Fax: (011) 81-27-220-8230
The incremental effect of positron emission tomography on diagnostic accuracy in the initial staging of esophageal carcinoma
Article first published online: 22 NOV 2004
Copyright © 2004 American Cancer Society
Volume 103, Issue 1, pages 148–156, 1 January 2005
How to Cite
Kato, H., Miyazaki, T., Nakajima, M., Takita, J., Kimura, H., Faried, A., Sohda, M., Fukai, Y., Masuda, N., Fukuchi, M., Manda, R., Ojima, H., Tsukada, K., Kuwano, H., Oriuchi, N. and Endo, K. (2005), The incremental effect of positron emission tomography on diagnostic accuracy in the initial staging of esophageal carcinoma. Cancer, 103: 148–156. doi: 10.1002/cncr.20724
- Issue published online: 17 DEC 2004
- Article first published online: 22 NOV 2004
- Manuscript Accepted: 9 SEP 2004
- Manuscript Revised: 31 AUG 2004
- Manuscript Received: 2 JUL 2004
- Japanese Ministry of Health, Labor, and Welfare. Grant Number: 13-18
- [18F] fluorodeoxyglucose;
- positron emission tomography;
- esophageal carcinoma;
- incremental value;
- diagnostic accuracy
The purpose of the current study was to assess whether [18F]fluorodeoxyglucose positron emission tomography (FDG-PET) provides incremental value (e.g., additional information on lymph node involvement or the presence of distant metastases) compared with computed tomography (CT) in patients with esophageal carcinoma.
The authors examined 149 consecutive patients with thoracic esophageal carcinoma. Eighty-one patients underwent radical esophagectomy without pretreatment, 17 received chemoradiotherapy followed by surgery, 3 underwent endoscopic mucosal resection, and the remaining 48 patients received definitive radiotherapy and chemotherapy. The diagnostic accuracy of FDG-PET and CT was evaluated at the time of diagnosis.
The primary tumor was visualized using FDG-PET in 119 (80%) of 149 patients. Regarding lymph node metastases, FDG-PET had 32% sensitivity, 99% specificity, and 93% accuracy for individual lymph node group evaluation and 55% sensitivity, 90% specificity, and 72% accuracy for lymph node staging evaluation. PET exhibited incremental value over CT with regard to lymph node status in 14 of 98 patients who received surgery: 6 patients with negative CT findings were eventually shown to have lymph node metastases (i.e., they had positive PET findings and a positive reference standard [RS]); 6 patients with positive CT findings were shown not to have lymph node metastases (i.e., they had negative PET findings and a negative RS); and 2 patients were shown to have cervical lymph node metastases in addition to mediastinal or abdominal lymph node metastases. Among the remaining patients, PET showed incremental value over CT with regard to distant organ metastases in six patients. The overall incremental value of PET compared with CT with regard to staging accuracy was 14% (20 of 149 patients).
FDG-PET provided incremental value over CT in the initial staging of esophageal carcinoma. At present, combined PET-CT may be the most effective method available for the preoperative staging of esophageal tumors. Cancer 2005. © 2004 American Cancer Society.
