| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (A.R.v.E., M.E.J.P., A.A.v.d.B.H., M.N.J.M.W., J.L.B.) and Oncological Surgery (C.J.H.v.d.V.), Leiden University Medical Center, Albinusdreef 2, PO Box 9600, 2300 RC Leiden, the Netherlands. Received August 3, 2001; revision requested September 10; revision received November 14; accepted January 7, 2002. M.E.J.P. supported by the Dutch Cancer Foundation, with a grant for work on the detection of hepatic metastases in patients with colorectal cancer. Address correspondence to A.R.v.E. (e-mail: a.r.van_erkel@lumc.nl).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: With use of a MEDLINE search (January 1994 to January 2001), articles were selected that contained original results on detection of hepatic metastases of colorectal cancer, categorized for size in at least two categories, with use of helical computed tomography (CT), helical CT at arterial portography, or magnetic resonance imaging. Results were compared with the size distribution of hepatic metastases in 47 consecutive patients with colorectal carcinoma, which were detected by using a combination of intraoperative ultrasonography (US) and palpation.
RESULTS: Seven studies met all predefined criteria. Four studies involved intraoperative US in all patients and demonstrated a significant negative correlation (-0.988) between detection rate and fraction of small metastases. These studies had a higher fraction and lower detection rate of small metastases and a lower overall detection rate. A majority (58% [145 of 252]) of metastases in the study population were smaller than 20 mm.
CONCLUSION: Few articles adequately describe the standard of reference and size distribution of hepatic lesions. Hepatic metastases of colorectal cancer are frequently smaller than 20 mm. When the standard of reference is suboptimal, many small metastases are excluded from analysis, and detection rates are therefore inflated.
© RSNA, 2002
Index terms: Colon, CT, 75.12115, 75.12119 • Colon, MR, 75.1214 • Colon, US, 75.12982 • Colon neoplasms, 75.321 • Liver neoplasms, metastases, 761.332 • Technology assessment
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
High detection rates for hepatic metastases have been reported for magnetic resonance (MR) imaging, helical computed tomography (CT), and helical CT at arterial portography (CTAP) (3–6). High (>50%) recurrence rates of metastatic disease distant from resected metastases (7,8), however, suggest that many metastases are small and are missed on preoperative images. Reports on high detection rates may be associated with a suboptimal standard of reference. The quality of the standard of reference is an issue because pathologic-anatomic examination of the entire liver is not possible for both practical and ethical reasons. Investigators in previous studies (9,10) have demonstrated that intraoperative ultrasonography (US) combined with intraoperative palpation of the liver is the best alternative standard of reference.
A suboptimal standard of reference may result in a too-small fraction of small lesions and overestimation of the detection rate. We suggest that the size distribution of the metastases has a major effect on the results and reflects the quality of the standard of reference.
The purpose of our study was to determine the relationship between the size of hepatic metastases, the standard of reference, and the reported detection rate in patients with colorectal cancer.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Lesions were categorized as either benign or malignant on the basis of characteristic appearance and compressibility at intraoperative US, consistency at palpation, and histologic findings. On the basis of intraoperative US, sharply defined, thin-walled anechoic lesions with increased through transmission were considered benign cysts. Hyperechoic geometrically defined lesions that were compressible at intraoperative US and soft at palpation were considered hemangiomas. Lesions that did not meet these criteria were considered potentially malignant. These were either resected or subjected to fine-needle aspiration biopsy. In addition, histologic confirmation of at least one representative lesion was obtained in each patient. The reference diagnosis was thus based on a combination of palpation and intraoperative US findings, and when the criteria of a benign lesion were not met, it was based on histologic findings. The total number of malignant and benign lesions per patient was recorded.
The maximal diameter of each lesion was recorded by using intraoperative US. The size distribution of malignant and benign lesions was determined by using the following size groups: smaller than 5, 5–9, 10–19, 20–29, 30–39, and 40 mm or larger. For the comparison between size and histologic nature, we used cutoff values of 10 and 20 mm and calculated positive and negative predictive values. Typical benign lesions at intraoperative US and palpation (cysts and hemangiomas) and histologically proven benign lesions that were not classified as typically benign at intraoperative US were combined and analyzed as benign lesions.
Literature Review
We searched MEDLINE literature for articles published between January 1994 and January 2001 that were limited to humans and the English language. We used the following search: (a) liver neoplasms (subject headings: diagnosis and radiography); (b) tomography, x-ray computed (all subject headings); (c) MR imaging (all subject headings); (d) combine a and b; (e) combine a and c; (f) combine d or e. The titles of the articles found were screened. An article was selected when the title mentioned detection or characterization of hepatic lesions with MR imaging and/or CT. The bibliographies of the selected articles were searched for articles on the same topic that were not found with the MEDLINE search.
