| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (A.J.B., M.F., J.S., E.J.R.), Nuclear Medicine (H.A.W.), Surgery (F.L., K.O.), and Pathology (K.B.); and 3rd Medical Department (F.L.), Technische Universitaet Muenchen, Klinikum rechts der Isar, Ismaninger Str 22, 81675 Munich, Germany. Received January 11, 2005; revision requested March 16; revision received April 27; accepted June 6; final version accepted June 28. Address correspondence to A.J.B. (e-mail: beer{at}roe.med.tum.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: The study protocol was approved by the local ethics committee. Written informed consent was obtained from all patients. Thirty-one patients with an AEG (stage T3 N0/1 M0 or T4 N0/1 M0) were examined with multi–detector row CT before and 2 weeks after the initiation of chemotherapy. There were seven women and 24 men with a mean age of 62 years ± 8.1 (standard deviation). The maximal transverse tumor diameter was measured and tumor volumetry was performed by three independent readers. The resulting changes were correlated with the histopathologic grade of regression in 21 patients. The differentiation of responders from nonresponders was assessed with receiver operating characteristic analysis in these 21 patients. Interobserver variability was determined in all 31 patients with the Spearman rank correlation. Survival without disease progression was estimated in all patients according to the Kaplan-Meier method. Statistical comparisons between different groups of patients were performed with the log-rank test.
Results: The interobserver variability for the diameter measurements (R = 0.13–0.20) was higher than that for the volumetric measurements (R = 0.70–0.82). The correlation of histopathologic grades of regression with changes in diameter was not statistically significant for the three readers, whereas the correlation of volume changes with histopathologic grades of regression was statistically significant for two of the three readers (P = .01, .05, and .08). Results of receiver operating characteristic analysis revealed an optimal cutoff level for tumor volumetry at a reduction of volume of 14.8%, which resulted in a sensitivity of 100% (six of six patients) and a specificity of 53% (eight of 15 patients). Although the probability of progression was higher in the nonresponder group than in the responder group (61% vs 40%, respectively), the differences were not statistically significant.
Conclusion: Tumor volumetry based on multi–detector row CT can help predict early response to treatment 2 weeks after the initiation of neoadjuvant chemotherapy in patients with AEG; however, the classic approach of tumor diameter measurement failed to show significant correlation with histopathologic tumor regression.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Published data indicate that patients who respond to neoadjuvant therapy ("responders") have a markedly better prognosis after surgery than do those who do not respond to therapy ("nonresponders") (4,5). Morbidity and mortality from tumor resection, however, may be increased in patients who do not respond to neoadjuvant treatment (6,7). It has been shown that positron emission tomography (PET) with fluorine 18 fluorodeoxyglucose (FDG) can help predict tumor response within 2 weeks after initiation of chemotherapy or combined radiation therapy and chemotherapy (8,9). Conversely, none of the routinely used clinical staging methods—including computed tomography (CT), esophagogastroscopy, and endoscopic ultrasonography (US)—has been proved to enable accurate response evaluation (10–12). All of the studies of these methods, however, have focused primarily on the response assessment after completion of neoadjuvant therapy, and no attention has been paid to response evaluation during therapy. To our knowledge, there are currently no data in the literature about early morphologic changes in adenocarcinomas of the esophagogastric junction (AEGs) after the induction of neoadjuvant chemotherapy and the correlation between early changes in tumor morphologic features during neoadjuvant therapy and histologic grades of tumor response.
The purpose of our study, therefore, was to prospectively evaluate multi–detector row CT in the assessment of early response during neoadjuvant chemotherapy for AEGs.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
The study protocol was approved by our local ethics committee. Written informed consent was obtained from all patients.
Preoperative Chemotherapy, Surgery, and Final Patient Population
Patients were treated with two courses of combination chemotherapy (cisplatin, leucovorin, and 5-fluorouracil) as described previously (8). On day 1, cisplatin was administered intravenously over 1 hour at a dose of 50 mg/m2 per body surface area. This was followed by the administration of leucovorin at a dose of 500 mg/m2 per body surface area over 2 hours and concluded with the administration of 5-fluorouracil at a dose of 2 g/m2 per body surface area over 24 hours. Infusion of cisplatin was repeated on days 15 and 29. Administration of leucovorin and fluorouracil was repeated on days 8, 15, 22, 29, and 36. The second course was initiated on day 49. In patients with no contraindications to a three-drug combination, 80 mg/m2 of paclitaxel per body surface area was infused over a period of 3 hours 1 day before the infusion of cisplatin.
