention of the colon allowed
satisfactory assessment of all segments in at least one position and resulted
in an appearance comparable with that of routine CTC images. The radiologist
interpreting the CTC images used the following descriptors for assessment:
optimal indicated effacement or near-effacement of interhaustral folds; fair,
some areas of less than optimal distention; and poor, opposite walls touching
in one or more areas or presence of deep interhaustral ridges. Four colonic
segments were assessed: right, transverse, left, and rectosigmoid in at least
one of the two scanning positions. Interpretive feasibility was judged as
ability to show coregistration of abnormalities between CTC and PET on fusion
images. Both observers used the cross-reference tool and digital measurement
cursor to rate interpretive feasibility on fusion images as excellent,
proximity within 2 cm, or separation greater than 2 cm.
Image Correlation with Colonoscopy (Reference Standard)
Polyps within the same segment or adjacent portions of adjacent segments
(e.g., distal descending and proximal sigmoid) were considered matched if they
were within 100% of the size measured by the other observer. Because we used
visual estimation, larger masses were considered matched if within 150% of the
other measurement.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Among 67 colonic segments rated for quality of distention (one patient had undergone hemicolectomy, and the right colon segment was absent), six sigmoid segments with inadequate distension due to diverticular disease were graded fair; two segments (one sigmoid segment with diverticular disease, one undistended cecum) were graded poor, and 59 segments were graded optimal. One patient had good distention but poor preparation. Preparation in all other patients was optimal.
Colonoscopy revealed 94 polyps. CTC depicted 10 (42%) of the 24 polyps measuring 0-5 mm, and PET depicted three (13%). CTC depicted 13 (42%) of the 31 polyps measuring 6-9 mm, and PET depicted one (3%). CTC depicted 27 (69%) of the 39 polyps 10 mm or larger (advanced adenoma), and PET depicted 23 (59%). The overall detection rates for the combined PET-CTC technique mimicked the rates for CTC, because PET did not depict additional polyps (Table 1).
|
Twenty-seven polyps or masses were FDG avid (Tables 2 and 3). The median polyp or mass diameter for FDG-avid polyps was 25 mm (range, 4-60 mm). The median SUV was 6.0 (range, 1.1-20.7). Sixteen polyps had completely benign histologic features (mean size, 23 mm; range 4-50 mm). These polyps had a mean SUV of 4.7 (range, 1.1-12.1). Nine of 27 polyps or masses were carcinoma (two cases of intramucosal component in otherwise benign polyps). All nine cases of carcinoma were detected with PET (mean size, 39 mm; range, 10-60 mm). The median SUV was 9.0 (range, 4.0-20.7). Two polyps were not histologically sampled. Six polyps were flat, and none of these was FDG avid. Sixteen polyps measuring > 10 mm were not seen on PET scans (Table 4).
|
|
|
Feasibility of Image Correlation
Twenty-three of 27 polyps exhibited excellent geometric coregistration at
fusion imaging (Figs. 1A,
1B,
1C,
1D,
2A, and
2B). Four (15%) of 27 FDG-avid
polyps were not perfectly superimposed on fusion images, but all were within 2
cm of one another (Figs. 3A,
3B,
3C,
3D,
3E,
3F,
3G, and
3H). These polyps measured 4,
20, 20, and 40 mm and were located in the distal sigmoid, proximal descending,
distal transverse, and proximal transverse colon segments.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In this pilot study, our primary aim was to assess the patient and
technological feasibility of the combination of PET and CTC and to correlate
polyp histologic findings with avidity for FDG. Challenges to patient
feasibility included the longer examination time for PET, which may prove too
taxing for claustrophobic patients and for patients trying to retain several
liters of colonic gas. Technical challenges included the time difference for
image acquisition between CTC ( 20-30 seconds per acquisition) and PET
( 9-16 minutes per full colonic acquisition depending on body length),
which can interfere with anatomic coregistration because of respiratory
misregistration and with maintenance of adequate colonic distention during
prolonged scanning. Strategies used to address these challenges included
administration of glucagon to reduce colonic motion, application of an
abdominal binder to minimize respiratory motion, continuous low-pressure
automated CO2 insufflation to maintain distention, and close
clinical supervision with verbal reassurance to eliminate claustrophobia.
From the standpoint of patient-related feasibility, outcome was excellent and better than expected. Ninety-four percent of patients tolerated scanning, 88% maintained the preferred arm position, 100% maintained gaseous distention in most of the colon, and 87% (59/67) of segments were optimally distended. Feasibility of image and polyp coregistration also was excellent: 85% of polyps of 10 mm or larger had excellent coregistration with CT images. The misregistered polyps were always within 2 cm of one another, and their spatial matching was facilitated by the high target-to-background focal signal of polyps. Misregistration was due to respiratory excursion and was most notable on supine images.
