The impact of fluor-18-deoxyglucose-positron emission tomography in the management of colorectal liver metastases

A Systematic Review and Metaanalysis

Authors


Abstract

Fluor-18-deoxyglucose-positron emission tomography (FDG-PET) has emerged as a promising diagnostic modality in recurrent colorectal carcinoma. Whole-body FDG-PET may be an accurate diagnostic modality to determine whether patients with recurrent hepatic disease are suitable candidates for curative resection. Reports on the use of FDG-PET in patients with recurrent colorectal carcinoma are scarce, especially those on colorectal liver metastases. To assess the usefulness of this emerging modality for the selection of patients to undergo resection for colorectal liver metastases, a systematic (meta)-analysis of the current literature was conducted. In the absence of randomized controlled clinical trials, a traditional meta-analysis could not be performed. An alternative strategy was designed to evaluate the current literature. After a literature search, an index score was devised to evaluate the articles with regard to the impact of FDG-PET in patients with colorectal liver metastases. The index scored articles on several items and, as such, could be considered an objective approach for the assessment of diagnostic, nonrandomized clinical trials. The proposed index proved to be an independent instrument for judging several research questions and was used systematically to address the sensitivity, specificity, and clinical impact of FDG-PET in patients with colorectal liver metastases. For FDG-PET, the pooled sensitivity and specificity results were 88.0% and 96.1%, respectively, for hepatic disease and 91.5% and 95.4%, respectively, for extrahepatic disease. For the 6 articles that reported the highest scores on the index, the sensitivity and specificity of FDG-PET for hepatic metastatic disease were 79.9% and 92.3%, respectively, and 91.2% and 98.4%, respectively, for extrahepatic disease, respectively. For computed tomography, the pooled sensitivity and specificity results were 82.7% and 84.1%, respectively, for hepatic lesions and 60.9% and 91.1%, respectively, for extrahepatic lesions. The percentage change in clinical management due to FDG-PET was 31.6% (range, 20.0–58.0%) in the articles that scored above the mean and reported this item. For the 6 highest scoring studies, the percentage change in clinical management was 25.0% (range, 20.0–32.0%). Despite apparent omissions in the literature, the combined sensitivity and specificity of FDG-PET clearly indicated that FDG-PET has added value in the diagnostic workup of patients with colorectal liver metastases. FDG-PET can be considered a useful tool in preoperative staging and produced superior results compared with conventional diagnostic modalities, especially for excluding or detecting extrahepatic disease. Cancer 2005. © 2005 American Cancer Society.

Colorectal carcinoma ranks second as the cause of death due to cancer in the Western world, with a 5-year survival rate of 55%. Liver metastases are the main cause of death in this patient group. Approximately 20% of patients already have hepatic metastases at the time the primary tumor is detected, and another 25% will develop metastatic lesions in the years after they undergo resection of their primary tumor. For selected patients who have recurrent disease that is confined to the liver, surgical resection of the metastases is the treatment of choice, with a 5-year survival rate of > 40% reported in the recent literature.1–3

The selection of patients for surgical resection of colorectal liver metastases, however, still poses a significant clinical problem.4, 5 First, significant numbers of patients (15–25%) who are scheduled for surgical resection have unresectable disease identified at the time of laparotomy. Second, after they undergo liver resection, 60% of patients will develop recurrent tumor within 3 years either in the liver or in extrahepatic areas. These findings strongly support the need for more effective preoperative imaging to improve staging and to avoid futile surgery.

Positron emission tomography (PET) using fluor-18-deoxyglucose (FDG) has emerged as a promising diagnostic modality in recurrent colorectal carcinoma. Whole-body FDG-PET may prove to be an accurate diagnostic technique for determining whether patients with recurrent disease in the liver are suitable candidates for curative resection. In contrast, with conventional diagnostic modalities, such as computed tomography (CT) scanning and magnetic resonance imaging, FDG-PET does not require anatomic alterations for the detection of malignancy but provides images based on the increased glucose uptake and the altered metabolism of the tumor.

