1Departments of Nuclear Medicine, 2Internal Medicine, 3Radiology, 4Radiation Oncology, 5Pathology and 6Thoracic Surgery, University Hospital Gasthuisberg, Katholieke Universiteit Leuven (KUL), Leuven, Belgium
Received 9 July 2001; revised 25 September 2001; accepted 25 October 2001.
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
This prospective study was designed to determine the utility of 18F-labelled deoxyglucose (FDG) in positron emission tomography (PET) (FDG-PET) for assessing the response to neoadjuvant chemoradiation therapy (CRT) in locally advanced oesophageal tumours.
Patients and methods
Thirty-six patients with locally advanced oesophageal cancer (clinical T4 stage) without organ metastases, underwent FDG-PET before and 1 month after CRT. Patients were classified as major responders by serial FDG-PET when the post-CRT PET demonstrated a strong reduction of FDG uptake at the primary tumour site (>80% reduction of tumour-to-liver uptake ratio) without any abnormal FDG uptake elsewhere in the body. PET response was compared with histology obtained during post-induction transthoracic oesophagectomy.
Results
A strong correlation was found between the extent of lymph node (LN) involvement as shown by the pre-CRT PET and the major response rate (P = 0.001): such response occurred in nine of 11 N0M0 patients (82%), in three of nine N12M0 patients (33%) and in two of 16 patients (13%) with distant lymphatic spread. Such a correlation was not found for computed tomography or endoscopic ultrasonography. The sensitivity of serial FDG-PET for a major CRT response was 10 of 14 (71%), its specificity 18 of 22 (82%). The concordance between the response assessment by PET and histopathology was 78%. The median survival time after CRT of PET major responders compared with PET non-major responders was 16.3 months and 6.4 months, respectively. The metabolic response as measured by serial FDG-PET is a stronger prognostic factor for overall survival (P = 0.002) than the extent of LN involvement seen on the pretreatment FDG-PET (P = 0.087).
Conclusions
These data indicate that CRT response as assessed by serial FDG-PET is strongly correlated with pathological response and survival.
Key words: induction therapy, oesophageal cancer, positron emission tomography, response assessment, survival
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
Materials and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Chemoradiation therapy
All patients received concomitant external beam radiotherapy (4 weeks, 5 days/week; 40 Gy total) and cisplatin [80 mg/m2 intravenously (i.v.); on day 1 and 21] and 5-fluorouracil (800 mg/m2/day i.v. continuous infusion; on days 14 and 2125). The radiation field included all regional LN stations.
Post-CRT preoperative evaluation
Four to five weeks after completion of the CRT, preoperative staging was performed including CT of chest and abdomen, EUS and a whole-body FDG-PET. The PET, CT and EUS findings, together with all other staging data, were reviewed and correlated at a multidisciplinary tumour conference involving thoracic surgeons, a medical oncologist, a pathologist, a radiation oncologist, a diagnostic radiologist and a nuclear medicine physician. Progressive disease indicated by a new suspect lesion appearing on the post-CRT PET affected patient management only after confirmation by biopsy or by dedicated radiographic techniques. If staging did not demonstrate persistent local irresectability of the primary tumour or overt progressive metastatic disease, a surgical exploration was performed (n = 30). If at that time resectable disease was found, an oesophagectomy with curative intent was performed.
Surgery
The surgical approach employed varied with the location of the tumour, as reported previously [8]. A subtotal oesophagectomy in conjunction with a partial gastrectomy and a cervical anastomosis was performed using a transthoracal approach. All lymphatic tissue localised in the upper abdominal compartment was removed. In the chest, a so-called posterior mediastinectomy was performed. As the surgical morbidity might be higher after preoperative CRT, a cervical lymphadenectomy was not routinely performed.
Spiral CT of chest and abdomen
Each patient underwent a spiral CT examination of the chest and abdomen for tumour staging before and 4 weeks after preoeprative CRT, using the following acquisition parameters: slice thickness, 5 mm; table speed, 9 mm/rotation, pitch, 1.8; reconstruction interval, 5 mm. The following i.v. contrast injection parameters were used: concentration, 350 mg iodine/ml; velocity of injection, 2 ml/s; injection time, 40 s; scan delay, 35 s; total volume of contrast, 80 ml. The patient was positioned supine with the arms above the head. A meal with an oral bolus of gastrografine was administered at the time of scanning in order to dilate the oesophagus for a better examination of the tumour. The scanning was performed within one breathhold. LNs measuring 10 mm or more at their maximum cross-sectional diameter were considered to be metastatic.