Eighteen patients completed two courses of preoperative chemotherapy and underwent surgical resection 3–4 weeks after the last application of chemotherapy. Three additional patients were clinically classified as nonresponders because they showed progressive symptoms such as dysphagia during chemotherapy. In these patients, surgical resection was performed after the first course of chemotherapy. Tumor progression was confirmed in these patients with CT and endoscopy with endoscopic US.
Twenty patients also underwent FDG PET (3). Ten of these 20 patients had a decrease in the standardized uptake value of the tumor of less than 35% after 14 days of chemotherapy and were considered metabolic nonresponders. These 10 patients underwent surgical resection after PET and therefore were excluded from the analysis of response prediction with multi–detector row CT. Thus, 21 patients (six women and 15 men; mean age, 62 years ± 8.1) were included in the study of correlation with histopathologic response, and all 31 patients were included in the receiver operating characteristic (ROC) analysis of interobserver variability and progression-free survival.
Histopathologic Analysis and Response Evaluation
Histopathologic tumor regression was evaluated by a pathologist (K.B.) with 12 years of experience in histopathologic response evaluation. The Mandard score, modified by Becker, was used as described previously (13,14). The grading of tumor regression in response to chemotherapy with use of the method developed by Becker is based on an estimation of the percentage of residual tumor tissue in relation to the macroscopically identifiable tumor bed, as follows: Grade 1 indicates complete or subtotal regression (0% to <10% residual tumor per tumor bed), grade 2 indicates partial tumor regression (10%–50% residual tumor per tumor bed), and grade 3 indicates minimal or no tumor regression (>50% residual tumor per tumor bed).
All patients who demonstrated grade 1 regression were considered histopathologic responders (n = 6). All patients who showed grade 2 or 3 regression were considered histopathologic nonresponders (n = 12). Three patients were classified as nonresponders when neoadjuvant chemotherapy was discontinued after the first course owing to progressive tumor disease. In these three patients, tumor resection was performed after the first course of chemotherapy; all three patients had more than 50% residual tumor per tumor bed in the resection specimen. Therefore, six of 21 patients were classified as histopathologic responders and 15 of 21 patients were classified as nonresponders (12 histopathologic and three clinical nonresponders).
Patient Follow-up
After surgical resection, patients were followed up at 3-month intervals during the 1st year, at 6-month intervals during the 2nd and 3rd years, and at 12-month intervals during the 4th and 5th years. CT of the chest and abdomen and endoscopic examinations of the residual esophagus and anastomosis were performed to detect local recurrence. Endoscopy was performed by staff members of the department of surgery at our institution (2–11 years of experience in endoscopy in patients with AEG). Local recurrence had to be proved with biopsy. The time to recurrence was calculated as the time from the initiation of neoadjuvant therapy to the detection of local recurrence or distant metastases.
CT Protocol
Multi–detector row CT was performed before (examination 1) and 14–17 days after (examination 2) initiation of chemotherapy in all patients (with VolumeZoom [12 patients] or Sensation 16 [19 patients] scanners; Siemens, Forchheim, Germany). Of the 18 patients who completed chemotherapy, 17 underwent a third multi–detector row CT examination immediately before surgical resection (examination 3). One patient was excluded from further analysis because only single-section spiral CT was performed before tumor resection.
Patient preparation included oral administration of 500 mL of water immediately before scanning followed by an injection of 40 mg of N-butylscopolamine for esophagus and stomach dilation. Iodinated contrast material (iomeprol, Imeron 300; Altana, Konstanz, Germany), at a dose of 120–150 mL, was administered intravenously at a flow rate of 3 mL/sec. The thorax and abdomen were scanned in the craniocaudal direction during the portal venous phase (delay, 70 seconds). The following imaging parameters were used: a tube voltage of 120 kV, a tube current of 180 mAs, a collimation of 16 x 0.75 mm (Sensation 16) or 4 x 1 mm (VolumeZoom), and section thicknesses and reconstruction intervals of 0.75 and 0.7 mm, respectively (Sensation 16), and 1.25 and 0.8 mm, respectively (VolumeZoom).
Image Analysis
Image analysis was performed by three independent readers (A.J.B. [reader 1], H.A.W. [reader 2], and M.F. [reader 3]) who were blinded to the clinical findings and the results of surgery or histopathologic examination. Reader 1 was a resident in radiology with 5 years of experience in gastrointestinal imaging. Readers 2 and 3 were residents in radiology with 2 years of experience in gastrointestinal imaging. Data from examinations 1–3 were always analyzed during the same reading session to ensure consistency in the evaluation of all three examinations.