Reports on the use of FDG-PET in patients with recurrent colorectal carcinoma are scarce, especially from patients with colorectal liver metastases. Furthermore, randomized, controlled clinical trials are not available. To assess the usefulness of this emerging modality for the selection of patients undergoing hepatic resection for colorectal metastases, a systematic (meta)-analysis of the current literature was conducted. Data on liver metastases were extracted from all reports of FDG-PET in patients with colorectal carcinoma and were analyzed with a systematic approach.

MATERIALS AND METHODS

Search Strategy

A systematic literature search for articles concerning recurrent colorectal liver metastases and PET imaging was performed using the Medline data base and EMBASE up to January 2004. Through the Medline and EMBASE data bases, 5 search strategies with the following Medical Subject Heading items were used: 1) diagnostic imaging (FDG, fluorodeoxygly-, fluorodeoxyglucose-F-18-, tomography-X-ray-, tomography emission computed, magnetic resonance imaging, and all subheadings thereof), 2) PET-scan (positr- near emissio-, positron-emiss-, PET, and all subheadings), 3) metastases (metasta-, recur-, reappear-, recidiv-, relaps- or recrudes- near cancer-, carcinom-, malign-, tumor-, tumour- or neoplasm-, “neoplasm-metastasis,” “neoplasm-recurrence-local,” “recurrence-,” “bile-duct-obstruction-extrahepatic,” and “bile-ducts-extrahepatic” with all subheadings), 4) liver metastases (Search Strategy 3 near liver- or hepat- or extrahepat- or bile- or bilia-), and 5) matching strategies (see Table 1). In total, 363 articles were identified in this first search. A second search was conducted using the up-dated strategy published by Mijnhout et al.6, 7 The results of these search strategies were combined.

Table 1. Search Strategies
  1. MIME: minor Medical Subject Headings (MeSH) headings; MJME: major MeSH headings; PET: positron emission tomography; FDG: Fluor-18-deoxyglucose.