PET imaging
Whole-body FDG-PET imaging was performed on a CTI-Siemens 931 or an HR+ scanner (SAMI; Knoxville, TN, USA) with axial fields of view of 10.1 and 15 cm, and spatial resolutions of 8 and 6 mm, respectively. The choice between PET scanners was determined by the availability. All patients fasted for at least 6 h preceding the examination. Sixty minutes after the i.v. injection of 6.5 MBq/kg FDG (to a maximum of 555 MBq) a whole-body emission scan was initiated. Imaging was started at the upper third of the thighs and reached to the head. The raw imaging data were reconstructed in a 128 x 128 matrix with use of an iterative reconstruction algorithm.
For the purpose of this study, all PET images were reviewed in batches by two experienced nuclear medicine physicians (A.M., J.P.C.), who were blind to the data concerning the patients. The pre- and post-CRT images (transaxial, coronal and sagittal views) were compared using a side- by-side display on a high resolution monitor. In the case of diagnostic discordances between the observers, a consensus diagnosis was generated.
The PET response was assessed firstly lesion by lesion, and then on a patient basis, by comparing lesion contrast between the pre-CRT and the post-CRT PET images. The observers expressed these PET changes as complete responses (CR) in the case of a complete normalisation of the lesion, quasi-complete responses (qCR) in the case of an almost complete disappearance of the lesion, partial responses in the case of a partial decrease of the lesion intensity, stable disease and progressive disease in the case of increase of the lesion intensity or of the appearance of a new lesion not seen on the pre-CRT PET.
Then, a patient-based classification of the PET response was performed. Patients were categorised as PET major responders compared with PET non-major responders. The PET major responders were those patients in whom the post-CRT PET showed a CR or a qCR of the primary tumour, a CR of all LN metastases observed on the pre-CRT PET, and the non-appearance of new lesions not seen on the pre-CRT PET. All other patients were classified as PET non-major responders.
A semi-quantitative analysis of the PET response, using tumour-to-liver uptake ratios (TUR) was performed on the primary tumour in order to define the PET response in the case the visual analysis revealed a quasi-complete normalisation of the FDG uptake at the primary tumour. A standardised methodology, illustrated in Figure 1, was used. First, the transaxial slices of the pre- and post-CRT PETs were visually aligned. A circular region of interest (ROI) of 16 cm2 was drawn centrally in the liver parenchyma on the pre-CRT PET and then copied to the corresponding post-CRT PET slice. The mean activities in the liver ROI were used for normalisation of the uptake of the primary tumour. The activity of the primary tumour was determined by the maximal pixel uptake in the tumour volume. The PET response of the primary tumour was semi-quantitatively expressed as deltaTUR, reflecting the therapy-induced decrease of the pre-CRT TUR: deltaTUR = (TUR pre-CRT-TUR post-CRT)/TUR pre-CRT) %.
|
A patient was classified as a CRT major responder when two criteria were fulfilled. First, the histology of the primary tumour indicated a pT02 stage, or a pT3 stage. The pT3 stage was considered as reflecting a major response only if the residual tumour consisted of small foci of viable tumour on a background of extensive histopathological response. Secondly, there was no sign of any tumoural viability beyond the primary tumour site (i.e. pN0M0 stage).
Thus, all patients with histological evidence of residual tumoural viability in a LN were considered as CRT non-major responders, regardless of the extent of histological signs of the CRT response. A pathological CR was defined as a pT0N0M0 stage.
Data analysis
Sensitivity and positive predictive values were calculated using standard statistical operations. Survival time was defined as the time interval from the end of CRT to the day of death (complete) or last contact (censored). Median survival times were calculated using the KaplanMeier method. Difference in survival between PET responders compared with PET non-responders was analysed using the log-rank test. Numerical data are presented as mean ± standard deviation. The difference of deltaTUR between the CRT responders compared with the CRT non-responders was analysed using a Students t-test. Chi-square testing was performed to check for a relationship between categorical parameters. If sample size was small, the overall total of the contingency table being <40, a Fishers exact test was used. The relationship between extent of LN involvement and CRT response was assessed using 3 x 2 tables and a Pearson chi-square test. Multivariate analysis by proportional hazard (Cox) regression was performed to evaluate the effect of the prognostic significance of the extent of LN involvement seen on pre-CRT PET and the PET response on overall survival. P values >0.05 were considered as not significant (NS).