[18F]Fluorodeoxyglucose positron emission tomography (FDG-PET) provides physiologic information that enables a diagnosis of cancer to be made on the basis of altered tissue glucose metabolism.1 The role and potential value of PET as a noninvasive imaging modality have been investigated extensively in recent years.2–5 A markedly increased uptake of FDG in patients with esophageal carcinoma has been documented in several studies.6–13 FDG-PET may facilitate diagnosis by helping to differentiate between benign and malignant tumors, assess tumor extension,6–12 detect tumor recurrence,14, 15 and monitor response to therapy.16–19 In a previous study, we reported that compared with computed tomography (CT), FDG-PET was more useful in the staging of esophageal squamous cell carcinoma (SCC) and offered higher sensitivity, specificity, and accuracy rates with regard to lymph node detection, particularly in the neck and the upper thoracic region.20 Preoperative FDG-PET imaging has led to the incidental detection of unsuspected distant metastases in > 20% of patients with esophageal carcinoma.6, 7, 21, 22
Accurate preoperative staging, particularly with regard to the depth of tumor invasion, lymph node involvement, and distant metastases, is vital in determining the most appropriate procedures for curative surgery for thoracic esophageal carcinoma.23 Conventional anatomic imaging procedures, including CT, magnetic resonance imaging (MRI), and endoscopic ultrasound (EUS), have been available for the preoperative staging of esophageal carcinoma; however, recent studies have reported on the efficacy of PET (particularly its improved accuracy relative to conventional imaging modalities) in this setting.6–10, 22, 24
The goals of the current study were to compare FDG-PET results with CT findings and to evaluate the accuracy of FDG-PET relative to CT, which currently is the most commonly used conventional imaging modality. We set out to determine whether FDG-PET provided incremental value (e.g., additional information on lymph node involvement or the presence of distant metastases) over CT in patients with esophageal carcinoma and to evaluate whether any additional information would lead to changes in indications for resection or choices regarding surgical procedures.
MATERIALS AND METHODS
Eligibility Criteria and Patients
Patients with histologically confirmed primary esophageal carcinoma were eligible for the current retrospective study of clinical data from a consecutive series of patients. All patients underwent pretreatment CT and FDG-PET for tumor staging at Gunma University (Maebashi, Japan). After providing written informed consent, patients were enrolled in the study. Patients were eligible for the study if they had received a CT scan within 1 month of receiving a PET scan, had histologically confirmed primary esophageal carcinoma, and had not received previous chemotherapy or radiotherapy. Patients were excluded from the study if they had concomitant malignancies or heart disease. Between January 1999 and September 2003, 149 consecutive patients with thoracic esophageal carcinoma who were treated at the Department of General Surgical Science (Surgery I), Gunma University Graduate School of Medicine, were enrolled in the current study. Primary tumors were located in the upper (n = 28), middle (n = 75), and lower esophagus (n = 46). Their malignancies included histologically confirmed squamous cell carcinoma (n = 134), adenocarcinoma (n = 7), and other carcinomas (n = 8). The median patient age was 63.2 years (range, 36–80 years). Tumor stage and disease grade were assigned according to the TNM system. Resectability was determined by conventional staging methods, which included CT of the neck, chest, and abdomen; bone scans; MRI; EUS; and esophagography.
Treatment and Clinical Outcome
Eighty-one of the 149 patients underwent radical esophagectomy without pretreatment, 17 received chemoradiotherapy followed by surgery, 3 underwent endoscopic mucosal resection, and the remaining 48 received definitive radiotherapy and chemotherapy (Table 1). These last 48 patients were considered to have inoperable tumors due to distant organ metastases, distant lymph node metastases, or severe organ dysfunction, or else they simply chose not to undergo surgery.
|Treatment||No. of patients|
|Neoadjuvant treatment followed by surgery||17|
|Reason for nonsurgical treatment|
|Distant organ metastasis|
|Refusal of surgery||6|
Two different procedures were used. For eligible patients, a standard esophagectomy was performed according to the McKeown method (i.e., right thoracotomy followed by laparotomy and neck incision with a cervical anastomosis), and three-field (thoracoabdominal and cervical) lymph node dissection was also performed if indicated. In other patients, Ivor Lewis esophagectomy (i.e., right thoracotomy and laparotomy with anastomosis in the chest) was used, and two-field (thoracoabdominal) lymph node dissection was performed. All patients underwent curative thoracic esophagectomy that included the esophagogastric junction.20 After surgery, the lymph nodes were separated from the resected esophagus and the adjacent tissue and assigned specific numbers indicating the localization of the lymph node, in accordance with the guidelines of the Japanese Society for Esophageal Diseases (JSED).25 Surgical specimens were fixed, embedded, stained with hematoxylin and eosin, and microscopically examined by two pathologists.