Our secondary aim was to perform a preliminary assessment (given the small study size) of the performance characteristics of PET-CTC in the detection of polyps. We used conventional colonoscopy as a reference standard, in particular to see whether PET would depict polyps not seen on CTC. Although CTC underperformed numerous results in the literature [2, 3], depicting only 27 of 39 polyps 10 mm or larger (and all cases of cancer), this detection rate is within the range reported in a 2005 multicenter study [13]. PET depicted more than one half of advanced adenomas and all malignant lesions but did not show polyps not seen on CTC. Our hypothesis that PET might reveal polyps hidden at CTC thus proved false. Five of six flat polyps, all with benign histologic features, were missed on CTC, and all six were missed on PET. The reason for this disappointing finding may be the use of 2.5-mm slices for CTC. Thinner collimation may eliminate this short-coming. That PET did not increase sensitivity for polyp detection is disappointing; however, because we found it feasible, this technique can be easily reproduced in a larger study of patients with large numbers of polyps to better elucidate the added value of PET.
Because the resolution of our PET machine was 4.0 mm in the axial plane, assessment of FDG uptake in small lesions was difficult to appreciate without anatomic correlation with a technique that has greater resolution. Therefore our goal was detection of polyps 6 mm and larger that had been missed on CTC. Even so, focal FDG avidity on PET correlated well with the presence of several small or intermediate polyps (4-6 mm), whereas most series in the literature report 10 mm as the lower limit for PET detection [4, 6, 7, 9]. Although the number of polyps was small, the increased sensitivity in our study was likely due to our method of fusing morphologic and functional images.
Several reports in the literature describe FDG avidity of colonic polyps [4-10]. Yasuda et al. [4] reported on 110 subjects who underwent PET and total colonoscopy. Fourteen (24%) of 59 polyps measuring 10-30 mm in 30 subjects undergoing total colonoscopy were detected on PET. The other 45 adenomas (size range, 5-16 mm) were not detected with PET. The positivity rate of PET increased with adenoma size. In 2005, Freidland et al. [10] performed prospective PET examinations on 45 patients with 58 colonic lesions referred for consideration for endoscopic resection. PET depicted only eight of 13 malignant tumors 2 cm or larger, three of 13 flat lesions, seven of 10 lesions larger than 3 cm, three of eight lesions measuring 2.0-2.9 cm, and two of 14 lesions 1.0-1.9 cm in diameter. In a retrospective review of our experience with routine whole-body PET over a 1-year period that involved 100 consecutive examinations with incidental abnormal abdominal uptake, seven of 14 cases of abnormal uptake were proved caused by tubular or tubulovillous adenomas measuring 8-15 mm [14]. Although published reports such as these indicate the ability of PET to depict polyps as incidental findings during scans for other reasons, these reports cannot address the true potential of FDG PET in depiction of colorectal polyps because of the confounding circumstances of unprepared and uncleansed colon. Accurate detection of colorectal polyps requires optimized patient and technical conditions. We believe our method addresses these needs and provides more accurate insight into how PET would perform in circumstances more akin to those resembling screening, diagnosis, and surveillance of colorectal carcinoma.
Among the most intriguing and surprising findings in our series was the lack of visualization on PET images of 16 polyps larger than 1 cm. For example, one patient had two 2-cm polyps in the transverse colon that were nearly identical colonoscopically and histologically and were clearly seen at CTC and colonoscopy, but only one of the polyps was FDG avid (Figs. 3A, 3B, 3C, 3D, 3E, 3F, 3G, and 3H). Although size may be one criterion for FDG avidity, as reported by Yasuda et al. [4], our data suggest that a more complex relation exists between polyps and their FDG visualization.
It is well established that FDG PET is useful in imaging primary and metastatic colorectal cancer [15]. We confirmed previous reports that FDG PET findings can be abnormal in premalignant lesions as well. Although frank adenocarcinomas, which were the largest lesions, had higher SUVs (mean, 9.5) than did adenomas (mean, 4.7), all histologic subtypes of premalignant polyps had FDG avidity, including tubular adenomas and even large hyperplastic adenomas (serrated adenomas) (Fig. 4) (Table 3). These results are encouraging, because they support use of another technique in addition to CTC and colonoscopy for identifying some cancer precursor lesions. Because of the functional nature of PET, this type of detection may reflect greater biologic importance. Further investigation is needed.
|
An encouraging finding was that several large right-sided polyps, currently known as serrated adenomas, were FDG avid. Unlike the highly prevalent, completely benign, small isolated hyperplastic polyps frequently found in the distal colon, serrated adenoma has predominantly hyperplastic histologic features but also has features that have been found to correlate with greater risk of malignant transformation. This relatively new entity of hyperplastic (or serrated) polyposis [16, 17] describes the condition of one of our patients. This patient had more than a dozen such polyps that appeared colonoscopically "malignant" and had associated right colon cancer with in situ serrated adenoma.
