Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online before print July 12, 2002, 10.1148/radiol.2243011362
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2243011362v1
224/3/748    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kinkel, K.
Right arrow Articles by Thoeni, R. F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kinkel, K.
Right arrow Articles by Thoeni, R. F.
(Radiology 2002;224:748-756.)
© RSNA, 2002

Evidence-based Practice

Detection of Hepatic Metastases from Cancers of the Gastrointestinal Tract by Using Noninvasive Imaging Methods (US, CT, MR Imaging, PET): A Meta-Analysis1

Karen Kinkel, MD, Ying Lu, PhD, Marcus Both, MD, Robert S. Warren, MD and Ruedi F. Thoeni, MD

1 From the Department of Radiology, Geneva University Hospital, Rue Micheli-du-Crest 24, CH 1211 Geneva 14, Switzerland (K.K.); Departments of Epidemiology (Y.L.), Radiology (K.K., Y.L., M.B., R.F.T.), and Surgery (R.S.W.), University of California, San Francisco; and Department of Radiology, Klinik für Diagnostische Radiologie de Christian-Albrechts-Universität zu Kiel, Germany (M.B.). Received August 10, 2001; revision requested October 9; revision received January 4, 2002; accepted January 29. Address correspondence to K.K. (e-mail: karen.kinkel@hcuge.ch).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
PURPOSE: To perform a meta-analysis to compare current noninvasive imaging methods (ultrasonography [US], computed tomography [CT], magnetic resonance [MR] imaging, and 18F fluorodeoxyglucose [FDG] positron emission tomography [PET]) in the detection of hepatic metastases from colorectal, gastric, and esophageal cancers.

MATERIALS AND METHODS: A MEDLINE literature search and review of article bibliographies and our institutional charts of patients with colorectal cancer identified data with histopathologic correlation or at least 6 months of patient follow-up. Two authors independently abstracted data sets and excluded data without contingency tables or data published more than once. Summary-weighted estimates of sensitivity were obtained and stratified according to specificity of less than 85% or 85% and higher. A covariate analysis was used to evaluate the influence of patient- or study-related factors on sensitivity.

RESULTS: Among 111 data sets, nine US (509 patients), 25 CT (1,747 patients), 11 MR imaging (401 patients), and nine PET (423 patients) data sets met the inclusion criteria. In studies with a specificity higher than 85%, the mean weighted sensitivity was 55% (95% CI: 41, 68) for US, 72% (95% CI: 63, 80) for CT, 76% (95% CI: 57, 91) for MR imaging, and 90% (95% CI: 80, 97) for FDG PET. Results of pairwise comparison between imaging modalities demonstrated a greater sensitivity of FDG PET than US (P = .001), CT (P = .017), and MR imaging (P = .055).

CONCLUSION: At equivalent specificity, FDG PET is the most sensitive noninvasive imaging modality for the diagnosis of hepatic metastases from colorectal, gastric, and esophageal cancers.

© RSNA, 2002

Index terms: Computed tomography (CT), comparative studies, 70.1211 • Gastrointestinal tract, neoplasms, 78.32, 78.3327 • Liver neoplasms, metastases, 761.3327 • Magnetic resonance (MR), comparative studies, 70.1214 • Positron emission tomography (PET), comparative studies, 70.12163 • Ultrasound (US), comparative studies, 70.1298


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
More than 59,000 patients in the United States will manifest hepatic metastases from cancers of the gastrointestinal (GI) tract each year (1,2). In patients with colorectal cancer, the early detection of hepatic metastases offers the opportunity to perform partial hepatic resection or intrahepatic arterial infusion chemotherapy. Both treatment options have been shown to increase survival (3–5).

To identify patients with hepatic metastases and guide treatment decisions in patients with GI cancer, a large number of noninvasive imaging methods are available and are used in conjunction with measurement of carcinoembryonic antigen (CEA) levels and hepatic function tests. Whereas CEA measurements suffer from low sensitivity (56%–59%) (6,7), reported sensitivities of imaging methods range from 57%–100% for ultrasonography (US), 36%–94% for computed tomography (CT), 69%–96% for magnetic resonance (MR) imaging, and 86%–99% for positron emission tomography (PET) by using the tracer fluorine 18 (18F) fluorodeoxyglucose (FDG) (8). There is no consensus on the most sensitive imaging method for the detection of hepatic metastases in patients with colorectal cancer. Although results of studies comparing US with CT suggested a higher sensitivity of CT, none reached statistical significance (6,9,10). Investigators in a multiinstitutional study comparing CT with unenhanced MR imaging (11) in 365 patients and a study comparing CT with MR imaging enhanced with superparamagnetic iron oxide (SPIO) particles (12) also found no significant differences in performance among imaging modalities. More recently, FDG PET has shown higher sensitivities than CT in the detection of hepatic metastases during initial staging (13) or during surveillance of patients with colorectal cancer (14). Possible reasons for the absence of differences among imaging methods could be related to insufficient sample size or equivalence in performance among the various imaging modalities.

To our knowledge, studies comparing all available imaging methods within the same patient population do not exist and are unlikely to be performed because of increased cost and longer work-up time. A meta-analysis enables comparison of various imaging methods through a systematic review of the literature by combining previously published work in a summary estimate of sensitivity and specificity for each imaging modality (15). The purpose of our study was to perform a meta-analysis to compare current noninvasive imaging methods (US, CT, MR imaging, and FDG PET) in the detection of hepatic metastases from colorectal, gastric, and esophageal cancers.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Data Sources
A comprehensive literature search of English-language abstracted studies of human subjects was performed by using the MEDLINE database and search algorithms with the key words "liver metastasis" and/or "liver metastases"; "hepatic metastasis" and/or "hepatic metastases"; "ultrasonography," "sonography," "ultrasound," and/or "US"; "computed tomography" and/or "CT"; "magnetic resonance imaging," "magnetic resonance," "MRI," and/or "MR"; and "positron emission tomography" and/or "PET." Articles published from December 1985 to December 2000 were searched and included publications in English, German, French, Italian, and Spanish. Only articles that presented raw data were included (see "Study Selection," below). Review articles without original data and case reports were excluded. Our own institutional database at the University of California, San Francisco, was analyzed to find patients with a diagnosis of colorectal cancer who underwent partial hepatic resection since 1993 (unpublished data set) for possible data inclusion.

