Metastases in the thoracic (TH 9/TH 12) and lumbar spine (L1) diagnosed with magnetic resonance imaging 3 months after transhiatal esophagus resection (adenocarcinoma stage, pT3 pN1 [4 of 17 lymph nodes involved] cM0 G4 R0).
Kneist W, Schreckenberger M, Bartenstein P, Menzel C, Oberholzer K, Junginger T. Prospective Evaluation of Positron Emission Tomography in the Preoperative Staging of Esophageal Carcinoma. Arch Surg. 2004;139(10):1043-1049. doi:10.1001/archsurg.139.10.1043
Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004
Positron emission tomography (PET) is a useful tool in the selection of patients with esophageal cancer who may not benefit from esophageal resection.
Tertiary care hospital.
Eighty-one patients with newly diagnosed esophageal cancer who underwent PET and computer tomography (CT) of the chest and abdomen (and of the neck in 45 patients) within 45 days were included.
Main Outcome Measures
We calculated the sensitivity and specificity in detecting metastatic sites on the basis of 31 histologically verified lesions. In addition to results obtained on CT, the information provided by PET was evaluated with a view to the choice of management strategies.
The PET findings had a higher specificity (89% vs 11%) but a lower sensitivity (38% vs 63%) than CT findings in the detection of metastatic sites. The CT results showed greater agreement with histopathological findings than did PET results. In 8 patients (10%), PET detected distant metastases that were not identified with CT. In 4 patients (5%), PET detected bone metastases only, but in all of these patients metastases in other locations were detected by CT. Although PET led to upstaging (M1) in 2 patients (2%), it did not enable the exclusion of esophageal resection.
Preoperative PET was not characterized by greater accuracy in the detection of metastatic sites previously identified by CT. Therefore, PET did not lead to a change in the indication for esophagectomy. An increase in the sensitivity and the combined use of CT and PET may lead to new indications for this staging procedure.
Positron emission tomography (PET) is a noninvasive diagnostic procedure with a high specificity rate in the demonstration of metabolically active tumor tissue in the entire body. 1- 3 The preoperative diagnosis of distant metastases in patients with esophageal carcinoma using PET enabled the identification of distant metastases that were not detected with standard computed tomography (CT).4- 13 Positron emission tomography has therefore been suggested as the most appropriate modality in the selection of patients for resection therapy.1,4,5,10,11 Many of these studies, however, have reported varying sensitivity rates for PET, thus producing significant differences in the number of newly detected distant metastatic sites (3%-28%).4,6,7,9,13 Whether PET enables an early diagnosis of distant metastases or detects only additional metastases in patients with advanced tumor disease and therefore does not exert an influence on the choice of the therapeutic strategy remains open to question. In this prospective study of patients with esophageal carcinoma, standardized preoperative CT was therefore used with PET, and a histopathological examination of the detected distant metastases was performed where possible to assess the accuracy of PET in the diagnosis of distant metastases and to determine the validity of the imaging technique in the selection of patients for operative therapy.
From January 21, 2001, to November 8, 2002, 90 consecutive patients with biopsy-proved esophageal carcinoma underwent CT of the thorax and abdomen for preoperative tumor staging at the Department of General and Abdominal Surgery at the Hospital of the Johannes Gutenberg-University, Mainz, Germany; in addition, CT of the neck was performed in 45 patients. All patients underwent PET with 18F-fluorodeoxyglucose (18F-FDG) injection. Two patients undergoing neoadjuvant chemotherapy were excluded from the study, as were 5 with intraoperative or postoperative classification of types II (cardia carcinoma) and III (subcardial carcinoma) carcinoma of the gastroesophageal junction14 and 2 with an interval of more than 45 days between the diagnostic measures. A total of 81 patients (41 with adenocarcinoma and 40 with squamous cell carcinoma) were thus eligible for inclusion in the study (Table 1).
Thirty-one patients underwent transhiatal esophagectomy with lymph node dissection in the abdomen and the lower and posterior mediastinum, and 23 underwent transthoracic resection with abdominal and mediastinal lymph node dissection. Contraindications for surgical therapy included insufficient vital organ function (n = 5), a locally nonresectable tumor (n = 9), distant metastases (n = 13), and a combination of these abnormalities.