Although only 2.5 per 1,000 polyps per year develop into cancer, there are no adequate predictive criteria for malignancy except for statistical correlation with size and similar statistical correlation with the presence of a villous component [18]. Our group of patients was very small, but findings such as ours may lead to more refined histochemical tests and molecular criteria for further differentiation of apparently benign FDG-avid polyps from those that do not pick up the tracer. New molecular approaches in various stages of development may make it possible to avoid conventional colonoscopy when PET-CT performed with more specific tracers is used for reliable identification of premalignant lesions.
We made some curious observations when viewing insufflated colon segments at times well-distended and at other times collapsed. Areas of nodular or linear uptake seen in collapsed colonic segments disappeared during scanning of the same segments in the opposite position with the colon distended. We cannot explain this phenomenon but believe it may represent an artifact for physicists to address. Furthermore, fluid or stool seemed to exhibit FDG avidity at times. We postulate that this phenomenon is related to mucosal cellular sloughing after FDG cellular trapping. Kim et al. [6] proved that stool assays may have positive results for FDG.
Although numerous opportunities for false-positive FDG uptake existed (e.g., colonic mobility, respiratory misregistration), unlike many investigators [4, 5, 7, 9], we encountered only one false-positive finding on PET. In this case, intense focal activity was found near the transverse colon on prone imaging. On fusion imaging this activity was found to represent increased small-intestinal uptake. This high specificity of FDG PET further underscores that with the more robust technique of fusing images from a cleansed, insufflated colon (CTC) with PET rather than images from an uncleansed, collapsed colon, one can obtain data that is anatomically more accurate and useful regarding abnormal uptake in the colon.
This study had several limitations. The number of large polyps in one patient skewed the average size of premalignant polyps upward compared with the findings in an average population. At times, notable size discrepancies were found, likely because of a combination of factors, including our use of only axial images for measurements and the use at colonoscopy of the visual estimate technique rather than the open biopsy forceps method of measuring polyps. We found polyp correlation easier between CTC and PET than between imaging and colonoscopy. Accurate segmental localization of polyps is a known pitfall of colonoscopy compared with CTC and surgery in long tortuous colons. Numerous polyps had to be excluded in two patients with familial adenomatous polyposis and one patient with hyperplastic (serrated) polyposis syndrome, both of whom had small polyps too numerous to count or correlate with CTC-PET findings.
Interpreting intestinal uptake of FDG can be challenging and fraught with error. Nonpathologic activity from lymphoid tissue, muscular contraction, inflammation, sloughed cells, errors in attenuation correction, motion misregistration, and even artifact caused by collapsed loops compose a partial list of problems. False-negative activity can be caused by blurring of small foci from respiratory or peristaltic motion, gross shifts in bowel position over a long scanning time, and errors in attenuation correction. We believe the critical limitation of this fusion technique was the difference between examination times for PET and CTC. We controlled for every aspect we could to eliminate these errors, but we suspect that despite our best efforts, smaller lesions were missed owing to a blurring effect. We know of no way at present to avoid the limitation of mismatched scanning time requirements, but we remain optimistic that technological advances will lead to development of more efficient collimator crystals that allow simultaneous PET-CT data acquisition.
In summary, using an innovative combination of FDG PET and CTC for imaging colorectal polyps, we found the technique feasible and well-tolerated in most patients. FDG-avid polyps and masses had excellent geographic coregistration on fusion images, and all cancers were detected on both CTC and PET. Although PET did not depict polyps not seen on CTC, a study with a larger series of patients would address ways to improve sensitivity and specificity. More important perhaps is the finding that the relation of FDG avidity to size and histologic features may be more complex than previously believed, because large equivalent-sized polyps were variably FDG avid. These intriguing findings should stimulate further investigation into this exciting new imaging strategy and after further refinements lead to prediction of the premalignant nature of specific colonic polyps.
Acknowledgments
We thank Hedvig Hricak for her mentoring in this investigation and help
with the manuscript. We also thank Yusuf E. Erdi, Phillip B. Paty, W. Douglas
Wong, Nancy Kemeny, Robert Kurtz, Jinru Shia, Heiko Schoder, Moshe Shike, Hans
Gerdes, and Mark Schattner for their assistance with patient management and
conceptual input to the manuscript.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  R. J. Stanley The Challenge of Managing Prostate Cancer Am. J. Roentgenol., January 1, 2007; 188(1): 1 - 1. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