Search Strategy I
 1. Explode “tomography-X-ray-+”/all subheadings in MIME, MJME; or explode “tomography-emission-computed-+”/all subheadings in MIME, MJME; or explode “magnetic-resonance-imaging-+”/all subheadings in MIME, MJME (414,785 records)
 2. “Fludeoxyglucose-F-18”/all subheadings in MIME, MJME (8176 records)
 3. Explode “bile-duct-obstruction-extrahepatic”/all subheadings in MIME, MJME; or explode “bile-ducts-extrahepatic”/all subheadings in MIME, MJME (19,559 records)
 4. Explode “neoplasm-recurrence-local”/all subheadings in MIME, MJME; or explode “recurrence-”/all subheadings in MIME, MJME (205,571 records)
 5. Explode “neoplasm-metastasis”/all subheadings in MIME, MJME (123,300 records)
 6. Fluorodeoxygl* (3305 records)
 7. (Metasta* or recurr* or reappear* or recidiv* or relaps* or recrudes*) near (cancer* or carcinom* or malign* or tumor* or tumour* or neoplas*) (276,686 records)
 8. Strategy I-7, or I-5, or I-4, or I-3 (461,092 records)
 9. Strategy I-6, or I-2, or I-1 (415,891 records)
 10. PET* (259,611 records)
 11. Positron-emiss* (747 records)
 12. Positr* near emissio* (21,386 records)
 13. Strategy I-12, or I-11, or I-10 (26,7521 records)
 14. Strategy I-8 near (liver* or hepat* or extrahepat* or bile* or bilia*) (33,772 records)
 15. Strategies I-9 and I-13 (32,398 records)
 16. Strategies I-14 and I-15 (244 records)
 17. FDG (6593 records)
 18. Strategy I-17, or I-6, or I-2, or I-1 (416,353 records)
 19. Strategies I-18 and I-13 (32,732 records)
 20. Strategies I-14 and I-19 (253 records)
Search Strategy II
 1. Explode “tomography-X-ray-+”/all subheadings in MIME, MJME; or explode “tomography-emission-computed-+”/all subheadings in MIME, MJME; or explode “magnetic-resonance-imaging-+”/all subheadings in MIME, MJME (248,884 records)
 2. “Fludeoxyglucose-F-18”/all subheadings in MIME, MJME (4525 records)
 3. Explode “bile-duct-obstruction-extrahepatic”/all subheadings in MIME, MJME; or explode “bile-ducts-extrahepatic”/all subheadings in MIME, MJME (14,860 records)
 4. Explode “neoplasm-recurrence-local”/all subheadings in MIME, MJME; or explode “recurrence-”/all subheadings in MIME, MJME (135,546 records)
 5. Explode “neoplasm-metastasis”/all subheadings in MIME, MJME (90,666 records)
 6. Fluorodeoxygl* (1792 records)
 7. (Metasta* or recurr* or reappear* or recidiv* or relaps* or recrudes*) near (cancer* or carcinom* or malign* or tumor* or tumour* or neoplas*) (182,550 records)
 8. Strategy II-7, or II-5, or II-4, or II-3 (307,154 records)
 9. Strategy II-6, or II-2, or II-1 (249,509 records)
 10. PET* (167,845 records)
 11. Positron-emiss* (444 records)
 12. Positr* near emissio* (12,063 records)
 13. Strategy II-12, or II-11, or II-10 (172,504 records)
 14. Strategy II-8 near (liver* or hepat* or extrahepat* or bile* or bilia*) (22,625 records)
 15. Strategies II-9 and II-13 (18,421 records)
 16. Strategies II-14 and II-15 (134 records)

Articles were included in this review only when they included either a description of the impact on clinical management by FDG-PET or a description of FDG-PET imaging results in patients with recurrent colorectal carcinoma. Systematic review articles that mentioned PET alongside CT or other diagnostic modalities were excluded, because the individual articles were included in our review.8–12

Analysis of the Literature

Because randomized clinical trails are missing, data with which to perform a traditional, systematic meta-analysis could not be obtained. To compare, weight, and summarize data from the different studies, an alternative approach was designed using the stepwise process described below (Table 2).

Table 2. List of Domains and Items
DomainClassification (quantity)aWeight factorb
  • CRC: colorectal carcinoma; pTNM: pathologic tumor, lymph node, metastasis classification; FDG-PET: fluor-18-deoxyglucose-positron emission tomography; CT: computed tomography; MRI: magnetic resonance imaging; SUV: standardized uptake value.

  • a

    Classification: 0.0, not mentioned; 0.5, partly mentioned; 1.0, adequately described.

  • b

    Weight factor: 1, not significant; 2, mildly significant; 3, average; 4, significant; 5, very significant.