Study Selection
One thousand two hundred sixty abstracts were analyzed for concordance with the inclusion criteria. Inclusion criteria for meta-analysis were as follows:

1. Patients with a diagnosis of colorectal or other GI cancer, which potentially generates hypointense hepatic metastases, such as cancer of the stomach or esophagus. Primary tumors of the GI tract other than colorectal cancer were included to increase the number of available data sets. Cancers with hyperintense hepatic metastases, such as islet cell tumors of the pancreas, cancer of the gallbladder or biliary tract, any neuroendocrine cancer of the GI tract, or sarcoma, were excluded. We excluded GI cancers that generated hyperintense hepatic metastases to avoid a bias in favor of imaging methods that use intravenous contrast media, since those methods are more likely to depict hyperintense than hypointense hepatic metastases.

2. For data presented on a per-patient basis, the standard of reference had to be a histopathologic finding (surgical specimen, core biopsy, or positive cytologic finding) of at least one site of hepatic metastasis or 6-month to 1-year follow-up. For data presented on a per-lesion basis, the reference standard had to be a histopathologic finding with or without use of intraoperative US for all hepatic metastases.

3. CT had to be performed with use of intravenous (IV) contrast media, and PET had to be performed with IV administration of FDG.

4. Interpretation of imaging data had to be performed while blinded to pathologic findings.

5. Per-patient data had to allow calculation of true-positive, true-negative, false-positive, and false-negative results for imaging methods. Per-lesion data had to allow calculation of true-positive and false-negative results for imaging methods.

6. The data or subsets of the data were not published more than once.

Exclusion criteria for meta-analysis were as follows:

1. The data included tumors other than GI cancers that generated hypointense hepatic metastases, or data were not presented separately for colorectal, gastric, and esophageal cancers.

2. The primary cancer was not specified.

3. No IV contrast material was used during CT.

4. Images were interpreted without being blinded to histopathologic findings.

5. Adequate standard of reference was missing.

6. Data were incomplete.

7. Data were presented as a case report.

8. Data were published twice. When data were published more than once and all inclusion criteria except for number 8 were met, we included the study with the most patients and excluded the study with the fewest patients.

From the 1,260 abstracts, 106 data sets seemed likely to fulfill the inclusion criteria and were retrieved for study selection and data abstraction. Four additional relevant data sets were found either by manually checking the reference lists of retrieved articles (two data sets) or by relying on the literature knowledge of an expert (reviewer of a journal who indicated two additional data sets). Therefore, 110 published data sets and one unpublished data set (from the University of California, San Francisco database) were analyzed for inclusion in this meta-analysis.

Data Extraction
Two readers (K.K., R.F.T.) abstracted the data from each article. Disagreement was resolved by means of consensus. The readers were not blinded to the authors, journal name, or year of publication. The following data were recorded for each article:

1. Author and year of publication.

2. Sample size (number of patients, metastases, and benign hepatic lesions).

3. Patient context: staging before treatment, secondary staging before resection of hepatic metastases, or mixed population.

4. Location of primary GI tumor (colorectal, gastric, or esophageal).

5. Imaging modality: US, CT, MR imaging, or FDG PET.

6. Imaging technique for US (with or without color Doppler imaging), CT (helical vs nonhelical, drip vs bolus injection), MR imaging (field strength, use and type of IV contrast media [gadolinium-based or SPIO]), and FDG PET (amount of tracer administered, type of image analysis). In patients who underwent imaging with more than one modality or technique, the result of each imaging technique was counted as a separate data set.

7. Congruent true-positive, true-negative, false-positive, and false-negative results for the differentiation of patients with hepatic metastases from patients without hepatic metastases; and congruent true-positive and false-negative results, and, when possible, false-positive results for the differentiation of hepatic metastases from other hepatic lesions in patients with at least one site of hepatic metastasis. Congruent data were defined as data that were presented without major errors (eg, different numbers in tables and text). We checked the concordance of the sensitivity stated in the text with the sensitivity that was calculated by using the numbers from tables. In the event of small errors (eg, 83.4% calculated from the table and 84% mentioned in the text), we used the calculated sensitivity from the raw data in the table.

8. Prospective or retrospective study.

9. Uni- or multimodality study.

10. Percentage of patients without histopathologic findings but with follow-up as standard of reference.

11. Length of follow-up.

Data on a per-Lesion Basis from the Authors’ Institutional Database
We retrospectively reviewed our own institutional database of 102 patients, who, following imaging, underwent surgical removal of hepatic metastases from colorectal cancer between 1992 and 1999. Approval was obtained from the institutional review board to review the patients’ records. Additional patient informed consent was not required, since patients had already agreed to be included in our database. This database included reports of preoperative imaging examinations and intraoperative US, histopathologic reports from hepatic resection specimens and biopsies, and follow-up until death. Among these 102 patients, 98 underwent CT during arterial portography, 34 underwent MR imaging with gadolinium-based enhancement, four underwent unenhanced MR imaging, two underwent MR imaging with SPIO enhancement, and one underwent MR imaging with manganese-based enhancement. Data concerning preoperative CT during arterial portography were excluded because they were based on an invasive method. Imaging methods that were performed in fewer than five patients were also excluded from the analysis to avoid random sampling errors. Only the group of 34 patients who had undergone MR imaging with gadolinium-based enhancement matched the inclusion criteria for our meta-analysis. To ensure blinded image interpretation, studies were included in the analysis only if the MR imaging report had been finalized before the date of any image-guided biopsy or the date of partial hepatic resection. The mean time of follow-up was 33 months (minimum, 10 months; maximum, 92 months). The mean number of follow-up examinations was 2.7 (range, 1–10 examinations). The sensitivity and false-positive rates were calculated by comparing the number of lesions identified at MR imaging with the number of metastases shown in the pathologic report of the resected hepatic specimen. Pathologic results in all 34 patients were reported with full knowledge of the results of intraoperative US to maximize the number of pathologically confirmed hepatic metastases.

Ranking of Data Sets
A quality rating of 1 (lowest) to 3 (highest) was applied to published data sets and depended on the presence of clearly defined inclusion criteria for the study group (one point) and diagnostic criteria for image interpretation (one point) and congruence of presented data (one point). The rating of each data set was obtained by adding the number of points.