Positron emission tomography was performed in 56 patients using an ECAT Exact PET scanner (CTI/Siemens, Knoxville, Tenn, and Erlangen, Germany) at the Department of Nuclear Medicine of the Johannes Gutenberg-University Hospital, and in the remaining 25 patients at the Johann Wolfgang Goethe-University, Frankfurt am Main, Germany, with the same model scanner. The PET scanner used has a 16.2-cm axial field of vision consisting of 47 image planes with a section thickness of 3.375 mm and a transaxial full width at half-maximum of 6 mm at the center of the visual field in the 2- (2-D) and 3-dimensional (3-D) modes. The patients fasted for a minimum of 6 hours, and blood glucose measurements were taken in 79 before the injection of 7.5 to 10.0 mCi (278-369 MBq) (mean ± SD, 9.1 ± 0.9 mCi [337 ± 35 MBq]) of intravenous 18F-FDG for the 2-D acquisition mode, and in 2 patients, before injection of 5.3 and 5.7 mCi (197 and 212 MBq) of intravenous 18F-FDG for the 3-D acquisition mode. To minimize radioactivity in the bladder and kidneys during PET image acquisition, each patient received intravenous furosemide, 20 mg, and an infusion of 500 mL of Ringer solution. Two patients with a blood glucose level of greater than 150 mg/dL (>8.3 mmol/L) received an initial intravenous injection of 2 to 6 IU of insulin 30 minutes before the injection of 18F-FDG; blood glucose concentrations were then checked at 10-minute intervals.
One hour after the 18F-FDG injection, PET was performed in all patients as a whole body scan (from the base of the skull to the proximal thigh) with 4 to 6 bed positions using the 2-D acquisition mode in all except 2 patients, in whom the 3-D acquisition mode was applied. The acquisition time per bed position in the 2-D mode was 7 minutes for the emission and 3 minutes for the transmission scan, ranging 5 and 7 minutes, respectively, for the 3-D scan. The images were reconstructed using an iterative (OSEM [ordered subsets expectation maximization]) algorithm in a 128 × 128 matrix and displayed as attenuation-corrected transaxial, coronary, and sagittal ultrasonic images. The PET images were evaluated independently by 2 experienced nuclear clinicians (M.S., P.B., and C.M.) blinded to each other's findings. In the presence of conflicting diagnoses, the opinion of a third specialist was obtained to reach a consensus. The clinicians evaluating the PET images had no knowledge of the previous CT findings.
In the qualitative assessment of the images (with consideration of the normal biodistribution of the tracer), foci with increased 18F-FDG uptake compared with the surrounding structures and tissues were defined as malignancies. In addition, a quantitative analysis was performed, based on the calculation of the standardized uptake value (SUV). An SUV of greater than 2.0 was defined as an additional quantitative malignancy criterion.
Uptake of 18F-FDG in adenocarcinomas and squamous cell carcinomas was assessed and compared with the maximum and median standardized uptake values.
Computed tomography was performed in 62 patients according to a standardized protocol at the Department of Radiology, Hospital of the Johannes Gutenberg-University, with a single-section or multisection Siemens spiral CT scanner. In 8 patients, preoperative staging was supplemented with a CT scan of the thorax or abdomen, whereas 10 of the patients had previously undergone these examinations at their respective referral hospitals. After nonenhanced spiral CT of the liver, intravenous injection of 120 mL of the contrast agent, reinjection of sodium chloride with a high-pressure injector, and imaging of the thorax and epigastrium during the arterial contrast agent phase, we performed scanning of the epigastric region during the portal venous phase with a second spiral CT.
A section thickness of 5 mm was used for all images. Size, number, and localization of the lymph nodes served as criteria in the assessment of the CT images with a view to thoracic or abdominal lymph node metastasis of known esophageal carcinoma.
Intrathoracic lymph nodes with a diameter of more than 6 mm or smaller lymph nodes occurring in clusters of more than 3 were considered to be positive for metastasis. Azygos or subcarinal lymph nodes represented an exception regarding size. For these lymph nodes, a size of 10 mm was determined to be normal. Lymph nodes of the lesser curvature and the celiac trunk of larger than 6 mm or clusters consisting of more than 3 smaller lymph nodes were defined as metastases.15- 18
The evaluation of the spiral CT images (slice thickness range, 3-10 mm) provided by the referral institution was based on the accompanying written diagnosis, in addition to a reassessment made on-site.
The median interval from CT to PET was 5 days (range, 0-32 days); from CT to surgery, 15 days (range, 1-45 days); and from PET to surgery, 8 days (range, 1-42 days).
The tumor stage was determined in accordance with the current TNM classification and based on the histopathological findings (pTNM). Regional (cervical, celiac, M1a) and nonregional lymph node metastases and organic involvement (M1b) were described as distant metastases.19 The findings of the histopathological examination for 31 suspicious lesions from 28 patients served as the basis for the diagnosis of distant metastasis. Distant lymph nodes were interpreted from pathological resection specimens. Preoperative fine-needle aspiration biopsy was not used.