1. Patient population  
 1.1 Patient characteristics 3
  Clear description of population's age and gender0/1 
 1.2 Localization and stage primary CRC 2
  Localization: colon vs. rectum or colon segment mentioned AND Stage: I–IV or pTNM1.0 
  One of the above0.5 
  None of the above0.0 
 1.3 Comorbidity/diabetes 4
  Diabetes is interfering with the correct assessment of the FDG-PET images0/1 
 1.4 Sample size; no. of livers/total patients 5
  > 35 patients with hepatic metastases1.0 
  < 35 patients with hepatic metastases0.5 
  Not mentioned0.0 
 1.5 Management decision after normal diagnostic workup (see 2.6, below) 4
  After normal workup, decision to operate was taken0/1 
 1.6 Excluded patients 4
  No. of excluded patients and how they were considered in the analysis 0/1
2. Study design  
 2.1 Research question 4
  A clear and decisive question should be stated of the study objective0/1 
 2.2 Inclusion criteria 4
  An exact summation of all inclusion and exclusion criteria1.0 
  The patient group characteristics in general0.5 
  None given0.0 
 2.3 Design 5
  Prospective1.0 
  Retrospective0.5 
  Not described0.0 
 2.4 Randomization 5
  Study design involved a randomization of PET used in diagnostic workup0/1 
 2.5 Study period 3
  Described beginning and ending of study0/1 
 2.6 Blinding 4
  Blinding PET results before above-mentioned decision (see 1.5)0/1 
3. Statistics  
 3.1 Descriptive statistics 2
  Clear description of statistics and rationale used0/1 
 3.2 Sensitivity/specificity of PET (OR false-positive/positive predictive value) liver 5
  Results of PET of hepatic metastases0/1 
 3.3 Sensitivity/specificity of CT/MRI (OR false-positive/positive predictive value) liver 5
  Results of CT or MRI of hepatic metastases0/1 
 3.4 Sensitivity/specificity of PET (OR false-positive/positive predictive value) extrahepatic 5
  Results of PET of extrahepatic metastases 0/1
 3.5 Sensitivity/specificity of CT/MRI (OR false-positive/positive predictive value) extrahepatic 5
  Results of CT or MRI of extrahepatic metastases0/1 
 3.6 Confidence interval 3
  Method used to describe 95% or 90% margin error0/1 
4. Preoperative technologies and image interpretation  
 4.1 Preoperative follow-up 1
  A clear outline that states follow-up of primary tumor, with set time intervals0/1 
 4.2 Surgical workup 5
  CT-chest, abdomen/MRI colonoscopy1.0 
  Chest X-ray, ultrasound0.5 
  Not given0.0 
 4.3 Interval CT and PET 4
  < 4 wks1.0 
  < 8 wks0.5 
  Not given0.0 
 4.4 Interval diagnostics and surgery 4
  <4 wks1.0 
  < 8 wks0.5 
  Not given0.0 
 4.5 Technical specifications and patient preparation PET 4
  Both1.0 
  One of two0.5 
  None0.0 
 4.6 Technical specifications and patient preparation CT 4
  Both1.0 
  One of two0.5 
  None0.0 
 4.7 Definition of positive and negative PET findings 4
  Scaling of PET finding1.0 
  Descriptive or SUV measurement0.5 
  Not given0.0 
 4.8 No. of reviewers (PET) 3
  Two or more1.0 
  One0.5 
  Not given0.0 
 4.9 Disease distribution 4
  Clear description of organs involved in metastatic tumor0/1 
 4.10 Lesions studied 3
  Lesion-by-lesion study and description of attribution to different outcomes0/1 
5. Final confirmation  
 5.1 Preoperative ultrasound 3
  Preoperative ultrasound to verify the findings in preoperative diagnostic workup0/1 
 5.2 Confirmation by histopathology or clinical follow-up 5
  Clinical follow-up > 12 mos1.0 
  Clinical follow-up > 6 mos and < less than 12 mos0.5 
  Not given and no histopathology0.0 
 5.3 Postoperative follow-up I 4
  Frequency of standard follow-up with intervals of 3 mos1.0 
  Frequency of standard follow-up with intervals of 6 mos0.5 
  Not given0.0 
 5.4 Postoperative follow-up II 4
  Diagnostic means of follow-up; CT-chest, abdomen/MRI colonoscopy1.0 
  Diagnostic means of follow-up; chest X-ray, ultrasound0.5 
  Not given0.0 
 5.5 Lost to follow-up 5
  No. of patients and consideration of attribution to outcome  
  Both1.0 
  One of two0.5 
  None0.0 
 5.6 Change in management 5
  Clinical implication of PET used in this setting0.1 

First, a panel of experts (a hepatic surgeon, a nuclear medicine physician, a methodologist, and a radiologist) constructed a concept study containing all items that should be included and weighted in an ideal study design. Table 2 shows that this team of experts selected 5 different domains, each containing several items. There was consensus that these domains and items ideally should be reported in the individual studies. Second, every item in a domain was weighted (5, very significant; 4, significant; 3, average; 2, mildly significant; 1, not significant) by each individual member of the expert team, after which, a consensus was achieved on the final weight factor of each item (Table 2).