After these diagnostic procedures were completed, 17 patients underwent neoadjuvant treatment consisting of concurrent radiotherapy and chemotherapy for 4 weeks. Patients received external radiotherapy at 2 grays (Gy) per fraction per day to a total dose of 40 Gy. Concurrent chemotherapy consisted of nedaplatin (or cisplatin) and 5-fluorouracil (5-FU). These 17 patients then underwent esophagectomy and regional lymph node dissection 3 weeks after receipt of the neoadjuvant treatment.
The 48 patients who did not undergo surgery received concurrent radiotherapy and chemotherapy. External radiotherapy was administered at 2 Gy per fraction per day to a total dose of 60–66 Gy. Concurrent chemotherapy consisted of nedaplatin (or cisplatin) and 5-FU, as described above.
We obtained PET images using the SET 2400 W scanner (Shimadzu Corporation, Kyoto, Japan), which has a 59.5 cm transaxial field of view and a 20 cm axial field of view. Using the simultaneous emission-transmission method, whole-body image acquisition was initiated 40 minutes after the injection of 275–370 megabecquerels (MBq) FDG.20 A total of 4–5 sections, spanning from head to thigh, were imaged for a duration of 8 minutes per section. Patients fasted for ≥ 4 hours before the FDG-PET scan was performed. The imaging protocols were approved by the institutional review board,26 and informed consent was obtained from all patients before they were examined. All PET images were evaluated qualitatively by two experienced nuclear medicine physicians (N.O. and K.E.). Functional images of standardized uptake value (SUV) were generated on the basis of attenuation-corrected transaxial images, the amount of FDG injected, the patient's body weight, and crosscalibration factors for PET and the dose calibrator. The SUV was defined as the concentration of radioactivity in the tissue or tumor specimen (MBq/mL) × patient body weight (g)/injected dose (MBq). Regions of interest (ROIs), which measured 1 cm in dimension and included the area of maximum uptake value, were drawn on images corresponding to lesions > 2 cm in dimension. If the lesion was ≤ 2 cm in dimension, the ROI was drawn to include the entire lesion, but the partial volume effect was not corrected. For primary lesions that were not visualized on PET imaging, ROIs were drawn on the corresponding area using a fusion image technique in conjunction with CT and MRI images. Similarly, for affected regional lymph nodes that were not visualized on PET, ROIs measuring 0.6 cm in dimension were drawn on the corresponding areas using the fusion image technique in conjunction with CT images. A background ROI of the same dimension as that of the lesion-based ROI was drawn on the corresponding symmetrically opposite area. If the lesion was located near the center of the body (as with the primary esophageal carcinoma), the background ROI was taken from the surrounding background area. The average value per pixel in the ROI that was used to assess the SUV was used for semiquantitative analysis. Regional lymph nodes evaluated using PET were assigned specific numbers to indicate localization, in accordance with the guidelines of the JSED. Blood glucose levels in all patients were < 100 mg/dL at the time of PET imaging.
All patients underwent helical CT scanning of the neck, chest, and abdomen. Continuous 10 mm scans were acquired from the neck to the bottom of the liver. A helical CT scanner was used, and scanning was performed before and after the injection of an intravenous contrast medium 3 minutes later. An oral contrast agent was not used. Lymph nodes were considered positive for metastatic disease if the long axis measured > 1 cm. These positive lymph nodes were assigned a specific number to indicate localization, in accordance with the guidelines of the JSED. CT hardcopy images were interpreted by two radiologists, who were blinded to PET findings.