After confirmation of distant metastasis in an organ, additional lymph nodes detected on imaging and suspected to be metastatic were also classified as distant metastases.
Changes in clinical staging and the therapeutic procedure as a result of the PET findings were recorded for all patients. The sensitivity, specificity, and accuracy of PET and CT were calculated by analyzing the lesions confirmed on results of histopathological examination and separately for adenocarcinomas and squamous cell carcinomas.
The secondary malignancies diagnosed on CT and PET were recorded and the tumor status was determined.
The sensitivity, specificity, and accuracy of PET and CT were calculated. The exact confidence intervals were indicated for sensitivity and specificity. A McNemar test was used to analyze significant increases in the number of false-positive or false-negative findings. In addition, a κ value was used to describe the correlation between histological findings and the 2 imaging techniques and between the imaging techniques with consideration of the random agreement. In case of complete agreement, the maximum κ can be a value of 1.00, ranging from 0.00 for random probability. Directions for interpretation have been provided by Landis and Koch.20 We performed statistical analysis with the Statistical Package for Social Sciences program (SPSS Inc, Chicago, Ill).
In 81 patients, no significant differences were noted between adenocarcinomas and squamous cell carcinomas as to 18F-FDG uptake in the primary tumor (mean ± SD SUVmax, 10.7 ± 10.4 vs 10.0 ± 5.3, P = .55; SUVmean, 5.4 ± 4.4 vs 4.0 ± 2.5, P = .08).
Preoperatively, distant metastasis was diagnosed in 38 patients (adenocarcinoma, 18 patients [43.9%]; squamous cell carcinoma, 20 [50%]) by means of imaging results. Lymphogenic distant metastases were detected in 24 patients; organ metastases were detected in 7; and lymphogenic and hematogenic distant metastases were suspected in 7. The CT images demonstrated distant metastases (cM1) in 36 (44%); PET images, in 20 patients (25%) (κ = 0.47; P<.001). Two patients (3%) underwent upstaging (from cM0 to cM1) and 18 (22%) underwent downstaging (from cM1 to cM0) after the addition of PET.
Histopathological examination demonstrated distant metastasis (pM1) in 20 of the 81 patients (adenocarcinoma in 12 and squamous cell carcinoma in 8). In 16 patients, this finding was in agreement with preoperative CT or PET results. Distant metastases were missed by CT and PET in 4 patients (peritoneal carcinosis in 1, truncal in 2, and cervical lymph node metastasis in 1). In 1 patient with a false-positive finding of truncal lymph node involvement with CT, a positive PET finding (cervical metastasis also diagnosed with CT) was not confirmed on histopathological findings. In 7 other patients, the clinical suspicion of distant metastasis (cM1) was not confirmed by histopathological examination.
We performed a histopathological examination for 31 lesions in 28 patients (Table 2 and Table 3). The sensitivity rate of PET imaging ranged from 38% and was lower than that of CT (63%), whereas the specificity of PET was higher (89% vs 11%). A higher value was calculated for the accuracy of PET imaging (52% vs 45%). The correlation between the histological findings and CT results was poor (κ<0.20) because of the lower sensitivity rate of PET imaging. There was no significant difference between the sensitivity for lymphogenic and distant metastases at other locations. In addition, the separate examination of both tumor types yielded a higher specificity for PET in the diagnosis of distant metastases and a higher sensitivity for CT, at a similar correlation with the histopathological finding (κ<0.20). The accuracy of PET was higher than that of CT for the detection of squamous cell carcinoma (56% vs 43%).
In 20 patients, CT scanning detected 31 additional lesions with the suspicion of distant metastases, although no histopathological evidence was obtained for these lesions. Distant metastases were previously demonstrated at different locations in 7 of these 20 patients. Seventeen of the 31 lesions were also described as distant metastases with PET.
Bone metastases were detected in 5 (6%) of 81 patients on PET imaging. Positron emission tomography led to the diagnosis of bone metastasis in 4 symptom-free patients. The diagnosis was proved by results of subsequent histopathological examination in 2 patients and in 3 patients using other imaging techniques (magnetic resonance tomography and scintigraphy). Four of these patients had an adenocarcinoma (4/41), whereas squamous cell carcinoma of the esophagus was present in 1 patient (1/40). Additional metastatic sites were identified on CT in all 5 patients.