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
Pre-CRT staging related to the response rate
All primary tumours were clearly visualised on the pre-CRT PET. The mean TUR of the primary tumours before CRT were normally distributed with a mean of 21 ± 10. The mean TUR of the primary tumour on the pre-CRT PET was not significantly different between the responders (24.4 ± 12.0) and the non-responders group (19.1 ± 7.8; P = NS).
The relationships between the disease extent as indicated by CT, EUS and FDG-PET and the response rate are summarised in Table 3. A significant relationship between the extent of LN involvement as seen by PET and the CRT response rate was found (P = 0.001). If pre-CRT PET indicated an N0M0 stage the major response rate was 82%, whereas in the case of an N1M0 stage it was 33%. Five out of the six patients in whom a pathological CR was found had an N0M0 disease stage on pre-CRT PET.
|
Figure 2 illustrates the added value of pre-CRT FDG-PET to predict the CRT response rate.
|
In the subset of 25 patients in whom LN involvement was observed on the pre-CRT PET, the post-CRT PET showed a complete normalisation in 10 patients (40%) and residual FDG uptake of the lesions in 15 patients (60%).
Discordances between the PET response at the primary tumour site with the PET response of the extratumoural site were found in eight of 36 patients (22%). In four patients the discordances consisted of a major PET response of the primary tumour but a PET non-major response at the extratumoural sites. All these extratumoural lesions were located outside the radiotherapy field (bone, lungs, retroperitoneal LNs and left hypochondrium). In the four other patients a non-response of the primary tumour was associated with response at the extratumoural site. Seven of the eight patients (88%) with discordant PET responses were CRT non-major responders.
Patient-based analysis. Table 4 shows the 2 x 2 contingency table of the PET responses compared with the CRT responses. The overall accuracy of PET for the assessment of CRT response on a patient basis was 28 of 36 (78%). PET underestimated the response in four patients (11%). This was based on one false-positive PET lesion in a regional LN, and on significant residual FDG-accumulation at the primary tumour site found in three patients (deltaTUR: 59, 56 and 70%). PET overestimated the response in four patients (11%). This was due to false-negative PET findings in one peritoneal nodule, and in three microscopically invaded loco-regional LNs. Figure 2 shows a patient in whom serial FDG-PET overestimated the CRT response. The accuracy of PET in assessing CRT response was similar for adenocarcinoma (22 of 27; 81%) compared with squamous cell carcinoma (six of nine; 67%; P = NS).
|
Survival analysis
Figure 3 shows the KaplanMeier survival curves for PET major responders compared with PET non-major responders. An intention-to-treat analysis showed that the median survival time after CRT was 16.3 months compared with 6.4 months, respectively (P = 0.005). Excluding the six patients with overt progressive disease who did not undergo surgery, median survival time after CRT was 16.3 and 8.0 months, respectively (P = 0.01). Surgery-related mortality was present in three of 30 patients (10%).
|
Cox regression analysis revealed the strong prognostic influence of PET response as measured by serial FDG-PET (P = 0.002). The extent of LN involvement as seen on the pre-CRT PET was a less strong prognostic factor for overall survival (P = 0.087).
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Pre-CRT PET predicts CRT response
We found a strong correlation between the extent of LN involvement as shown by pre-CRT PET and the major response rate of patients treated with neoadjuvant CRT. If pre-CRT PET was considered as the only staging modality, a major CRT response occurred in 82% N0M0 patients, in 33% of N1M0 patients and in only 13% of the patients with distant lymphatic spread (M+ly). Importantly, this correlation between pre-CRT disease extent and response rate was not found using CT or EUS as the staging modalities. Interestingly, these results are in line with those reported recently using EUS [15]. In that study using EUS as the TNM staging modality, no correlation between disease extent as indicated by pre-CRT EUS and the CRT response was found. A pathological CR was found in nine of 24 patients (38%) with N0 and in 13 of 30 patients (43%) with N1 disease (P = NS). Thus, PET seems the only imaging modality which can identify those patients with cT4 tumours who are very likely to attain a major, or even a pathologically complete CRT response. The major reason for this predictive capacity of PET is its increased sensitivity for detection of distant LN metastases compared with CT and EUS, which we demonstrated in a previous study [8]. Indeed, in the current study, PET detected 16 distant LN metastases, compared with nine using CT and five using EUS. It is well known that the presence of these metastases severely worsens the overall prognosis of these patients, and therefore this could have contributed to the superior predictive value of PET.