In a third step, all articles were screened for the presence of the selected items. When an item was not available, no points were awarded. In articles in which an item was covered partially, 0.5 points were assigned. When the item was present and reported adequately, 1 point was awarded. The awarded points for each item were multiplied by the weight factor (1–5) to achieve the final value per item, and the total number of points per domain was calculated. Each of the 5 domains contributed 20% to the total score. In Figure 1, the scores for each domain and the total score for each article are presented as standard deviations from the mean.

Figure 1.

The scores for each domain and the total scores for each study are presented as standard deviations from the mean.

The method described above can be considered a conceivable approach for the assessment of diagnostic, nonrandomized clinical trials to determine the impact of FDG-PET in oncology, in line with the Standards for Reporting Studies of Diagnostic Accuracy (STARD) statement13, 14 for diagnostic trails, the Consolidated Standards of Reporting Trials (CONSORT)15 statement for randomized trials, and the Methodological Index for Non-Randomized Studies16 for nonrandomized surgical clinical trials. We used the STARD initiative as a starting point and adopted it specifically to suit our objective of addressing the impact of FDG-PET in patients with potentially resectable colorectal liver metastases. Like with any new trial design that has no definitive gold standard, it is feasible that the proposed list has internal flaws. To validate the approach used, the results from the systematic assessment were compared with the results from a search for the best articles by two surgeons who were not aware of the items on the list.

The objective of this proposed approach was to conceive a method by which most critical research questions on FDG-PET and hepatic colorectal metastases could be addressed in a systematic and semiquantitative manner. Using this approach, the following questions were addressed: 1) How do the descriptive statistics (sensitivity and specificity) for FDG-PET compare with those for CT in the assessment of both hepatic and extrahepatic metastatic involvement? 2) Does FDG-PET have a significant impact on changes in the management of these types of metastases? 3) Which domain of research is lacking in the literature?

Not all studies were designed specifically for the analysis of hepatic recurrences of colorectal carcinoma. Therefore, a low total score does not automatically indicate that a study was poor, because the score merely recognizes the study's contribution to the evaluation of FDG-PET in colorectal liver metastases.

Statistical Analysis

All statistical analyses were performed using the SPSS version 11.0 software package (SPSS Inc., Chicago, IL). To demonstrate a correlation between different aspects of the study and the validity of the proposed domain list, a Pearson correlation test was used. Furthermore, pooled sensitivity and specificity results were calculated from the true-negative results, true-positive results, false-negative results, and false-positive results that were reported in the different articles. Pooled sensitivity and specificity were determined after correcting for the variable number of patients per study.

RESULTS

The search strategies identified 32 studies, which were included in this systematic analysis.17–49 An overview describing the content of each article can be found in Table 3. All selected articles were scored according to the weighting procedure, as shown in Table 2. By using this approach, it was possible to determine the weight for each domain and the total weight of for different articles. Although large variations were observed between the different articles, there was remarkable consistency in the domain scores and in the final total scores for the both lower scoring and higher scoring articles. At least six articles consistently showed high scores for all of the domains, whereas six other studies consistently scored low for the different domains. Both the results of the weighted evaluation and the total scores are summarized in Figure 1, which presents the score as a deviation from the mean for each article.

Table 3. Study Scores
StudySamplingaDesign studyScore
  • a

    Sampling: the number of patients with liver metastases divided by the total patients in the study.