The articles were reviewed and selected by A.R.v.E. The abstract, methods, and results sections of all selected papers were read, and an article was definitely selected for analysis when all of the following criteria were met: (a) The article contained original results on the detection of hepatic metastases of colorectal cancer; (b) helical CT (in portal and/or arterial phase following administration of intravenous contrast agents), helical CTAP, and/or MR imaging were used as the diagnostic modalities; (c) The study involved at least 10 patients; (d) size-specific detection of hepatic lesions was provided for at least two size categories (eg, for lesions < 2 or 
2 cm). For each category, the detection rate and actual number of lesions were specifically stated.
For the analysis, only the mean number, detection rates, and size distribution of hepatic metastases were included. Benign or other malignant lesions were not included. The mean number of metastases, the overall detection rate, and detection rate per size stratum were compared with the size distribution of hepatic metastases in patients with colorectal carcinoma, as determined in the patient population of the current study. Since the threshold value of 10 mm was used most consistently in the articles that met all criteria, this threshold value was used for the analyses per size stratum. When detection rates of more than one imaging modality were reported, the maximal detection rate reported was used for analysis. When two imaging modalities were combined and separate detection rates could not be discerned from the report, the combination of modalities was used.
Statistical Analysis
To determine the relationship between the size and histologic nature of the hepatic lesions in the patients in the current study, the sizes of malignant and benign lesions, as measured with intraoperative US, were compared by using the median test. To determine if, in published reports, the reported detection rates correlated with the fraction of small (<10-mm) metastases, the two-tailed Pearson correlation test was used.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
Literature Review
The MEDLINE search resulted in a total of 1,442 articles. From these and the bibliographies of the selected articles, 155 articles were selected for further reading on the basis of their titles. Detection rates for hepatic metastases of colorectal cancer (criterion a) were not provided in 141 of the articles. In 41 of these 141 articles, detection rates were reported for hepatic lesions not solely consisting of metastases of colorectal cancer. Of these 41 articles, 13 included size-categorized detection rates.
In 14 studies (3–5,11–21), detection rates for hepatic metastases of colorectal cancer were reported. One study (17) was excluded on the basis of criterion b. In this study, the detection rate for dynamic incremental CT was reported. Five more studies (4,18–21) were excluded because no size-stratified detection rates were reported. Only eight studies (3,5,11–16) met all predefined criteria. The article by Schmidt et al (16) was excluded from the analysis, since it was based on the same data as the earlier article by Strotzer et al (13). The results of the seven remaining studies are summarized in the Table. It is remarkable that all seven studies involved intraoperative US as the standard of reference—in four studies (11,12,14,15) in all patients and in three studies (3,5,13) in part of the patient population. The overall recalculated detection rate of these seven studies was 75% (389 of 516 metastases). The mean number of metastases per patient in the seven reviewed studies was 2.3 (516 divided by 220; range, 1.5–3.6). In the three studies in which intraoperative US was performed in only part of the population (3,5,13), the mean number of metastases per patient was 1.7 (175 divided by 106). In the four studies in which intraoperative US was performed in all patients (11,12,14,15), the mean number of metastases per patient was 3.0 (341 divided by 114).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
The quality of the standard of reference is crucial in reports on detecting hepatic metastases of colorectal cancer. Pathologic examination of the entire liver is not an option, and pathologic examination of the resected hepatic specimen alone will result in overestimation of the detection rate, since part of the liver is not subjected to the standard of reference (verification bias). The best alternative standard of reference consists of surgical inspection and palpation, in combination with intraoperative US performed by an experienced radiologist and histologic confirmation when the nature of a lesion is unclear (9,10). It is, therefore, no surprise that the seven studies on the detection of hepatic metastases that fulfilled all criteria for analysis (3,5,11–15) all involved intraoperative US as the standard of reference. This standard of reference by no means enables detection of all hepatic metastases. The lower limit in our patient group was 2 mm, and only 8% (19 of 252) of metastases were smaller than 5 mm. Also, lesions larger than 5 mm may occasionally be missed. Even with this standard of reference, the detection rates of the evaluated imaging modalities will be overestimated.
The size distribution of hepatic metastases is indicative of the quality of the standard of reference. In reporting the detection rate of an imaging modality for hepatic metastases, it is therefore equally important to adequately describe the standard of reference used and the size distribution of hepatic metastases, preferably with size-stratified detection rates. In recent literature, only a few studies (3,5,11–16) meet these criteria.
Of the eight studies that met our predefined criteria (3,5,11–16), the study by Schmidt et al (16) was excluded from the analysis, since it was based on the same data as the study by Strotzer et al (13). The seven remaining studies (3,5,11–15) demonstrated great variance in the fraction of metastases smaller than 10 mm (25%–53%) and in detection rates (55%–98%). The four studies in which intraoperative US was performed in all patients (11,12,14,15) demonstrated a significant negative correlation between detection rate and fraction of small metastases (two-tailed Pearson correlation test, -0.988). Although this correlation was statistically significant (P = .012), it was based on only four studies; thus, it was heavily influenced by each value. A greater fraction of small metastases was associated with a lower detection rate of hepatic metastases. In general, the studies in which intraoperative US was performed in all patients (11,12,14,15) had a higher fraction of small hepatic metastases, a lower detection rate for small metastases, and a lower overall detection rate, as compared with the studies in which intraoperative US was performed in a part of the population (3,5,13). Only Lencioni et al (14) reported a high detection rate of 94% (34 of 36) for helical CTAP in a study with a complete standard of reference (14). The fraction of small (10-mm) metastases in that study was, however, smaller than the fraction of small metastases in the three MR imaging studies in which intraoperative US was performed in all patients (11,12,15). The lower detection rates in these studies, as compared with that in the study by Lencioni et al (14), can be explained by the smaller fraction of small metastases in the population described by the latter authors.