As a reference standard (RS), distant metastases were assessed via histologic methods or clinicoradiologic follow-up for ≥ 6 months. Histologic evidence, progressively increasing tumor size, or the appearance of a new tumor was considered indicative of disease. The RS was obtained using comprehensive procedures, which included CT and MRI, ultrasonography, bone scintigraphy, FDG-PET, and specific X-ray studies. The sensitivity, specificity, and accuracy rates associated with FDG-PET and CT were calculated using the standard definitions.27 Accuracy rates for these two modalities were compared using a McNemar test.28
The primary tumor was visualized in 119 of 149 patients (80%) on FDG-PET. FDG uptake was detected in 21 of 49 patients with T1 tumors (43%), 9 of 10 patients with T2 tumors (90%), 50 of 51 patients with T3 tumors (98%), and 39 of 39 patients with T4 tumors (100%). Among the 81 patients who initially underwent surgery, FDG uptake was detected in 17 of the 40 who had pT1 tumors (43%). FDG accumulation was observed in 3 of 17 patients with pT1a tumors (remaining within the muscularis mucosae) (18%) and in 14 of 23 patients with pT1b tumors (involving the submucosa) (61%). FDG uptake rates in patients with pT2, pT3, and pT4 tumors were 83%, 97%, and 100%, respectively.
Lymph Node Status
Of the 81 patients who initially underwent surgery, 42 had histopathologically confirmed lymph node metastases. During surgery, a total of 5350 lymph nodes (1523 cervical, 2430 thoracic, and 1397 abdominal) were dissected in these 81 patients, with an average of 66 lymph nodes dissected per patient. A total of 1266 lymph node groups (according to the JSED classification) were found in these 81 patients. Histopathologically, lymph node metastases were present in 153 lymph nodes and in 115 lymph node groups. Diagnostic sensitivity, specificity, and accuracy rates for FDG-PET and CT in the detection of lymph node metastases are shown in Table 2. In the evaluation of individual lymph node groups, FDG-PET exhibited 32% sensitivity, 99% specificity, and 93% accuracy, compared with 23%, 97%, and 91%, respectively, for CT. The specificity and accuracy rates for FDG-PET were significantly higher than those associated with CT (P < 0.01 and P < 0.05, respectively). Of the 78 false-negative lymph node groups identified by PET, 32 were located in the immediate vicinity of the primary tumor, whereas the remaining 46 went undetected.
|Sensitivity (%)||Specificity (%)||Accuracy (%)|
|Lymph node group evaluation|
|PET||32 (37/115)||99 (1140/1151)a||93 (1177/1266)b|
|CT||23 (26/115)||97 (1120/1151)a||91 (1146/1266)b|
|Staging evaluation (N0/N1)|
|PET||55 (23/42)||90 (35/39)||72 (58/81)|
|CT||48 (20/42)||79 (31/39)||63 (51/81)|
In the evaluation of lymph node staging, FDG-PET exhibited 55% sensitivity, 90% specificity, and 72% accuracy, compared with 48%, 79%, and 63%, respectively, for CT (Table 2).
Results of the diagnostic evaluation of well, moderately, and poorly differentiated carcinomas by FDG-PET and CT are summarized in Table 3. In both staging and lymph node group evaluations, the sensitivity of both modalities (but especially CT) with regard to well differentiated carcinomas was low. The sensitivity of CT was lower for well differentiated carcinomas compared with poorly differentiated carcinomas (P < 0.05). In the evaluation of staging, the sensitivity of PET tended to be lower for well differentiated carcinomas compared with poorly differentiated carcinomas (P =0.07). We subsequently examined the sizes of cancer nests in metastatic lymph nodes in well, moderately, and poorly differentiated carcinomas (Fig. 1). Cancer nests in lymph node metastases were smaller for patients with well differentiated carcinomas than for patients with moderately or poorly differentiated carcinomas (P < 0.05).