In 8 (10%) of the 81 patients, PET identified the presence of distant metastases that were missed on CT (uptake in cervical lymph nodes, n = 4; with CT of the neck, n = 2; with CT of the bone, n = 4; with CT of the liver, n = 2; and with CT of the lower abdomen, n = 1) (Table 4). Distant metastasis (lower abdomen) was excluded in 1 patient with a benign rectal adenoma. Histopathological evidence of distant metastasis was available in 6 patients, and further metastases were detected at various locations in addition to the locations demonstrated by PET findings; therefore PET did not lead to a change in M staging in these patients. The PET finding (uptake in cervical lymph nodes and/or liver) could not be confirmed by other techniques in 2 patients. In 1 patient, the increased cervical uptake has remained unchanged for longer than 1 year, enabling the exclusion of the presence of distant metastasis with a high degree of certainty. The PET findings thus led to a change in tumor stage in only 1 patient (from cM0 to cM1).
Secondary tumors were identified in 4 (5%) of 81 patients. In 3 patients (4%), these tumors were diagnosed exclusively on PET. In all patients, the finding led to therapeutic consequences. The PET findings demonstrated a lesion with suspected malignancy in the rectal region of 1 patient with adenocarcinoma of the distal esophagus. An adenoma was diagnosed after transanal endoscopic microsurgery. Colonic cancer was diagnosed and resected (pT3 pN0 M0 G2 R0) in a second patient with distal adenocarcinoma. Two patients with squamous cell carcinoma had additional esophageal carcinomas, which were identified after exploratory thoracotomy in one patient and detected on CT findings in the other. Furthermore, in one of these patients, PET findings demonstrated a carcinoma of the epiglottis, which was removed by local excision.
In 2 patients, PET findings resulted in tumor upstaging based on the demonstration of increased cervical (n = 1) and hepatic (n = 1) uptake, which was not supported by histopathological findings. One of these patients was a 76-year-old woman with distal Barrett esophageal carcinoma (pT1) who underwent transhiatal resection without cervical lymph node resection in view of her advanced age. One year after surgery, the patient remained asymptomatic. Positron emission tomography findings confirmed an unchanged cervical uptake. The second patient (with increased uptake in the liver) was medically unfit for surgery because of impaired cardiac function. Therefore, the indication for surgery was not influenced in these patients by PET-provided information. Independent of the demonstration of distant metastatic sites, the information provided by PET on cardiac function (cardiomyopathy in 1 patient) served as the basis for the exclusion from surgical therapy.
In addition to the assessment of the general state of health and resectability of the primary tumor, the absence or the presence of distant metastases is a decisive factor in the selection of patients with esophageal carcinoma for surgical therapy. Positron emission tomography also enables the demonstration of tumor manifestations in organs other than the abdomen and thorax visualized by CT, and therefore has been suggested as a useful tool in the selection of patients for operative therapy.1,4,5,10,11 In the evaluation of the statements made by those authors, consideration needs to be given to the fact that most of these investigations are retrospective studies with small patient numbers, which are characterized by differences in the quality of preoperative diagnosis, the absence of precise information on the incidence of biopsy-proved distant metastases, or therapeutic consequences resulting from PET findings (Table 5).
In the present study including 90 consecutive patients with esophageal carcinoma undergoing therapy during the study period, standardized preoperative diagnostic CT was used with the addition of PET. The investigators were blinded to the results obtained via the respective other examination procedures.
Previous studies7,10,11,21 considered patients with carcinoma of the esophagogastric junction (Table 5). Beside adenocarcinomas of the distal esophagus (type I), they may include patients with carcinoma of the cardia and the upper third of the stomach (type II and III) that, owing to different tumor biology and metastatic patterns, might exert an influence on the findings of the imaging techniques regarding esophageal carcinomas.22 To eliminate possible bias, we excluded patients with true carcinoma of the cardia and subcardial carcinoma from our study.
Histopathological evidence of distant metastasis was obtained in 20 patients, and in 15 the diagnosis was based on the characteristic findings of the described imaging techniques.
Positron emission tomography detected the presence of distant metastatic disease missed on CT in 8 (10%) of the 81 patients. A total of 19 lesions were visualized with PET in this group. The most frequent finding was increased uptake in the skeleton (n = 4), liver (n = 4), and cervical lymph nodes (n = 6).
The number of bone metastases (6%) detected on PET is consistent with results reported by other studies (1.9%-8.8%).5,6,8,11,13,21 Positron emission tomography did not demonstrate early and isolated bone metastasis in our patient population, but detected metastases in patients with additional distant metastasis in other organs. The limited ability of PET in the early detection of bone metastasis was confirmed by the clinical course of 1 patient with a normal preoperative PET finding, who exhibited bone metastasis 3 months after operative therapy (Figure 1).