The outlier in the seven studies was the study by Strotzer et al (13). The combination of helical CT and CTAP resulted in detection rates of 96% (51 of 53) for all metastases and 93% (26 of 28) for metastases 10 mm or smaller. Despite the fact that in the study by Strotzer et al intraoperative US was performed in only 21 of the 35 patients, the fraction of metastases 10 mm or smaller was as high as 53% (28 of 53). One can only speculate about the reason for this. The combination of helical CT and CTAP may indeed lead to increased detection of small metastases. A standard of reference not used independently of the tested imaging modalities might then result in a high fraction of small metastases. However, the low mean number of metastases per patient of 1.5 (35 of 53) in that study may also reflect a different patient population.
The mean number of metastases per patient (5.4) in the current study population was high, as compared with that (2.3) in the seven reviewed studies (3,5,11–15). For the three studies in which intraoperative US was performed in only part of the population (3,5,13), this difference can be explained by the quality of the standard of reference. In the four remaining studies (11,12,14,15), intraoperative US was performed in all patients, and the standards of reference were identical. Patient selection might play a role in the differences in the number of metastases per patient. In 62% (29 of 47) of the patients in the current study, resection of metastases was considered impossible. In the study by Hagspiel et al (11), all 13 patients underwent partial hepatic resection. In the other three studies (12,14,15), data were insufficient to determine the fraction of patients considered able to undergo surgery.
The current study was limited by the fact that only seven articles could be used for final analysis. Since only English-language literature was analyzed, relevant articles in other languages may have been missed. Only four studies (11,12,14,15) that fulfilled all criteria for inclusion involved intraoperative US in all patients and were used to determine the correlation between size and detection rates.
A secondary finding in the evaluation of the patients in the current study with colorectal cancer was that size is partially helpful in characterizing focal hepatic lesions in patients with colorectal cancer and hepatic metastases. Lesions 20 mm or larger were most likely to be metastases, with a positive predictive value for malignancy of 91% (107 of 118 lesions). The negative predictive value, however, was only 39% (92 of 237 lesions). Characterization of focal hepatic lesions smaller than 20 mm is not possible on the basis of size alone, because a majority of benign lesions are, as metastases, smaller than 20 mm.
Our review of the literature demonstrates that only a few articles, in which the detection rate of modern imaging modalities for hepatic metastases is reported, have adequate description of the standard of reference and size distribution of the studied lesions. In studies on the detection rate of hepatic metastases in patients with colorectal cancer, the quality of the standard of reference determines the size distribution of hepatic metastases and therefore the reported detection rates. Detection rates are lower in studies with a high-quality reference standard and therefore a larger fraction of small metastases.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, A.R.v.E.; study concepts and design, A.R.v.E., M.E.J.P., A.A.v.d.B.H., M.N.J.M.W.; literature research, A.R.v.E., M.E.J.P.; clinical studies, A.R.v.E., M.E.J.P., M.N.J.M.W., C.J.H.v.d.V.; data acquisition, A.R.v.E., M.E.J.P.; data analysis/interpretation, A.R.v.E., M.E.J.P., A.A.v.d.B.H.; statistical analysis, A.R.v.E., M.E.J.P., A.A.v.d.B.H.; manuscript preparation, A.R.v.E., J.L.B.; manuscript definition of intellectual content, all authors; manuscript editing, A.R.v.E., J.L.B., M.N.J.M.W.; manuscript revision/review, A.R.v.E., M.E.J.P., J.L.B., M.N.J.M.W.; manuscript final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. J. Kim, K. W. Kim, J. H. Byun, H. J. Won, Y. M. Shin, P. N. Kim, M.-S. Lee, and M.-G. Lee Comparison of mangafodipir trisodium- and ferucarbotran-enhanced MRI for detection and characterization of hepatic metastases in colorectal cancer patients. Am. J. Roentgenol., April 1, 2006; 186(4): 1059 - 1066. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. W. Kim, A. Y. Kim, T. K. Kim, S. H. Park, H. J. Kim, Y. K. Lee, M.-S. Park, H. K. Ha, P. N. Kim, J. C. Kim, et al. Small (l.e. 2cm) Hepatic Lesions in Colorectal Cancer Patients: Detection and Characterization on Mangafodipir Trisodium-Enhanced MRI Am. J. Roentgenol., May 1, 2004; 182(5): 1233 - 1240. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