Beets et al., 19941815/35Prospective5.9
Schiepers et al., 19953973/763.3
Vitola et al., 19964524/24Prospective4.5
Lai et al., 19963234/34Prospective10.3
Ogunbiyi et al., 19973523/58Retrospective2.3
Delbecke et al., 19972052/525.7
Flanagan et al., 19982522/22Retrospective− 5.1
Delbecke et al., 19982153/110Prospective0.0
Yasuda et al., 1998478/8Retrospective− 17.2
Flamen et al., 19992345/103Retrospective− 1.7
Fong et al., 19992640/40Prospective11.0
Valk et al., 19994457/115Prospective28.7
Boykin et al., 19991914/14Retrospective− 25.2
Imdahl et al., 20002928/71Prospective14.1
Whiteford et al., 20004635/105Retrospective− 3.8
Zhuang et al., 20004828/28Retrospective− 20.1
Staib et al., 2000413/100Prospective9.7
Imbriaco et al., 200028Not mentionedRetrospective− 11.0
Iwata et al., 2000308/78− 21.7
Strasberg et al., 20014943/43Prospective− 7.9
Flamen et al., 20012415/50Retrospective4.4
Topal et al., 20014391/91Prospective2.2
Hung et al., 2001272/32Retrospective− 5.5
Arulampalam et al., 20011715/42Prospective3.0
Johnson et al., 20013126/41Retrospective− 1.9
Ruers et al., 20023751/51Prospective20.1
Lonneux et al., 20023433/68Retrospective
Rydzewski et al., 20023847/47Retrospective− 6.7
Rohren et al., 20023623/23Retrospective− 3.1
Langenhoff et al., 20023323/23Prospective13.7
Simo et al., 20024031/145Retrospective− 9.7
Desai et al., 20032225/114Prospective1.7

The proposed list appeared to be valid, because 5 of the 6 articles also were selected by the surgeons who were blinded to the items identified on the list (κ = 0.839). The 6 articles that had scores > 10 points above the mean and, therefore, are regarded as the best studies that evaluated the research questions addressed in this systematic review.

Descriptive Statistics

To use the designed domain list as a separate instrument for analysis, the list should be independent from the main endpoints: the sensitivity and specificity of CT scanning and PET imaging and the change in clinical management that is attributable to FDG-PET. For this reason, the total score of the domain list was compared with the presence of sensitivity and specificity data in each article from a Pearson correlation test. The correlations of the total score with sensitivity and specificity for liver metastases of FDG-PET and CT were 0.21 and 0.05, respectively. For extrahepatic disease, these values were 0.13 for FDG-PET and 0.01 for CT. This indicates that the sensitivity and specificity are not correlated with the total scores of the domain list, underlining the fact that the proposed list is not dependent on the outcomes of sensitivity and specificity, thus securing the domain list as an independent measuring tool.

Change in clinical management was another endpoint of interest, because it shows the clinical implications of FDG-PET imaging in patients with liver metastases. No significant correlation (r) was found between the total score and the reported change in management (r = − 0.20), supporting the use of the proposed domain list as an independent tool for assessing the different studies.

Sensitivity and Specificity

Lesion-based sensitivity and specificity were calculated for the different diagnostic modalities as derived from the different studies. The results presented here may differ from the sensitivities and specificities provided in the articles because of the lesion-based nature of the current analysis. Figure 2 shows the sensitivity and specificity according to each individual article. The pooled overall scores (Fig. 2, “Combined”) were derived directly from the combined true-negative, true-positive, false-negative, and false-positive results from all studies. If FDG-PET was inconclusive in the lesion-based analysis, then the results were considered false positive. For hepatic lesions, the pooled sensitivity and specificity results from FDG-PET were 88.0% and 96.1%, respectively, with 95% confidence intervals (95% CIs) of 88.–98.0 and 70.4–104.3, respectively. For extrahepatic lesions, the pooled sensitivity and specificity results from FDG-PET were 91.5% and 95.4%, respectively (95% CIs, 84.3–96.2 and 71.4–98.4, respectively). When the same analyses were performed for CT scanning, a sensitivity of 82.7% (95% CI, 64.2–88.6) and a specificity of 84.1% (95% CI, 68.2–97.0) were found for hepatic lesions. For extrahepatic lesions, the sensitivity and specificity results were 60.9% and 91.1%, respectively (95% CIs, 44.4–68.9 and 66.0–92.8, respectively). Compared with CT, sensitivity for the detection of extrahepatic disease was superior for FDG-PET, emphasizing the utility of FDG-PET for the detection of extrahepatic disease.