Statistical Analysis
Statistical analyses were performed by using statistical packages SAS version 8.2 (SAS Institute, Cary, NC) and S-plus (Insightful, Seattle, Wash). A goodness-of-fit test, which was based on the analysis of deviance for generalized linear models (16), was used to evaluate the appropriateness of using summary receiver operating characteristic (ROC) curves to perform meta-analyses (15,17,18). In the current study, we tested the linear relationship between D (logit true-positive rate - logit false-positive rate) and S (logit true-positive rate/logit false-positive rate) (15,19). Because our data from all included data sets failed the goodness-of-fit test for linear regression models of summary ROCs (no linear relationship between D and S), we compared sensitivities of different modalities by controlling specificities through stratification. We stratified our data sets into three categories: those with specificities lower than 85%, between 85% and 95%, and 95% or higher. By comparing sensitivities with the use of specificity strata to control differences, we evaluated the overall performance of the methods. Because a survey of surgeons and oncologists from our institution indicated that an imaging method is clinically useful only if its specificity is higher than 85%, we also repeated our analyses by excluding studies with a specificity lower than 85%.

To compare the sensitivity of imaging modalities, we first obtained normalized weighted sensitivities through a variance-stabilizing transformation of sensitivity. For a sensitivity p, we created a new variable, Z = arcsin({surd}p). After this transformation, Z is approximately normal with a variance of 1/(4n), where n is the total number of true-positive cases (16). We compared imaging modalities and effects of covariates on the basis of normal variables (Z values) by using weighted generalized linear random effects models. The weight used in these analyses was the inverse of the variance of Z. Least squared means of Z in the generalized linear random effects models were used to estimate the means of the transformed sensitivities and their 95% CIs for each modality. The inverse transformation of Z provided an estimation of the mean sensitivity and its 95% CIs. We used the random effects models to control the clustering effect of data sets from the same articles that resulted from multimodality examinations (20) and/or the sensitivities calculated on the basis of lesions or patients. The Tukey-Kramer procedure for adjustment of multiple tests was used for the pairwise comparison of sensitivities between imaging modalities (21). A P value of less than .05 was considered to indicate a significant difference.

Covariate Analysis
To determine whether imaging results were significantly affected by heterogeneity between individual studies, the following covariates were extracted and their influence on sensitivity analyzed: year of publication, prospective or retrospective study, uni- or multimodality study, type of data analysis (per lesion or per patient), methodologic quality rating, hospital type (university vs community), country of origin, prevalence of hepatic metastases, number of lesions per patient, patient context, location of primary GI tumor (colorectal, gastric, or esophageal), and length of follow-up. A subgroup analysis compared gray-scale US with color Doppler US, helical CT with nonhelical CT, unenhanced MR imaging with gadolinium-enhanced or SPIO-enhanced MR imaging, and FDG PET with and without the use of CT.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Literature Search
For the detection of hepatic metastases with the use of US, CT, MR imaging, or PET in patients with GI cancer, 111 data sets were retrieved from 78 published articles and one unpublished source (the authors’ institutional database) (Table 1). Fifty-four data sets (47%) from 32 articles and the authors’ institutional database fulfilled the inclusion criteria (6,9–14,22–46). In 15 articles, one imaging modality alone was evaluated (unimodality studies), versus 18 articles in which more than one modality was evaluated (multimodality studies). Data were presented per patient in 38 data sets and per lesion in 16 data sets. Four articles provided information for both a per-patient and a per-lesion analysis and were therefore counted as separate data sets for the covariate analysis (12,37,42,43). The most common reasons for exclusion from this meta-analysis were an inadequate standard of reference, incomplete data, and/or hepatic metastases from primary tumors other than colorectal, gastric, or esophageal cancers.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Results of MEDLINE Search for Data Sets to Evaluate Detection of Hepatic Metastases in Patients with GI Cancer

 
For each imaging modality, included data sets with corresponding years of publication and pooled patient characteristics, as well as percentages of patients with colorectal cancers, are presented in Table 2. Among the 54 included data sets (nine US, 25 CT, 11 MR imaging, and nine FDG PET), eight data sets from five studies and our own institutional database did not report true-negative results (25,38–41). In these data sets, results were analyzed by lesion, and true-negative results were considered to represent a potentially infinite number of "absent hepatic metastases" (in patients without hepatic metastases at long-term follow-up examination). Therefore, specificity could not be assessed for these studies. In articles providing both per-patient and per-lesion data sets, the per-patient data set was chosen for the comparison of sensitivity according to specificity strata, and both types of data sets were used for the covariate analysis comparing per-lesion results with per-patient results.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Study Characteristics of Patients Undergoing Imaging for Detection of Hepatic Metastases

 
Figure 1 shows the distribution of sensitivity and specificity in the 39 data sets with complete data. Results of goodness-of-fit tests for summary ROC models suggested that the use of summary ROC was inappropriate for analysis of our data, whether we included all studies in one linear model, regardless of the imaging modality (P < .005), or used a generalized linear model that allowed different intercepts and slopes for each imaging modality (P < .001).



View larger version (31K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Distribution of true-positive rate (sensitivity) and false-positive rate (1 - specificity) for data sets analyzing hepatic metastases from colorectal, gastric, or esophageal cancers with US, CT, MR imaging, and FDG PET. Points indicate individual data sets. The two dotted lines show the stratification of data into different areas of specificity (left = specificity above 95%; between the two dotted lines = specificity of 85%-95%; right = specificity below 85%). With the exception of one US and one CT study (upper right corner), the range of specificity is relatively small. However, sensitivity data vary substantially.

 
Comparison of Sensitivities of Noninvasive Imaging Techniques
The mean weighted sensitivity of each noninvasive imaging technique with its 95% CIs is presented in Table 3. The Table shows the sensitivity of all 54 included data sets, and this group is further subdivided according to corresponding results of specificity (stratification group). The stratification group included all 39 data sets for which specificity results were available, and results were evaluated per patient (eight US, 20 CT, four MR imaging, and seven FDG PET data sets). A specificity of at least 85% was considered clinically useful, leading to a subgroup of data sets with an 85% specificity or higher.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Sensitivity of Noninvasive Imaging Techniques for the Detection of Hepatic Metastases in Patients with GI Tumors: Data Stratification by Specificity

 
The comparison of sensitivities showed significant differences between imaging methods in all data sets (P = .007), in the stratification group (P = .043), and in the subgroup of data sets with a specificity of at least 85% (P = .007). In all data sets, the sensitivity of FDG PET was significantly higher than that of all other imaging modalities, without differences between US, CT, and MR imaging. In the stratification group, FDG PET was not different from MR imaging (P = .353) but demonstrated greater sensitivity than US (P = .011) or CT (P = .012).

Within the subgroup of data sets with an 85% specificity or higher, pairwise comparison between techniques showed that the sensitivity of FDG PET was significantly higher than that of US (P = .001) and CT (P = .012). The difference in sensitivity became marginally significant when FDG PET was compared with MR imaging (P = .055). There were no significant differences between US, CT, and MR imaging.

