The objective of the NEOPEC-trial is to assess the accuracy and negative predictive value of FDG-PET and CT-scan for the early prediction of non-response to preoperative chemoradiotherapy in patients with potentially curable esophageal cancer. These parameters are defined as primary endpoints.
Secondly, the correlation between tumor response as predicted by FDG-PET and CT-scan or as measured by histopathologic assessment in the resection specimen and long term survival will be determined. This correlation will be used as a secondary endpoint.
In September 2004 a multicenter randomized phase-III trial (175 patients in each arm) was initiated (CROSS trial) in the Netherlands, comparing neoadjuvant chemoradiotherapy followed by surgery with surgery alone in patients with potentially curable esophageal cancer. Long term survival and quality of life are the primary endpoints in this therapy-related trial. (ISRCTN80832026)
The present research proposal comprises serial FDG-PET and CT-scan before and during neoadjuvant therapy in the multimodality treatment arm of the randomized trial. Finally, resection specimens will be assessed for tumor response. Figure describes the logistic integration of the NEOPEC-trial within the CROSS-trial.
integration NEOPEC-trial within CROSS-trial. CRT chemoradiotherapy, PET positron emission tomography, CT computed tomography.
Patients with histologically proven squamous cell carcinoma or adenocarcinoma of the esophagus or gastro-esophageal junction will undergo extensive preoperative staging, including spiral CT-scan of the chest and abdomen. The tumor must not extend more than 2 cm into the gastric cardia. Longitudinal tumor length must not exceed 8 cm, radial size must not exceed 5 cm. cT1N0 tumors are not eligible. In the absence of local irresectability and/or distant dissemination patients with an acceptable general condition (ECOG performance status 0, 1, 2; weight loss < 10%) will be invited to participate in the randomized trial. After written, voluntary, informed consent and stratification for tumor type, treatment center, clinical lymph node status, and WHO performance the patients will be randomized between the two treatment-arms (neoadjuvant chemoradiation followed by surgery versus surgery alone). Patients who are randomized for the neoadjuvant arm will consequently be asked to participate in the NEOPEC-trial.
Concurrent chemoradiotherapy CROSS-trial
Over a 5 week period concurrent chemoradiotherapy is applied on an outpatient basis.
Paclitaxel (50 mg/m2) and carboplatin (area-under-curve = 2) are administered by i.v. infusion on days 1, 8, 15, 22, and 29. Additional medication consists of dexamethason, clemastine, ranitidine, and ondansetron. Vital signs are monitored every fifteen minutes during paclitaxel administration.
External beam radiation with a total dose of 41.4 Gy is given in 23 fractions of 1.8 Gy, 5 fractions a week, starting the first day of the first cycle of chemotherapy. Radiation therapy will be delivered using a multiple field technique. All patients will undergo 3-D planning using a 3-D conformal technique.
Surgery, i.e. either transhiatal (1 field lymphadenectomy) or transthoracic (2 field lymphadenectomy) esophagectomy including lymph node dissection around the celiac axis, will be performed within 6 weeks of randomization (surgery alone arm) or within 6 weeks after completion of the chemoradiation (multimodality treatment arm). For esophageal surgery a clear correlation has been shown between hospital volume and quality of care. Therefore, only centers which perform at least 10 esophagectomies per year participate in this trial.
Evaluation of tumor response by FDG-PET and spiral CT
PET-scan and CT scan will be performed at baseline and at day 15. FDG uptake in the tumor will be assessed semi quantitatively. PET-scans will be performed in several PET centers, using standardized acquisition and image reconstruction, with centralized data-analysis (VUMC PET data-analysis center) and after inter institute calibration of scanners. [19
For quantitative analysis of FDG uptake the standardized uptake value (SUV) and the simplified kinetic method will be assessed using previously validated methodology. [9
] Serum glucose levels will be measured using the standard hexokinase method. Fractional changes of FDG-uptake between baseline and early during treatment will be correlated with tumor response in the resection specimen. Further ROC analysis will be performed in order to obtain an optimal cut-off value of decrease in FDG uptake for the prediction of non-responders.