|Sensitivity (%)||Specificity (%)||Accuracy (%)||Sensitivity (%)||Specificity (%)||Accuracy (%)|
|Lymph node group evaluation|
|G1||26 (5/19)||99 (271/275)||94 (276/294)||5 (1/19)a||96 (263/275)b||90 (264/294)|
|G2||25 (14/55)||99 (575/580)||93 (589/635)||24 (13/55)||98 (570/580)||92 (583/635)|
|G3||44 (18/41)||99 (294/296)||93 (312/337)||29 (12/41)||97 (287/296)||89 (299/337)|
|Staging evaluation (N0/N1)|
|G1||38 (3/8)||91 (10/11)||68 (13/19)||25 (2/8)a||64 (7/11)||47 (9/19)a|
|G2||48 (10/21)||86 (18/21)||67 (28/42)||43 (9/21)||81 (17/21)||62 (26/42)|
|G3||77 (10/13)||100 (7/7)||85 (17/20)||69 (9/13)||100 (7/7)||80 (16/20)|
Of 17 patients who underwent neoadjuvant treatment, 12 had histopathologically confirmed lymph node metastases. The results of lymph node staging using PET were consistent with those obtained using CT. However, two patients had cervical lymph node metastases that were detected by PET but were not identified by CT. These two patients had cervical lymph node metastases in addition to mediastinal or abdominal lymph node metastases.
Distant Organ Metastases
Forty-eight patients received definitive radiotherapy and chemotherapy: 11 due to the presence of extensive local tumors (T4), 25 due to the presence of distant organ metastases, and 6 due to the presence of severe organ dysfunction (Table 1). In addition, six patients chose not to undergo surgery. Of the 25 patients with distant metastases, 9 had metastases in the distant lymph nodes, 7 had liver metastases, 6 had bone metastases, and 3 had lung metastases.
Distant lymph node metastases were detected by both PET and CT in all nine patients who had them. In addition, all seven cases of liver metastases were detected correctly by both PET and CT. These cases were confirmed by subsequent imaging studies demonstrating increases in lesion size and number. Of the six patients identified as having bone metastases, two had biopsy specimens that were positive for metastatic disease, and another underwent further imaging (bone and MRI) to confirm the presence of the metastatic lesion. Only one patient's bone metastasis was detected by CT. Three patients were found to have lung metastases; CT identified lung lesions in two of these three patients, and the remaining patient was considered to have an inflammatory lesion. PET identified lung metastases in two of these three patients, whereas the remaining patient's metastasis was too small to be detected by this method.
Incremental Value of FDG-PET
PET showed incremental value over CT with regard to lymph node status in 14 of 98 patients: 6 patients with a negative CT findings were shown to have lymph node metastases (i.e., they had positive PET findings and a positive RS); 6 patients with positive CT findings were shown not to have lymph node metastases (i.e., they had negative PET findings and a negative RS); and 2 patients were shown to have cervical lymph node metastases in addition to mediastinal or abdominal lymph node metastases (Fig. 2). Therefore, the incremental value of PET relative to CT with regard to lymph node status was 14% (14 of 98 patients).
PET also showed incremental value over CT with regard to distant organ metastases in six patients. CT incorrectly identified these patients as having M0 tumors, whereas PET accurately identified them as having M1 tumors. Of these six patients, five had bone metastases, and the remaining patient had metastases to the lung.
Overall, the incremental value of PET in comparison to CT with regard to staging accuracy was 14% (20 of 149 patients).
The optimal method for assessing disease extension in patients with esophageal carcinoma is unknown. CT has represented the standard in the staging of esophageal carcinoma. CT is noninvasive and is the least costly of the available methods, but it has major limitations in terms of the accurate assessment of staging. FDG-PET imaging also is noninvasive but provides qualitatively different information compared with CT imaging, due to its reliance on the metabolic function of tumors rather than tumor size alone.1 In the current study, we compared FDG-PET results with CT findings and examined whether FDG-PET provided incremental value (e.g., additional information on lymph node involvement or the presence of distant metastases) compared with CT in patients with esophageal carcinoma.