Covariate Analysis
The influence of cofactors on sensitivity was tested for the stratification group (data sets with available specificity results, regardless of whether results were evaluated per patient or per lesion) and for the subset of data with an 85% specificity or higher. Within the 47 data sets for which specificity results were available (including eight data sets from four studies presenting results per lesion), the type of data analysis (per lesion or per patient) affected sensitivity significantly (P = .012), with lower sensitivities when results were analyzed by lesion (sensitivity, 0.62; 95% CI: 0.49, 0.75) than by patient (sensitivity, 0.81; 95% CI: 0.73, 0.87). This difference became marginally significant for the subset of data with an 85% specificity or higher (P = .062).

The results of technical cofactors are presented in Table 4. The comparison between nonhelical and helical CT or between unenhanced MR imaging and MR imaging with SPIO enhancement showed no differences in mean weighted sensitivity. MR imaging studies with IV gadolinium-based enhancement could not be included in this covariate analysis because of absent values of specificity for all three MR imaging studies (40,41, and institutional database). However, the mean weighted sensitivity of the three studies in which gadolinium-based enhancement was used was 0.69 (95% CI: 0.53, 0.83). There was no significant difference in sensitivity between studies in which FDG PET was used alone versus those in which FDG PET was used in combination with helical CT. All other cofactors (year of publication, prospective or retrospective study, uni- or multimodality study, methodologic quality rating, hospital type [university vs community], country of origin, prevalence of hepatic metastases, number of lesions per patient, patient context, location of primary GI tumor [colorectal, gastric, or esophageal], and length of follow-up) had no influence on sensitivity.


View this table:
[in this window]
[in a new window]

  		TABLE 4. Influence of Technical Cofactors on the Sensitivity of Noninvasive Imaging Techniques for the Detection of Hepatic Metastases in Patients with GI Tumors

 
Within the subset of data with an 85% specificity or higher, we observed a significant effect of the prevalence of metastases in the study (P = .018). Figure 2 shows the association between prevalence of metastases and sensitivity of imaging modalities. The sensitivity increased with the prevalence of metastases for all imaging modalities. This relationship was not affected by further subgroups of specificity (85%–95% and higher than 95%). At equivalent prevalence, significant differences were maintained between imaging techniques (P = .027).



View larger version (27K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Association between true-positive rate and prevalence of metastases for data sets with an 85% specificity or higher. Observed sensitivities for US, CT, MR imaging, and FDG PET increased with prevalence of metastases (P = .018). The expected sensitivities derived from the generalized linear random effects model were plotted in lines according to imaging techniques. Points indicate individual data sets.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
 
In patients with GI tumors, particularly in those with colorectal cancer, the detection of hepatic metastases is crucial for treatment planning and improved outcome. Therefore, choosing an imaging modality with the highest sensitivity at an acceptable rate of specificity is of utmost importance. To our knowledge, this meta-analysis is the first report that assesses and compares summary estimates of sensitivity for four noninvasive imaging methods that are currently used for depicting hepatic metastases in patients with GI cancer. Through the comparison of sensitivities stratified by specificity, we also compared studies with a clinically relevant specificity (>=85%, as well as analyses using three strata of <85%, 85%–95%, and >=95%). The purpose of this subgroup analysis was to identify an imaging method with high sensitivity and an acceptable rate of false-positive diagnoses for benign hepatic lesions to limit inadequate treatment decisions or unnecessary hepatic biopsies. In this clinical context, if FDG PET is compared with US, CT, and MR imaging, the results of our meta-analysis demonstrate that FDG PET is the most sensitive noninvasive imaging method for the detection of hepatic metastases from colorectal, gastric, or esophageal cancers.

One major drawback of PET is the lack of detailed anatomic information in the area of significant tracer uptake. Clinical decision making requires confirmation of the intrahepatic location of abnormal tracer uptake, the hepatic segment involved, and the relationship between the lesion and the surrounding vessels. Results of our study suggest that high sensitivity values are obtained when FDG PET results are interpreted in conjunction with the use of CT. For the diagnosis of hepatic metastases, however, the results of this meta-analysis did not demonstrate significant differences between FDG PET results interpreted with or without the use of CT.

Because of increased cost, FDG PET cannot be used in all patients with GI tumors. Usually, a diagnostic test has the greatest utility when the pretest probability of disease is in the intermediate range of 20%–50% (47). The prevalence of hepatic metastases at the time of initial diagnosis is 20%–40% for colorectal, 50% for esophageal, and 12% for gastric cancer, versus 100% for all patients at autopsy (48,49). In patients harboring hepatic metastases, however, the chance of increased survival is usually limited to patients with metastases of colorectal origin. Indeed, the liver is the first site of metastases in patients with colorectal cancer (50), unless the tumor is located in the lower third of the rectum. The disease course of gastric or esophageal cancer is different because of the presence of other sites of metastases when hepatic metastases are identified. Therefore, FDG PET should be used, particularly for patients with colorectal cancer when hepatic resection is an issue. Park et al (51) compared the cost-effectiveness of CT with that of the combined "CT + FDG PET strategy" in patients with colorectal cancer. The CT + FDG PET strategy was found to be cost effective for patients with elevated CEA levels who were candidates for hepatic resection. The authors determined that this strategy increased the mean cost by $429 per patient but resulted in an increase in mean life expectancy of 9.527 days. Several authors (52–54) recently reported the high incremental value of FDG PET in patients with normal conventional imaging findings and increasing CEA levels after treatment for colorectal cancer. The information provided by using FDG PET changed treatment decisions in 61%–94% of these patients (52,53). Preliminary findings regarding survival of patients evaluated with FDG PET before hepatic resection for metastatic colorectal carcinoma are promising. Findings showed an increase in the 3-year disease-free survival rate to 40% (55) from 15%–28% in previously published series (56).

Another theoretical advantage of using PET during staging is the possibility of detecting extrahepatic foci of metastatic disease, particularly in areas in which CT has failed to demonstrate high sensitivity, such as in detecting lymph node invasion in the upper abdomen. The overall sensitivity of CT for depicting metastases in lymph nodes of patients with colorectal cancer is about 45% (57). However, the role of FDG PET in the evaluation of extrahepatic sites in patients suspected of having metastases from colorectal cancer has not yet been fully established and requires further evaluation. Experts in nuclear medicine conducted most of these FDG PET studies in an optimal clinical setting. Reproducibility of this new technology remains an issue that has, to our knowledge, not yet been sufficiently addressed in the current literature.