Figure 2.

The sensitivity and specificity (Sens/spec) of diagnostic modalities for the detection of colorectal liver metastases are charted for each study and for all outcomes (COMBINED). PET: positron emission tomography; CT: computed tomography.

When the same pooling procedure was applied only to the 6 highest scoring articles, the analysis showed similar results for FDG-PET, as illustrated in Table 4. For hepatic disease, FDG-PET had a sensitivity of 79.9% and a specificity of 92.3%; for extrahepatic disease, these values were 91.2% and 98.4%, respectively. For CT, the sensitivity and specificity results were 85.8% and 88.3%, respectively, for hepatic disease and 55.3% and 95.6%, respectively, for extrahepatic disease. Again, this illustrates the added value of FDG-PET imaging compared with CT scanning for the detection of extrahepatic disease in patients who have hepatic metastases from colorectal carcinoma.

Table 4. Sensitivity and Specificity by Diagnostic Modality for the Six Studies with the Highest Scores
StudyFDG-PET hepatic lesionsFDG-PET extrahepatic lesionsCT hepatic lesionsCT extrahepatic lesions
SensitivitySpecificitySensitivitySpecificitySensitivitySpecificitySensitivitySpecificity
  1. FDG-PET: fluor-18-deoxyglucose-positron emission tomography; CT: computed tomography.

Lai et al., 1996320.930.440.920.951.000.14
Fong et al., 1999260.710.930.850.910.881.000.300.81
Valk et al., 1999440.951.000.920.990.840.950.610.96
Imdahl et al., 2000290.840.860.860.88
Ruers et al., 2002370.650.950.970.800.650.94
Langenhoff et al., 2002331.000.98
Combined studies0.800.920.910.980.860.880.550.96

Weighted Change in Clinical Management

Among the studies that reported scores below the mean according to the proposed index score, only 5 of 14 studies listed a change in management on a patient basis, with a mean percentage change of 35.7% (range, 20.0–58.0%) (Fig. 3). In contrast, 13 of 17 studies that scored above the mean mentioned the change in clinical management. In those series, a percentage change in clinical management of 31.6% (range, 20.8–43.0%) was reported. After correcting for the number of patients in each study, the change in clinical management was 32.9% for the studies that scored below the mean and 30.8% for the studies that scored above the mean.

Figure 3.

The changes in the clinical management (%) of colorectal liver metastases are charted on a patient basis for each study and for all studies combined.

Five of the 6 highest scoring articles focused on the percentage change in clinical management and reported a mean change of 25.0% (range, 20.0–32.0%), which is lower compared with the percentage change in clinical management reported in all other articles. After correcting for the number of patients with liver metastases in the individual studies, the mean change in management that was observed in those 5 articles was 25.4%.

Omissions in the Literature

Because there were no randomized trials in the literature on the subject, the scoring list that was used shows the reported items systematically, thereby identifying omissions (Fig. 1). All articles failed to report some of the items that are useful in the assessment of the research questions discussed above. Items such as the handling of excluded patients and descriptive statistics, and, more specifically, the 95% CIs, the workup of patients for resection, technical aspects and patient preparation for both PET and CT, means and intervals of postoperative follow-up, and the impact of PET on the clinical management of this patient group all are described poorly or inadequately in some of the articles, although all of these items may have a significant impact on the outcome of the study.