Serial FDG-PET to assess the CRT response
The PET response assessment in this study was entirely based upon visual side-by-side comparison of FDG-PET images acquired before and 1 month after CRT. A semi-quantification technique (TUR) was only used to determine the degree of FDG uptake decrease corresponding to the given visual score. Given the two categories for response used in this study, more elaborate quantification techniques were not necessary. For the assessment of CRT response, the concordance between serial FDG-PET and histology was 28 of 36 (78%). The accuracy of PET in assessing CRT response was not significantly different for adenocarcinoma (81%) compared with squamous cell carcinoma (67%; P = NS).
Misclassifications of CRT-responsiveness based on PET were found in 22% of the patients and were equally based on an overestimation as on an underestimation of the response. The overestimation found was due to false-negative PET findings in the case of residual micrometastatic disease, and the underestimation of CRT response was based on false-positivity at the primary tumour site because of therapy-induced invasion of FDG-avid inflammatory and scavenging cells. Our results are in line with the results reported by Brücher et al. [12]. In that study 27 patients were studied before and 3 weeks after neoadjuvant CRT. At a threshold of 52% decrease of FDG uptake compared with baseline, sensitivity and specificity of serial FDG-PET was 100% and 55%, respectively.
Our study also showed that a complete pathological CR cannot be accurately diagnosed by FDG-PET. The sensitivity and positive predictive value of a completely normal post-CRT PET for the diagnosis of a pathological CR was only four of six (67%) and four of eight (50%), respectively. Therefore, an R0 resection is still required in those patients with a complete normalisation of the post-CRT PET.
An important and unique advantage of PET for measuring the CRT response is its capability of performing whole-body imaging and, by this, of measuring the response of all lesions together in one single examination. This is confirmed by this study. Discordances between the PET response at the primary tumour site with the PET response of the extratumoural sites were found in eight of 36 patients (22%). Importantly, seven of eight (88%) of these patients were CRT non-responders, which indicates that the least-responding PET lesion, whether it is the primary tumour or an extratumoural lesion, determines the overall responsiveness (and, therefore, prognosis).
PET response and prognosis
This study demonstrated a strong relationship between the CRT response as assessed by PET and survival: the median survival according to PET response, in an intention-to-treat analysis was 16.3 months compared with 6.4 months. By means of a Cox regression analysis it was shown that the metabolic response as measured by serial FDG-PET is a stronger prognostic factor (P = 0.002) than the extent of LN involvement seen on the pretreatment FDG-PET (P = 0.087).
The low survival time (5.9 months, starting from the last day of CRT) in the patient subgroup with LN metastases seen on the pre-CRT PET in combination with a poor CRT-response observed on PET, clearly raises questions on the value of post-CRT oesophagectomy in these patients. Similar findings have been reported recently by Brücher et al. [12]. In that study, patients with a CRT-induced decrease of the FDG uptake of >50% had a significantly shorter median survival time (8.8 months) compared with patients with an uptake decrease of >50% (22.5 months). A significant correlation between the response and survival has also been demonstrated recently using serial EUS [16]. The difference in median survival between the EUS responders compared with the EUS non-responders, however, was much smaller (17.6 months compared with 14.5 months), suggesting the marked superiority of FDG-PET in this setting.
Considerations on the future clinical application of PET in oesophageal cancer
The results of PET for preoperative staging oesophageal cancer as well as before or after neoadjuvant CRT strongly support its use as a surrogate endpoint in future trials comparing different multimodality treatment schemes. PET offers an objective and accurate means for establishing the severity of disease as well at entry into the treatment protocol (to assure balanced randomisation of patients between study arms of randomised controlled trials), as well as after chemotherapy and/or radiotherapy. Therefore, PET could be the preferred technique to compare the results of alternative treatment modalities or conflicting reports of efficacy for the same modality between different centres. Secondly, the results of this study provide a strong basis for future clinical research protocols using PET as a tool for prediction of response at an earlier stage during preoperative CRT for oesophageal cancer; for example, after one or two cycles. One preliminary report by Weber et al. [13] using quantitative FDG-PET in oesophageal cancer has indeed shown promising results in this regard, with good correla-tions between early response measurements (2 weeks after start of polychemotherapy) and histopathological responses (P = 0.001). If these exciting results are confirmed by other investigators, FDG-PET could thus predict response at a very early stage during CRT, providing a strong basis to proceed directly to immediate surgery or to definitive chemoradiotherapy in the case of non-responding disease, or to continue preoperative CRT in the case of responding disease.