Comparison of our data with the results of a recent meta-analysis of patients suspected of having recurrent hepatic colorectal cancer shows that our 88% summary estimate of the sensitivity of FDG PET is lower than the 96% sensitivity of FDG PET reported in the meta-analysis of Huebner et al (8). One explanation might be a lower performance of PET in patients referred for initial staging compared with patients referred for suspected recurrence. This hypothesis was not confirmed by the results of our covariate analysis, which did not show any relationship between patient context and the sensitivity of FDG PET. Interestingly, the data in the meta-analysis of Huebner et al (8) also failed to fit the appropriateness for the use of summary ROC curves but limited the summary estimate for the sensitivity of PET to a pooled weighted sensitivity without further stratification.

Our statistical method differs from other published meta-analyses that compared the performance of imaging modalities. The standard method consists of the comparison of summary ROC curves with their corresponding Q* points (points on the summary ROC curve where sensitivity equals specificity) to limit the influence of different trade-offs between sensitivity and specificity. However, this approach requires that sensitivities and specificities of investigated studies follow a specific regression relationship. Failure to satisfy such a relationship will lead to a biased and inefficient meta-analysis. In the present study, we compared sensitivities of different imaging methods on the basis of equivalent specificities. We limited a potential influence of specificity by stratifying the sensitivity results according to corresponding specificities. Our method has the advantage that no model assumption between sensitivity and specificity is necessary. It can include all sensitivity results in the overall evaluation of imaging methods, even when some studies had missing specificities. We excluded those studies with missing specificities when we performed the stratified analysis. Although not presented here, we also performed a summary ROC curve analysis to allow comparison between our results and the comparison of summary ROC curves. The comparison of the summary ROC curves showed a similar trend in the performance of imaging methods (FDG PET is the best imaging method); however, the differences in Q* points failed to reach statistical significance because the summary ROC model was inappropriate and had less power in our study.

A limitation of the present study is the incomplete reporting of study methods and results in the published literature, resulting in a relatively small number of included data sets and technical subgroup analyses, particularly for MR imaging and PET. By adding gastric and esophageal tumors to colorectal tumors, we could increase the number of data sets for CT and US by 25%. Results of covariate analysis showed that the origin of the primary tumor had no influence on the detection of hepatic metastases with any imaging modality. Insufficient reporting of study methods, such as patient characteristics or imaging methods, has been noted by others (8) and is particularly evident when imaging literature is published in the nonradiologic literature. Although the mean quality rating did not differ between imaging modalities, studies dealing with PET lacked detailed description of image acquisition and analysis. Therefore, any technical cofactor in PET, such as tracer amount, threshold for significant tracer uptake, or the type of image analysis, could not be assessed.

Despite a 15-year range of publication, we did not identify an influence of year of publication on lesion detection. We have no explanation why changes in technology over the past 15 years had no significant effect on the sensitivity of hepatic metastases detection.

Because of the restricted availability and high cost of FDG PET, the question of an alternative imaging modality remains relevant. Investigators assessing the reproducibility of MR imaging to differentiate hemangiomas from hepatic metastases reported excellent results (58). Our meta-analysis estimated the sensitivity of MR imaging with SPIO enhancement to be 67%, representing, after helical CT, an alternative to the PET approach. The large 95% CI, ranging from 33% to 93%, reflects the heterogeneity of reported results in the literature. Said et al (42) compared the value of unenhanced MR sequences with that of ferumoxide-enhanced MR sequences in the detection of early hepatic recurrence in 19 patients with colorectal cancer. Both strategies demonstrated a low sensitivity of 42%, without any additional metastases identified with the SPIO-enhanced sequence (42). Earlier studies reported higher sensitivities of 72% and 80% for SPIO-enhanced MR imaging (12,39). The value of MR imaging enhanced with mangafodipir (Mn-DPDP, Teslascan; Nycomed Amersham, Oslo, Norway), another liver-specific MR imaging contrast agent, could not be assessed in this meta-analysis, because breast and pancreatic cancer were evaluated together with colorectal carcinoma (59).

Results of our covariate analysis demonstrated a significantly lower sensitivity in studies analyzing results per lesion than in studies analyzing results per patient. The proportion of studies analyzing results per lesion was 73% (eight of 11) for MR imaging, 21% (five of 24) for CT, 22% (two of nine) for PET, and 11% (one of nine) for US. Therefore, because of the high proportion of MR imaging studies analyzing results per lesion, our meta-analysis is likely to underestimate the sensitivity of MR imaging. Our meta-analysis also demonstrated that the sensitivity of an imaging modality to depict hepatic metastases increases with the prevalence of hepatic metastases among the patient population. This relationship is known as spectrum effect and corresponds to differences in severity of the disease among study populations (60). However, these differences in prevalence do not explain differences in sensitivity between imaging methods. Indeed, Figure 2 shows that at equivalent prevalence of hepatic metastases, differences in sensitivity among PET and other imaging modalities persist.

Many large institutions continue to use CT during arterial portography in patients who might be candidates for hepatic resection. This modality is reported to be the most sensitive invasive method for the detection of hepatic metastases from colorectal cancer, with a sensitivity of 93% and a false-positive rate of 25% (12,40). It is an invasive technique, however, because of catheterization of the superior mesenteric or splenic artery and is not practical for screening. Furthermore, a false-positive rate of 25% is unacceptably high and necessitates careful examination of the liver with intraoperative US. If presurgical evaluation is based on CT findings during arterial portography, surgical resection may not be considered in some patients with benign lesions in addition to metastases. Results of a recent study (61) have shown that radiologists’ performances at combined unenhanced, gadolinium-enhanced, and ferumoxide-enhanced MR imaging were similar to their performances at combined helical CT during arterial portography and biphasic CT during hepatic arteriography for the preoperative detection of malignant hepatic tumors. This makes a strong argument for using enhanced MR imaging in these patients when PET is not available or is considered too costly.

In conclusion, on the basis of results of our meta-analysis, PET represents the most sensitive noninvasive imaging modality for the detection of hepatic metastases from colorectal, gastric, or esophageal cancers with equivalent specificity. This technique might be particularly helpful in patients with increasing CEA levels and normal imaging findings when hepatic resection is planned. Standardized protocols of image acquisition and interpretation of FDG PET studies are necessary before this promising method can be widely used.