DISCUSSION

In the current review, we identified several gaps in the literature that addresses the use of FDG-PET in patients with recurrent colorectal carcinoma and, more specifically, in patients with liver metastases. Most apparent was the lack of one or more randomized clinical trials, which complicates a systematic meta-analysis. Combining the six highest scoring articles may be considered as a surrogate for a randomized trial, because it can provide the most accurate overview of the current data from the literature.

Furthermore, most studies failed to report a number of items, making it difficult to present an appropriate interpretation of the results. Data regarding comorbidity (e.g., diabetes) and patient-selection criteria were not always described appropriately and even were absent, although both items may have had a significant impact on both the imaging results and the study outcome. The same holds true for the characteristics of the primary tumors, because a higher stage results in a greater likelihood of metastatic disease and has a subsequent impact on the diagnostic workup of these patients. The different items that may have a significant impact and that are described inadequately in the literature should be addressed in future research.

Despite the apparent omissions in the literature of the results from imaging recurrent and metastatic colorectal carcinoma, these studies demonstrate the potential utility of FDG-PET in this specific patient group. Most publications (18 of 32 studies) demonstrated the general usefulness of FDG-PET for recurrent colorectal carcinoma, although they did not specifically address hepatic involvement.

Overall, the pooled sensitivity and specificity for FDG-PET compared favorably with the pooled results from CT scanning, clearly indicating that the FDG-PET diagnostic technique is appropriate in this clinical setting, especially for the assessment of extrahepatic disease. For extrahepatic lesions, the pooled sensitivity and specificity results for FDG-PET were 91.5% and 95.4%, respectively, compared with 60.9% and 91.1% for CT scanning. Furthermore, the data showed a ≥ 25% change in clinical management after FDG-PET, indicating that this imaging modality has a significant impact on the treatment of patients with colorectal liver metastases. This is attributed mostly to the detection of extrahepatic disease with FDG-PET, which generally precludes liver resection, as discussed above. The presence of extrahepatic disease may be more important than the detection of additional hepatic lesions by FDG-PET. Such additional lesions detected by FDG-PET seldom turn resectable disease into unresectable disease and, therefore, hardly effect clinical management.22, 36, 37, 40

The current systematic analysis emphasizes the importance of joint interpretation of FDG-PET results and CT scan images. Future developments in technology may enhance further the joint interpretation of multiple diagnostic modalities. This may be accomplished by using combined PET and CT scanners or by fusing software to combine different diagnostic techniques. At the least, multidisciplinary oncologic meetings should be held at which several clinical and nonclinical specialist can perform a joint assessment of all data to review clinical information and diagnostic imaging on a patient-by-patient basis. Only in this way can we guarantee the optimal translation of imaging results into clinical management decisions.

Our proposed index may serve as a tool to distinguish poor studies from more appropriate diagnostic studies in the literature and also may serve as a guideline for reporting this type of study. Obviously, when evaluating trials and trial design without a gold standard, it is possible that the proposed approach may have internal flaws. However, the validity of our approach was tested by searching for the top five articles in the field of study, and those articles coincided with the highest rated articles according to the proposed index. Moreover, the proposed index may be used as a separate instrument for analysis, independent from other endpoints.

In conclusion, despite the apparent omissions in the literature, the pooled results clearly indicate that FDG-PET is useful in the diagnostic workup of patients with potentially resectable hepatic metastases from colorectal carcinoma. Most evident is the detection of extrahepatic disease, in which the FDG-PET demonstrated superior sensitivity and specificity compared with CT scanning. A clear influence of FDG-PET on clinical management was observed in the majority of studies.

Nevertheless, randomized, controlled clinical trials should be performed to investigate the role played by FDG-PET scanning in patients with colorectal liver metastases and its actual impact (e.g., on survival parameters). An additional assessment of the cost-effectiveness of FDG-PET may serve to strengthen its role further.

Ancillary