The maximal tumor diameter was determined in the transverse plane at a workstation (Leonardo; Siemens). The transverse section portion of the maximal tumor extension at the baseline examination was determined, and the maximal wall thickness perpendicular to the lumen of the esophagus was measured by each of the readers. In the follow-up examinations, the wall thickness was determined at the same tumor level as in the baseline examination.
To perform tumor volumetry, software (Volume; Siemens) that implements a semiautomatic segmentation algorithm for volume measurements was used (Fig 1). The lower threshold was set to 5 HU and the upper threshold to 1000 HU. The user must define a region of interest at the beginning and end of the tumor and at each area of the tumor where major changes in tumor morphologic features occur. For the sections in which no contours are drawn, the algorithm defines the contours around the tumor by means of interpolation between the sections and creates a volume of interest. All voxels in the volume of interest with HU values between the lower and upper threshold are used for calculating the final volume.
|
|
Statistical Analyses
Quantitative values are expressed as the mean ± 1 standard error of the mean or standard deviation. Within-patient comparisons of absolute values and changes in tumor volume and diameter (means for the three readers) and of the times needed for volumetry were performed with a paired t test.
The Spearman rank correlation was used for estimating the correlation between the readers for the different measurements (relative changes between the examinations) and for assessing statistical significance. The relative percentage changes in diameter and volume were correlated with the Becker score by using Spearman rank correlation.
The diagnostic accuracy of multi–detector row CT in the prediction of subsequent response to chemotherapy calculated by using changes in diameter and volume was evaluated with ROC analysis (15). The optimal cutoff value for differentiating responders from nonresponders was defined as the point on the ROC curve with minimal distance from the 0% false-positive rate and the 100% true-positive rate. The area under the ROC curve (Az) was determined with the corresponding 95% confidence interval. The statistical significance of differences between the ROC curves was based on the 95% confidence interval of the difference between the Az values. Survival without disease progression was calculated from the 1st day of chemotherapy, and survival rates were estimated according to the Kaplan-Meier method. Statistical comparisons between different groups of patients were performed with the log-rank test.
All statistical tests were performed at the 5% level of statistical significance by using the StatView (version 5.0; SAS Institute, Cary, NC) or MedCalc (version 6.15.000; MedCalc Software, Mariakerke, Belgium) program.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
|
|
|
|
|
|
|
Correlation of Morphologic Tumor Changes with Histopathologic Regression
The correlation of histopathologic regression and relative change in tumor diameter (mean for all three readers) between examinations 1 and 2 reached statistical significance (P = .05). When we reviewed the results for the individual readers, however, no statistically significant correlation with histopathologic regression could be observed (Fig 5a).
|
|
For the diameter changes (mean for all three readers), the optimal cutoff level was determined to be –13.2%, with resulting sensitivities and specificities for differentiating responders from nonresponders of 67% (four of six patients) and 93% (14 of 15 patients), respectively. The positive predictive value was 67% (four of six patients), the negative predictive value was 77% (14 of 18 patients), and the accuracy was 86% (18 of 21 patients). The results for the individual readers, however, were worse than the mean results and are summarized in Figure 6a.
|
|
|
|
With use of a volume reduction cutoff level of 14.8%, patients were classified as responders or nonresponders. Corresponding progression-free survival times were then calculated for these subgroups. The mean follow-up time for all patients was 14.9 months ± 7.4 (range, 5–30 months). During the observation period, 13 of the 31 patients (42%) demonstrated tumor progression or recurrence. Of the 14 patients in the responder group, five (36%) demonstrated tumor progression. Eight of the 17 patients in the nonresponder group (47%) demonstrated tumor progression. Eighteen of the 31 patients (58%) demonstrated no tumor progression or recurrence during the observation period. The mean progression-free survival time for these patients was 17.5 months ± 7.4 (range, 7–30 months). In the responder group, the mean progression-free survival time was 19 months ± 6.2 (range, 13–30 months). In the nonresponder group, the mean progression-free survival time was 17 months ± 7.7 (range, 7–28 months). The probability of progression (as estimated according to the Kaplan-Meier method) was higher in the nonresponder group than in the responder group (61% vs 40%, respectively). The differences between the responder and nonresponder groups were not statistically significant (P = .48, log-rank test) (Fig 7).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Previous studies of the response assessment of esophageal cancer have focused on the changes in tumor morphologic features after the completion of neoadjuvant therapy (12,16,17). To our knowledge, only one other study has examined the value of tumor volumetry in response assessment after neoadjuvant chemotherapy of esophageal cancer (18,19); no significant correlation was found between histopathologic tumor regression and changes in tumor volume after the completion of chemotherapy. To the best of our knowledge, our study is the first of its kind, having investigated changes in tumor morphologic features as early as 14 days after the initiation of neoadjuvant chemotherapy. The interval of 14 days after the start of chemotherapy was chosen with reference to former studies involving FDG PET of esophageal cancer, which showed good results at this time point with regard to early response evaluation (8,9). Changes in tumor volume showed a statistically significant correlation with histopathologic regression at this point.