   FOOTNOTES
 
Abbreviations: CEA = carcinoembryonic antigen, FDG = 18F fluorodeoxyglucose, GI = gastrointestinal, IV = intravenous, ROC = receiver operating characteristic, SPIO = superparamagnetic iron oxide

Author contributions: Guarantors of integrity of entire study, K.K., Y.L., R.F.T.; study concepts, K.K., R.S.W., R.F.T.; study design, K.K., Y.L., R.F.T.; literature research, K.K., M.B.; clinical studies, R.S.W.; data acquisition, K.K., M.B., R.F.T.; data analysis/interpretation, Y.L., K.K.; statistical analysis, Y.L.; manuscript preparation, K.K., Y.L.; manuscript definition of intellectual content, K.K., Y.L., R.F.T.; manuscript editing, R.F.T.; manuscript revision/review, R.S.W., R.F.T.; manuscript final version approval, all authors.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 

  1. American Cancer Society. Cancer facts and figures: 1999 Atlanta, Ga: American Cancer Society, 1999.
  2. Fong Y, Cohen AM, Fortner JG, et al. Liver resection for colorectal metastases. J Clin Oncol 1997; 15:938-946.[Abstract/Free Full Text]
  3. Lehnert T, Knaebel HP, Duck M, Bulzebruck H, Herfarth C. Sequential hepatic and pulmonary resections for metastatic colorectal cancer. Br J Surg 1999; 86:241-243.[CrossRef][Medline]
  4. Nakamura S, Suzuki S, Baba S. Resection of liver metastases of colorectal carcinoma. World J Surg 1997; 21:741-747.[CrossRef][Medline]
  5. Tanada M, Saeki T, Takashima S, Mogami H, Hyoudou I, Jinno K. Intrahepatic arterial infusion chemotherapy for the colon cancer patients with liver metastases: a comparison of arterial embolization chemotherapy versus continuous arterial infusion chemotherapy. Gan To Kagaku Ryoho 1996; 23:1440-1442[Japanese].[Medline]
  6. Ohlsson B, Tranberg KG, Lundstedt C, Ekberg H, Hederstrom E. Detection of hepatic metastases in colorectal cancer: a prospective study of laboratory and imaging methods. Eur J Surg 1993; 159:275-281.[Medline]
  7. Moertel CG, Fleming TR, Macdonald JS, Haller DG, Laurie JA, Tangen C. An evaluation of the carcinoembryonic antigen (CEA) test for monitoring patients with resected colon cancer. JAMA 1993; 270:943-947.[Abstract]
  8. Huebner RH, Park KC, Shepherd JE, et al. A meta-analysis of the literature for whole-body FDG PET detection of recurrent colorectal cancer. J Nucl Med 2000; 41:1177-1189.[Abstract/Free Full Text]
  9. Leen E, Angerson WJ, Wotherspoon H, Moule B, Cook TG, McArdle CS. Detection of colorectal liver metastases: comparison of laparotomy, CT, US, and Doppler perfusion index and evaluation of postoperative follow-up results. Radiology 1995; 195:113-116.[Abstract/Free Full Text]
  10. Carter R, Hemingway D, Cooke TG, et al. A prospective study of six methods for detection of hepatic colorectal metastases. Ann R Coll Surg Engl 1996; 78:27-30.[Medline]
  11. Zerhouni EA, Rutter C, Hamilton SR, et al. CT and MR imaging in the staging of colorectal carcinoma: report of the Radiology Diagnostic Oncology Group II. Radiology 1996; 200:443-451.[Abstract/Free Full Text]
  12. Strotzer M, Gmeinwieser J, Schmidt J, et al. Diagnosis of liver metastases from colorectal adenocarcinoma: comparison of spiral-CTAP combined with intravenous contrast-enhanced spiral-CT and SPIO-enhanced MR combined with plain MR imaging. Acta Radiol 1997; 38:986-992.[Medline]
  13. Abdel-Nabi H, Doerr RJ, Lamonica DM, et al. Staging of primary colorectal carcinomas with fluorine-18 fluorodeoxyglucose whole-body PET: correlation with histopathologic and CT findings. Radiology 1998; 206:755-760.[Abstract/Free Full Text]
  14. Whiteford MH, Whiteford HM, Yee LF, et al. Usefulness of FDG-PET scan in the assessment of suspected metastatic or recurrent adenocarcinoma of the colon and rectum. Dis Colon Rectum 2000; 43:759-767; discussion 767–770.[CrossRef][Medline]
  15. Irwig L, Tosteson AN, Gatsonis C, et al. Guidelines for meta-analyses evaluating diagnostic tests. Ann Intern Med 1994; 120:667-676.[Abstract/Free Full Text]
  16. McCullaph P, Nelder JA. Generalized linear models In: Cox DR, Hinkley DV, eds. Monographs on statistics and applied probability. London, England: Chapman & Hall, 1983.
  17. Cook DJ, Sackett DL, Spitzer WO. Methodologic guidelines for systematic reviews of randomized control trials in health care from the Potsdam Consultation on Meta-Analysis. J Clin Epidemiol 1995; 48:167-171.[CrossRef][Medline]
  18. Rockette HE, Gur D, Campbell WL, Thaete FL. Use of meta-analysis in the evaluation of imaging systems. Acad Radiol 1994; 1:63-69.[CrossRef][Medline]
  19. Irwig L, Macaskill P, Glasziou P, Fahey M. Meta-analytic methods for diagnostic test accuracy. J Clin Epidemiol 1995; 48:119-130; discussion 131–132.[CrossRef][Medline]
  20. Diggle PJ, Liang KY, Zeger SL. Analysis of longitudinal data In: Atkinson AC, Copas JB, Pierce DA, Schervish MJ, Titterington DM, eds. Statistical science series. Oxford, England: Clarendon, 1994.
  21. Hochber Y, Tamhane A. Multiple comparison procedures New York, NY: Wiley, 1987.
  22. Tempero MA, Williams CA, Anderson JC. The value of hepatic ultrasound and biochemical liver tests in screening for liver metastases. J Clin Oncol 1986; 4:1074-1078.[Abstract/Free Full Text]
  23. Grace RH, Hale M, Mackie G, Marks CG, Bloomberg TJ, Walker WJ. Role of ultrasound in the diagnosis of liver metastases before surgery for large bowel cancer. Br J Surg 1987; 74:480-481.[Medline]
  24. Watt I, Stewart I, Anderson D, Bell G, Anderson JR. Laparoscopy, ultrasound and computed tomography in cancer of the oesophagus and gastric cardia: a prospective comparison for detecting intra-abdominal metastases. Br J Surg 1989; 76:1036-1039.[Medline]
  25. Soyer P, Levesque M, Elias D, Zeitoun G, Roche A. Preoperative assessment of resectability of hepatic metastases from colonic carcinoma: CT portography vs sonography and dynamic CT. AJR Am J Roentgenol 1992; 159:741-744.[Abstract/Free Full Text]
  26. Stell DA, Carter CR, Stewart I, Anderson JR. Prospective comparison of laparoscopy, ultrasonography and computed tomography in the staging of gastric cancer. Br J Surg 1996; 83:1260-1262.[CrossRef][Medline]
  27. Freeny PC, Marks WM, Ryan JA, Bolen JW. Colorectal carcinoma evaluation with CT: preoperative staging and detection of postoperative recurrence. Radiology 1986; 158:347-353.[Abstract/Free Full Text]
  28. Balthazar EJ, Megibow AJ, Hulnick D, Naidich DP. Carcinoma of the colon: detection and preoperative staging by CT. AJR Am J Roentgenol 1988; 150:301-306.[Abstract/Free Full Text]
  29. Freitag J, Rosenkranz M, Freitag G, Bathge M, Willgeroth C. Computed tomographic diagnosis of rectal tumors. Radiol Diagn (Berl) 1989; 30:257-262[German].