A target CT will be performed using a multi-slice CT scanner (minimal four-slice CT-scanner) after administration of 100 ml low osmolar intra venous contrast medium. After a 25 second delay a spiral scan CT is performed targeted at the esophagus and esophagogastric junction including the maximal extension of the tumor (maximum slice thickness 3 mm, maximum pitch ≤ 1.5, 150 mAs) with inferior-superior scanning. Scan settings concerning spatial resolution and contrast timing will be identical for pre- and post-treatment CT-scans. The tumor area will be calculated quantitatively by 3D volume measurement and multiplication of the greatest tumor diameter and the greatest perpendicular distance in the axial plane, as this is the clinically accepted method. [23
] A decrease in tumor area or tumor volume of at least 50% is defined as tumor response. However, as for the FDG-PET data, ROC analysis will be performed to assess the optimal cut-off value for tumor response.
Evaluation of tumor response in the resection specimen: reference standard
The 'reference standard' of the post-chemoradiation disease stage will be based on the histology in the surgical resection specimen. Treatment response will be classified according to the Mandard-score which is based on the percentage of viable residual tumor cells in relation to fibrosis/necrosis. [25
] Responders are defined as patients with < 10% viable residual tumor cells; non-responders are defined as patients with > 10% viable residual tumor cells.
The accuracy of early FDG-PET and CT-scan in distinguishing responders from non-responders will be determined by using ROC-analysis to show the effect of different thresholds in early response to identify non-responders. Additional analyses will focus on the best cut-off points for PET- and CT-measurements to obtain the most effective approach for costs and quality of care.
Sample size calculation
The sample size calculation is based on two considerations. First, to compare the accuracy between PET and CT in distinguishing responders from non-responders, the area under the ROC curve (AUC) will be used. Based on results of a previously publicized meta-analysis, the expected difference in AUC between PET and CT is 0.87 vs. 0.54. [7
] To detect such a difference in an unpaired design, approximately 80 patients are required (with a power of 80% and a significance level of 5%), assuming that the proportion of responding patients is 50% (i.e. 40 responders in a group of 80 patients). [6
Second, to be useful in clinical practice the best strategy to measure early treatment response (supposedly PET) should have a sufficiently high negative predictive value. At this moment, it is anticipated that the negative predictive value should be at least 0.80. Based on an expected sensitivity (true responders correctly identified by PET) of 95% and specificity (true non-responders correctly identified by PET) of 70%, 130 patients are needed to ensure that the lower limit of the exact 95% confidence interval around the negative predictive value will not reach below 0.80 (expected value of negative predictive value is then 0.93; 95% CI 0.83 to 0.99). The total number of 130 patients is based on the assumption that the proportion of responders will be 50%.
Therefore, a sample size of 130 evaluable patients is needed to answer both primary research questions. Anticipating a drop-out of 20 patients for various reasons after initial inclusion, we plan to include a total number of 150 patients in 3 years.
The use of FDG-PET and CT-scan in all patients will increase costs however early discontinuation will reduce the costs of neoadjuvant treatment in those timely detected as non-responders. Therefore, the economic evaluation will be designed as a cost-effectiveness study.
The reference strategy is neoadjuvant treatment in all patients, without imaging to predict early response. The alternative strategies are FDG-PET or CT-scan in all patients and early discontinuation of neoadjuvant treatment in patients who test as negative (non-response). We will compare the incremental costs of these alternative strategies and the associated changes in survival and quality of life, relative to the reference strategy.
We will estimate the direct medical costs and indirect (time) costs for all strategies by documenting resource utilization by counting volumes and using unit prices. The relevant cost items include: FDG-PET, CT-scan, neoadjuvant chemoradiotherapy (including the management of associated morbidity), surgery (distinguishing between patients with and without neoadjuvant treatment), and after hospital discharge: out-patient consultations, out-of-hospital consultations and hospital re-admissions.
Volume data will be collected through patient-specific case record forms and data retrieval from hospital information systems. Standard unit prices will be derived from the Dutch guideline on unit costing for economic evaluations in health care, from actual unit costing in the participating hospitals related to Dutch Diagnosis Related Groups (DBC) billing system, from previous research, and from charges in case of volume data of lesser significance.[27
] The unit costs derived from multiple centers will be averaged.
A decision analytic approach will be used to compare the overall survival, quality of life and costs of the three strategies: neoadjuvant chemoradiotherapy in all patients, without prediction of treatment response, FDG-PET to predict tumor response, with early discontinuation of chemoradiotherapy in patients with non-response and CT-scan to predict tumor response, with early discontinuation in patients with non-response.
The present study was approved by the Medical Ethics Committee of the Academic Medical Center in Amsterdam, the Netherlands, with reference number MEC 04/210.