The discrepancy between the negative results of former studies and our positive findings may be explained in part by our use of multi–detector row CT, which has a higher temporal and spatial resolution than does single-section CT. These results indicate that tumor volumetry with multi–detector row CT at an early stage after the start of neoadjuvant chemotherapy could be used in the early differentiation of responders from nonresponders. Theoretically, nonresponders would stop receiving neoadjuvant chemotherapy at this time point and undergo surgical tumor resection. Only responders would receive the full two courses of neoadjuvant chemotherapy before surgery. Although the sensitivity of early volume changes in the prediction of tumor response at the cutoff level of –14.8% was excellent, the specificity was low. Because the percentage of patients with tumors in the upper gastrointestinal tract during neoadjuvant chemotherapy who are nonresponders is usually high, approximately 70% (20), however, a substantial number of nonresponders could be identified despite the low specificity. Within the framework of our study (using a cutoff of 14.8% volume reduction), eight of 15 nonresponders would have been identified with CT volumetry and could have been spared the costs and side effects of chemotherapy with marginal effectiveness. It must be emphasized, however, that these results apply to only our selected patient group with AEG and cannot be applied to patients with other tumors (eg, squamous cell cancer of the esophagus).
We analyzed the progression-free survival of CT responders and nonresponders with a cutoff of 14.8% volume reduction. Although the probability of progression was higher in the nonresponder group than in the responder group (61% vs 40%, respectively), the difference was not statistically significant. The observation period was short, however, and only 13 patients demonstrated tumor progression during this period. The observation period was also not long enough to calculate survival times for the patients. Therefore, these results must be interpreted with caution and further studies with longer observation periods are necessary to determine the relationship between response demonstrated at CT and patient survival.
In this study, we also investigated the interobserver variability of volume changes compared with changes in the maximal transverse diameter. The results of the three readers with regard to diameter changes showed no significant correlation with each other. Conversely, the correlation between the readers with regard to volume changes was statistically significant. This was most likely due to the fact that the absolute values were larger for tumor volume than for diameter and the different measurement techniques used. The measurement of maximal tumor diameter in the transverse plane was difficult to standardize in esophageal cancer because the tumor wall is very complex and irregular. In addition, the tumor very often has a long craniocaudal extension. Therefore, the choice of the image section with the maximal tumor extension can vary considerably from reader to reader. This helps explain some of the differences in the readers' outcomes. Although the use of a single maximal tumor diameter might have been a limitation of our study, it is becoming increasingly common to use only one maximal diameter for response evaluation. Tumor volumetry, conversely, is more standardized, and the only difficulty for the reader is in defining the cranial and caudal borders of the tumors.
We also analyzed the time needed for volumetry with the software. With a mean time of 8.6 minutes (all readers) for the first measurement (examination 1) and 8.2 minutes for the second measurement (examination 2), the procedure is more time consuming than conventional diameter measurements on transverse sections, which only take a few seconds. Two consecutive volume measurements for response evaluation, therefore, take about 17 minutes. The most experienced reader, however, needed significantly less time than did the two less experienced readers. This suggests that work flow can still be optimized with growing experience in gastrointestinal imaging.
In contrast to the few published studies about CT volumetry of esophageal cancer and the lack of studies about early response assessment during neoadjuvant therapy with CT, there is an increasing amount of literature about early metabolic response assessment of esophageal and gastric cancer with FDG PET (8,9,20). Weber et al (8) reported that a reduction in FDG uptake of more than 35% 14 days after the start of neoadjuvant chemotherapy for AEG results in a sensitivity of 93% and a specificity of 95% for response evaluation. Similar exceptional results have been obtained for gastric cancer and squamous cell carcinoma of the esophagus (9,20). Future studies in which findings from FDG PET and multi–detector row CT are compared in the same patient population are necessary to determine the final role of multi–detector row CT volumetry in response evaluation.