  30. Chapuis P, Kos S, Bokey L, Dent O, Newland R, Hinder J. How useful is pre-operative computerized tomography scanning in staging rectal cancer? Aust N Z J Surg 1989; 59:31-34.[Medline]
  31. Acunas B, Rozanes I, Acunas G, Celik L, Sayi I, Gokmen E. Preoperative CT staging of colon carcinoma (excluding the recto-sigmoid region). Eur J Radiol 1990; 11:150-153.[CrossRef][Medline]
  32. Cance WG, Cohen AM, Enker WE, Sigurdson ER. Predictive value of a negative computed tomographic scan in 100 patients with rectal carcinoma. Dis Colon Rectum 1991; 34:748-751.[CrossRef][Medline]
  33. Winzelberg GG, Grossman SJ, Rizk S, et al. Indium-111 monoclonal antibody B72.3 scintigraphy in colorectal cancer: correlation with computed tomography, surgery, histopathology, immunohistology, and human immune response. Cancer 1992; 69:1656-1663.[CrossRef][Medline]
  34. Maerz LL, Deveney CW, Lopez RR, McConnell DB. Role of computed tomographic scans in the staging of esophageal and proximal gastric malignancies. Am J Surg 1993; 165:558-560.[CrossRef][Medline]
  35. Adachi Y, Sakino I, Matsumata T, et al. Preoperative assessment of advanced gastric carcinoma using computed tomography. Am J Gastroenterol 1997; 92:872-875.[Medline]
  36. Davies J, Chalmers AG, Sue-Ling HM, et al. Spiral computed tomography and operative staging of gastric carcinoma: a comparison with histopathological staging. Gut 1997; 41:314-319.[Abstract/Free Full Text]
  37. Delbeke D, Vitola JV, Sandler MP, et al. Staging recurrent metastatic colorectal carcinoma with PET. J Nucl Med 1997; 38:1196-1201.[Abstract/Free Full Text]
  38. Valls C, Lopez E, Guma A, et al. Helical CT versus CT arterial portography in the detection of hepatic metastasis of colorectal carcinoma. AJR Am J Roentgenol 1998; 170:1341-1347.[Abstract/Free Full Text]
  39. Ward J, Naik KS, Guthrie JA, Wilson D, Robinson PJ. Hepatic lesion detection: comparison of MR imaging after the administration of superparamagnetic iron oxide with dual-phase CT by using alternative-free response receiver operating characteristic analysis. Radiology 1999; 210:459-466.[Abstract/Free Full Text]
  40. Soyer P, Levesque M, Caudron C, Elias D, Zeitoun G, Roche A. MRI of liver metastases from colorectal cancer vs. CT during arterial portography. J Comput Assist Tomogr 1993; 17:67-74.
  41. Peterson MS, Baron RL, Murakami T. Hepatic malignancies: usefulness of acquisition of multiple arterial and portal venous phase images at dynamic gadolinium-enhanced MR imaging. Radiology 1996; 201:337-345.[Abstract/Free Full Text]
  42. Said B, McCart JA, Libutti SK, Choyke PL. Ferumoxide-enhanced MRI in patients with colorectal cancer and rising CEA: surgical correlation in early recurrence. Magn Reson Imaging 2000; 18:305-309.[CrossRef][Medline]
  43. Boykin KN, Zibari GB, Lilien DL, McMillan RW, Aultman DF, McDonald JC. The use of FDG-positron emission tomography for the evaluation of colorectal metastases of the liver. Am Surg 1999; 65:1183-1185.[Medline]
  44. Imdahl A, Reinhardt MJ, Nitzsche EU, et al. Impact of 18F-FDG-positron emission tomography for decision making in colorectal cancer recurrences. Langenbecks Arch Surg 2000; 385:129-134.[CrossRef][Medline]
  45. Franke J, Rosenzweig S, Reinartz P, Hoer J, Kasperk R, Schumpelick V. Value of positron emission tomography (18F-FDG-PET) in the diagnosis of recurrent rectal cancer. Chirurg 2000; 71:80-85[German].[Medline]
  46. Valk PE, Abella-Columna E, Haseman MK, et al. Whole-body PET imaging with [18F]fluorodeoxyglucose in management of recurrent colorectal cancer. Arch Surg 1999; 134:503-511; discussion 511–513.[Abstract/Free Full Text]
  47. Lind SE, Singer DE. Diagnosing liver metastases: a Bayesian analysis. J Clin Oncol 1986; 4:379-388.[Abstract/Free Full Text]
  48. Morgan-Parkes JH. Metastases: mechanisms, pathways, and cascades. AJR Am J Roentgenol 1995; 164:1075-1082.[Abstract/Free Full Text]
  49. Huguier M, Lacaine F. Hepatic metastases in gastrointestinal cancer: diagnostic value of biochemical investigations. Arch Surg 1981; 116:399-401.[Abstract]
  50. Oxley EM, Ellis H. Prognosis of carcinoma of the large bowel in the presence of liver metastases. Br J Surg 1969; 56:149-152.[Medline]
  51. Park KC, Schwimmer J, Shepherd JE, et al. Decision analysis for the cost-effective management of recurrent colorectal cancer. Ann Surg 2001; 233:310-319.[CrossRef][Medline]
  52. Maldonado A, Sancho F, Cerdan J, et al. FDG-PET in the detection of recurrence in colorectal cancer based on rising CEA Level: experience in 72 patients. Clin Positron Imaging 2000; 3:170.[CrossRef][Medline]
  53. Staib L, Schirrmeister H, Reske SN, Beger HG. Is (18)F-fluorodeoxyglucose positron emission tomography in recurrent colorectal cancer a contribution to surgical decision making? Am J Surg 2000; 180:1-5.[CrossRef][Medline]
  54. Flamen P, Hoekstra OS, Homans F, et al. Unexplained rising carcinoembryonic antigen (CEA) in the postoperative surveillance of colorectal cancer: the utility of positron emission tomography (PET). Eur J Cancer 2001; 37:862-869.
  55. Strasberg SM, Dehdashti F, Siegel BA, Drebin JA, Linehan D. Survival of patients evaluated by FDG-PET before hepatic resection for metastatic colorectal carcinoma: a prospective database study. Ann Surg 2001; 233:293-299.[CrossRef][Medline]
  56. Taylor M, Forster J, Langer B, Taylor BR, Greig PD, Mahut C. A study of prognostic factors for hepatic resection for colorectal metastases. Am J Surg 1997; 173:467- 471.[CrossRef][Medline]
  57. Thoeni RF. Colorectal cancer: radiologic staging. Radiol Clin North Am 1997; 35:457-485.[Medline]
  58. Lee MG, Baker ME, Sostman HD, et al. The diagnostic accuracy/efficacy of MRI in differentiating hepatic hemangiomas from metastatic colorectal/breast carcinoma: a multiple reader ROC analysis using a jackknife technique. J Comput Assist Tomogr 1996; 20:905-913.[CrossRef][Medline]
  59. Hamm B, Vogl TJ, Branding G, et al. Focal liver lesions: MR imaging with Mn-DPDP—initial clinical results in 40 patients. Radiology 1992; 182:167-174.[Abstract/Free Full Text]
  60. Ransohoff DF, Feinstein AR. Problems of spectrum and bias in evaluating the efficacy of diagnostic tests. N Engl J Med 1978; 299:926-930.[Abstract]
  61. Kondo H, Kanematsu M, Hoshi H, et al. Preoperative detection of malignant hepatic tumors: comparison of combined methods of MR imaging with combined methods of CT. AJR Am J Roentgenol 2000; 174:947-954.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Ann OncolHome page
M. de Wit, W. Brenner, M. Hartmann, J. Kotzerke, D. Hellwig, J. Lehmann, C. Franzius, S. Kliesch, M. Schlemmer, K. Tatsch, et al.
[18F]-FDG-PET in clinical stage I/II non-seminomatous germ cell tumours: results of the German multicentre trial
Ann. Onc., May 16, 2008; (2008) mdn170v2.
[Abstract] [Full Text] [PDF]