A further advantage of PET is the visualization of cervical lymph node metastases missed on CT. Detection rates reported in the literature range from 7.7%7 to 14%.4 The detection rate depends on the quality and extent of CT. Computed tomography of the neck is the only procedure that enables a significant statement on cervical lymph node involvement. A search of the available literature showed only 1 study in which CT of the neck was performed in addition to CT of the thorax.12 In our patient population, CT of the neck was performed in 45 patients, and normal findings were obtained in 33, whereas in 2 patients increased uptake was detected in the cervical lymph nodes. However, no positive evidence of metastasis was detected in the clinical course of 1 of the patients. Positron emission tomography therefore detected cervical lymph node involvement in only 1 patient with a normal CT finding (3%).
Positron emission tomography led to upstaging after the demonstration of previously missed distant metastases in only 1 (1%) of 81 patients. In the remaining patients, the CT scan enabled visualization of distant metastatic sites, whereas the PET image merely provided supplementary information on the presence of additional distant metastases.
The relatively small percentage of distant metastases detected exclusively by PET in our patient population is a result of the low sensitivity rate of the examination (38%). A number of other studies have reported higher sensitivity rates.4,8,9,11,13 Most of these studies, however, have not differentiated between clinically suspected and histologically proved metastases8,9,11,13 and have included results of the clinical course, without providing detailed information on the retrospective confirmation of the PET finding.8,9,11,13 The results reported by Kim et al12 and Lerut et al10 are based on histologically proved lymphogenic distant metastases; the authors calculated sensitivity rates of 69% and 77%, respectively, in small sample sizes (50 and 42 patients, respectively). Data on the sensitivity of biopsy-proved distant metastases are yet to be published. In our patient population, 5 of 14 lymph node metastases (M1ly; distant disease) and 3 of 6 distant metastases in other organs were detected on PET images. Although these findings have been obtained in a small number of patients, they nevertheless validate the values recorded for biopsy-proved lymph node metastases.5,6,8,10,23
A possible cause of the low sensitivity rate may be an excessively high borderline value. To exclude this possibility, the SUV of patients with a negative finding was lowered from 2.0 to 1.5 at the follow-up examination. This did not lead to a higher detection rate, owing to fact that the described borderline value no longer allowed the differentiation between unspecific tracer uptake and metastases. Further trials are required to determine whether the differences in tumor biological behavior might be responsible for the differences in the sensitivity rates recorded for PET, which have previously been observed for other tumors.
Because the use of a predetermined lymph node size threshold (10 mm in diameter) as the rigid diagnostic criterion for nodal metastasis at CT may have been problematic,15- 18,24,25 we used a smaller cutoff size to define involved abdominal lymph nodes. This probably explains, in part, why specificity for the detection of distant metastatic sites with CT is so low and sensivitivity is higher compared with those of other studies.
The results of the present study have confirmed the high specificity of PET in the demonstration of lymphogenic and hematogenic distant metastases. In addition, the identification of secondary tumors in 4 patients (5%) on PET images only is in accordance with results reported in the literature. Eight secondary tumors (1.6%) were detected in 11 series (Table 5) that included a total of 499 patients. An additional carcinoma was identified in 5 of the 8 patients and 3 of 4 patients in our own patient population. Although the identification of the secondary tumor was of therapeutic consequence, it did not influence the decision about the therapeutic procedure for esophageal carcinoma, and the demonstration of metastases with PET in our own patient population did not have an effect on the indication for surgery.
The results of this study confirm that PET is able to identify distant metastases missed on CT and represents a valuable aid in the demonstration of secondary tumors (1.6%) in preoperative staging of esophageal cancer. However, most of the patients in this study had advanced tumor disease with additional distant metastases in other organs. The incidence of distant metastases demonstrated exclusively by PET was low (1/81). Accordingly, the results of PET imaging did not exert an influence on the therapeutic procedure. In addition, because of the increased financial costs the prerequisite for the use of PET in the selection of patients with esophageal carcinoma for surgical therapy is an improvement in the sensitivity of the method in the early diagnosis of distant metastases.
Correspondence: Werner Kneist, MD, Department of General and Visceral Surgery, Hospital of the Johannes Gutenberg-University, Langenbeckstraße 1, 55131 Mainz, Germany (email@example.com).
Accepted for publication February 18, 2004.