Although the findings of our study are encouraging, several limitations must be noted. The study population was small, and owing to limitations arising from a small sample size, only six of the 31 patients (19%) were classified as responders. Moreover, the optimal cutoff value for response evaluation was derived from the data obtained in the present study. This post hoc definition may cause the diagnostic accuracy of multi–detector row CT to be overestimated. Thus, the optimal cutoff value for predicting response must be confirmed in a larger study population. Moreover, histopathologic response evaluation is a surrogate parameter, and our results still must be correlated with patient survival after a longer follow-up period. The studies by Mandard et al (14) and Weber et al (8), however, support the use of histopathologic tumor regression for response evaluation in esophageal cancer and show that patients with complete or nearly complete tumor regression are most likely to benefit from neoadjuvant therapy.
Finally, it cannot be excluded that the results could have been improved with more experienced readers, especially with regard to the interobserver variability. The use of readers with intermediate experience, however, is realistic with regard to the introduction of volumetry into a routine clinical setting, because in most departments, residents are probably the first to perform the measurements, after which they discuss the results with fully trained faculty radiologists. Therefore, the method must be robust enough to work in the hands of less experienced users.
In conclusion, early morphologic tumor changes during neoadjuvant chemotherapy for AEG can be measured with multi–detector row CT. These preliminary results indicate that an early response assessment based on volumetric measurements is feasible and may be an effective alternative to FDG PET. The optimal time point for response evaluation and the optimal cutoff value must still be determined prospectively in a larger patient population. With the increase in the use of combined PET/CT scanners, a combination of CT volumetry and standardized uptake value measurements might improve the accuracy of early response evaluation in the future.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: AEG = adenocarcinoma of the esophagogastric junction • Az = area under the ROC curve • FDG = fluorine 18 fluorodeoxyglucose • ROC = receiver operating characteristic
Author contributions: Guarantors of integrity of entire study, A.J.B., H.A.W., F.L., M.F., J.S., E.J.R.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, A.J.B., H.A.W., F.L., M.F., J.S.; clinical studies, A.J.B., H.A.W., F.L., K.O., M.F.; statistical analysis, A.J.B., H.A.W., M.F., K.B., J.S.; and manuscript editing, all authors
Authors stated no financial relationship to disclose.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  B. J. Krause, K. Herrmann, H. Wieder, and C. M. zum Buschenfelde 18F-FDG PET and 18F-FDG PET/CT for Assessing Response to Therapy in Esophageal Cancer J. Nucl. Med., May 1, 2009; 50(Suppl_1): 89S - 96S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Allen-Auerbach and W. A. Weber Measuring Response with FDG-PET: Methodological Aspects Oncologist, April 1, 2009; 14(4): 369 - 377. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. R. Benz, M. S. Allen-Auerbach, F. C. Eilber, H. J.J. Chen, S. Dry, M. E. Phelps, J. Czernin, and W. A. Weber Combined Assessment of Metabolic and Volumetric Changes for Assessment of Tumor Response in Patients with Soft-Tissue Sarcomas J. Nucl. Med., October 1, 2008; 49(10): 1579 - 1584. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Ott, K. Herrmann, F. Lordick, H. Wieder, W. A. Weber, K. Becker, A. K. Buck, M. Dobritz, U. Fink, K. Ulm, et al. Early Metabolic Response Evaluation by Fluorine-18 Fluorodeoxyglucose Positron Emission Tomography Allows In vivo Testing of Chemosensitivity in Gastric Cancer: Long-term Results of a Prospective Study Clin. Cancer Res., April 1, 2008; 14(7): 2012 - 2018. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Suzuki, H. Jacobsson, T. Hatschek, M. R. Torkzad, K. Boden, Y. Eriksson-Alm, E. Berg, H. Fujii, A. Kubo, and L. Blomqvist Radiologic Measurements of Tumor Response to Treatment: Practical Approaches and Limitations RadioGraphics, March 1, 2008; 28(2): 329 - 344. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. A. Weber and R. Figlin Monitoring Cancer Treatment with PET/CT: Does It Make a Difference? J. Nucl. Med., January 1, 2007; 48(1_suppl): 36S - 44S. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