Home page
 Postgrad. Med. J.Home page
Y-x He and Q-y Guo
Clinical applications and advances of positron emission tomography with fluorine-18-fluorodeoxyglucose (18F-FDG) in the diagnosis of liver neoplasms
Postgrad. Med. J., May 1, 2008; 84(991): 246 - 251.
[Abstract] [Full Text] [PDF]

Home page
 JNMHome page
J. W. Fletcher, B. Djulbegovic, H. P. Soares, B. A. Siegel, V. J. Lowe, G. H. Lyman, R. E. Coleman, R. Wahl, J. C. Paschold, N. Avril, et al.
Recommendations on the Use of 18F-FDG PET in Oncology
J. Nucl. Med., March 1, 2008; 49(3): 480 - 508.
[Abstract] [Full Text] [PDF]

Home page
 Ann OncolHome page
L.F. de Geus-Oei, H. W. M. van Laarhoven, E. P. Visser, R. Hermsen, B. A. van Hoorn, Y. J. L. Kamm, P. F. M. Krabbe, F. H. M. Corstens, C. J. A. Punt, and W. J. G. Oyen
Chemotherapy response evaluation with FDG-PET in patients with colorectal cancer
Ann. Onc., February 1, 2008; 19(2): 348 - 352.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
T. R. O'Rourke, P. Tekkis, S. Yeung, J. Fawcett, S. Lynch, R. Strong, D. Wall, T. G. John, F. Welsh, and M. Rees
Long-Term Results of Liver Resection for Noncolorectal, Nonneuroendocrine Metastases
Ann. Surg. Oncol., January 1, 2008; 15(1): 207 - 218.
[Abstract] [Full Text] [PDF]

Home page
 JNMHome page
H. Guo, H. Zhu, Y. Xi, B. Zhang, L. Li, Y. Huang, J. Zhang, Z. Fu, G. Yang, S. Yuan, et al.
Diagnostic and Prognostic Value of 18F-FDG PET/CT for Patients with Suspected Recurrence from Squamous Cell Carcinoma of the Esophagus
J. Nucl. Med., August 1, 2007; 48(8): 1251 - 1258.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
B. Wiering, T. J. M. Ruers, P. F. M. Krabbe, H. M. Dekker, and W. J. G. Oyen
Comparison of Multiphase CT, FDG-PET and Intra-Operative Ultrasound in Patients with Colorectal Liver Metastases Selected for Surgery
Ann. Surg. Oncol., February 1, 2007; 14(2): 818 - 826.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
C. Charnsangavej, B. Clary, Y. Fong, A. Grothey, T. M. Pawlik, and M. A. Choti
Selection of Patients for Resection of Hepatic Colorectal Metastases: Expert Consensus Statement
Ann. Surg. Oncol., October 1, 2006; 13(10): 1261 - 1268.
[Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
W. C. Chapman, P. M. Hoff, and S. M. Strasberg
Selection of Patients for Resection of Hepatic Colorectal Metastases: Expert Consensus Statement by Charnsangavej et al.
Ann. Surg. Oncol., October 1, 2006; 13(10): 1269 - 1270.
[Full Text] [PDF]

Home page
 Ann. Thorac. Surg.Home page
N. Rizk, R. J. Downey, T. Akhurst, M. Gonen, M. S. Bains, S. Larson, and V. Rusch
Preoperative (18)[F]-Fluorodeoxyglucose Positron Emission Tomography Standardized Uptake Values Predict Survival After Esophageal Adenocarcinoma Resection.
Ann. Thorac. Surg., March 1, 2006; 81(3): 1076 - 1081.
[Abstract] [Full Text] [PDF]

Home page
 RadioGraphicsHome page
J. S. Lim, M. J. Yun, M.-J. Kim, W. J. Hyung, M.-S. Park, J.-Y. Choi, T.-S. Kim, J. D. Lee, S. H. Noh, and K. W. Kim
CT and PET in Stomach Cancer: Preoperative Staging and Monitoring of Response to Therapy
RadioGraphics, January 1, 2006; 26(1): 143 - 156.
[Abstract] [Full Text] [PDF]