The conclusions of this study are: (i) the presence and extent of LN involvement seen on the pre-CRT PET is strongly correlated to the response rate; (ii) visual comparison of PET images obtained before and 1 month after CRT allows a highly accurate estimate of the CRT response; (iii) a pathologically complete CRT response cannot accurately be diagnosed by FDG-PET; and (iv) the response assessment provided by serial FDG-PET is strongly correlated to survival. Its predictive power for survival is stronger than the pretreatment extent of LN involvement.
![]() |
Footnotes |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2. Bates BA, Detterbeck FC, Bernard SA et al. Concurrent radiation therapy and chemotherapy followed by oesophagectomy for localized oesophageal carcinoma. J Clin Oncol 1996; 14: 156163.[Abstract]
3. Flood WA, Forastiere AA. Esophageal cancer. In Benson AB III (ed.): Gastrointestinal Oncology. Kluwer Academic Publishers 1998; 140.
4.
Walsh TN, Noonan N, Hollywood D et al. A comparison of multimodal therapy and surgery for oesophageal adenocarcinoma. N Engl J Med 1996; 335: 462467.
5. Bosset J-F, Gignoux M, Triboulet J-P et al. Chemoradiotherapy followed by surgery compared with surgery alone in squamous cell cancer of the esophagus. N Engl J Med 1997; 33: 161167.
6. Zuccaro G, Rice TW, Goldblum J et al. Endoscopic ultrasound cannot determine suitability for esophagectomy after aggressive chemoradiotherapy for esophageal cancer. Am J Gastroenterol 1999; 94: 906912.[ISI][Medline]
7. Adelstein DJ, Rice TW, Becker M et al. Concurrent chemotherapy (CCT), accelerated fractionated radiation (AFR) and surgery for esophageal cancer. Proc Am Soc Clin Oncol 1996; 15: 203.
8.
Flamen P, Lerut T, Van Cutsem E et al. The utility of positron emission tomography (PET) for the staging of patients with potentially operable esophageal carcinoma. J Clin Oncol 2000; 18: 32023210.
9.
Luketich JD, Friedman DM, Weigel TL et al. Evaluation of distant metastases in esophageal cancer: 100 consecutive positron emission tomography scans. Ann Thorac Surg 1999; 68: 11331137.
10. Yeung HW, Macapinlac HA, Mazumdar M et al. FDG-PET in esophageal cancer: incremental value over computed tomography. Clin Pos Imag 1999; 2: 255260.
11. Kole AC, Plukker JT, Nieweg OE et al. Positron emission tom-ography for staging of esophageal and gastro-esophageal malignancy. Br J Cancer 1998; 9: 18631873.
12. Brücher B, Weber W, Bauer M et al. Neoadjuvant therapy of esophageal squamous cell carcinoma: response evaluation by positron emission tomography. Ann Surg 2001; 233: 300309.[ISI][Medline]
13.
Weber W, Ott K, Becker K et al. Prediction of response to preoperative chemotherapy in adenocarcinomas of the esophagogastric junction by metabolic imaging. J Clin Oncol 2001; 19: 30583065.
14. Sobin LH, Wittekind C. TNM Classification of Malignant Tumors, 5th edition. New York, NY: Wiley-Liss 1997.
15. Mallery S, DeCamp M, Bueno R et al. Pretreatment staging by endoscopic ultrasonography does not predict complete response to neoadjuvant chemoradiation in patients with esophageal carcinoma. Cancer 1999; 86: 764769.[ISI][Medline]
16. Chak A, Canto MI, Cooper GS et al. Endosonographic assessment of multimodality therapy predicts survival of esophageal carcinoma patients. Cancer 2000; 88: 17881795.[ISI][Medline]