In the current study, the primary tumor was visualized by FDG-PET in 119 of 149 patients (80%). For lymph node metastases, FDG-PET exhibited 32% sensitivity, 99% specificity, and 93% accuracy in individual lymph node group evaluation and 55% sensitivity, 90% specificity, and 72% accuracy in lymph node staging evaluation. The overall incremental value of PET relative to CT with regard to staging accuracy was 14% (20 of 149 patients).
Accurate preoperative staging, particularly with regard to depth of tumor invasion, involvement of lymph nodes, and distant metastases, is essential for determining the most appropriate treatment strategy for patients with esophageal carcinoma. Table 4 summarizes the existing literature on the diagnostic accuracy of PET in the staging of esophageal carcinoma. Primary tumor detection has been reported in 69–100% of patients with esophageal carcinoma. Some studies have reported that all patients with esophageal carcinoma had abnormal FDG accumulation in the primary tumor8, 10, 13, 32; the average detection rate in those reports was > 90%. Himeno et al.35 reported that PET imaging could detect primary esophageal tumors with an invasion status of T1b or greater (i.e., tumors involving the submucosa), whereas T1a tumors (i.e., tumors invading the muscularis mucosae) were undetectable. False-negative PET findings were always related to small (Tis or T1) tumors, suggesting limitations in the spatial resolution of the PET scanner. Furthermore, PET provides no definition of the esophageal wall or paraesophageal tissue and has no value in the assessment of tumor invasion.
|Study||Primary tumor||Lymph node metastases||Distant metastases|
|No. of patients||Detection rate (%)||Evaluation status||No. of patients||Sensitivity (%)||Specificity (%)||Accuracy (%)||No. of patients||Sensitivity (%)||Specificity (%)||Accuracy (%)|
|Block et al.6||58||97||Staging||35||52||79||63|
|Luketish et al.7||35||97||Staging||35||45||100||48||35||88||93||91|
|Flanagan et al.8||36||100||Staging||29||72||82||76||36||71||100||94|
|Kole et al.9||26||96||Staging||21||92||88||90|
|Rankin et al.10||25||100||Staging: periesophageal||19||38||90||67|
|Rankin et al.10||25||100||Staging: arteria gastrica sinistra||19||11||90||53|
|Kobori et al.11||33||72||Staging: mediastinal||33||34|
|Kobori et al.11||33||72||Staging: abdominal||79|
|Yeung et al.12||67||99||Lymph node group||67||28||99||79|
|McAteer et al.13||16||100||10||0|
|Luketish et al.21||91||69||93||84|
|Lerut et al.22||Staging: locoregional lymph nodes||42||22||91||48|
|Lerut et al.22||Staging: distant lymph node||42||77||90||86|
|Flamen et al.29||74||95||Staging||39||33||89||59|
|Meltzer et al.30||47||87||Staging||47||41||75||49||47||71||93||89|
|Choi et al.31||Lymph node group||48||57||97||86|
|Jager et al.32||16||100||12||66||12||100|
|Kim et al.33||53||96||Lymph node group||50||52||94||84||50||69|
|Kato et al.20||32||78||Staging||32||78||93||84|
|Wren et al.34||Staging||24||71||86||24||67||92|
|Himeno et al.35||31||69||Lymph node group||31||42||100||92|
|Kneist et al.36||Staging||58||6–42||94–100||59–82||58||35||87|
|Rasanen et al.37||42||83||Staging||42||37||100||63||42||47||89||74|
In the evaluation of lymph node metastases, the sensitivity of PET was lower than its specificity (Table 4). The median sensitivity rate for for lymph node metastases was 57% (range, 0–92%), whereas the median specificity rate was 90% (range, 75–100%). When the results were subjected to metaanalysis using the Mantel–Haenszel method, the synthesized odds ratio (OR) was 7.31 (95% confidence interval [CI], 4.27–12.54).
Many studies have demonstrated that the diagnostic sensitivity, specificity, and accuracy rates for PET in the detection of individual lymph node metastases are significantly better compared with the corresponding rates for CT.6–8, 10, 31 This finding is attributable to the significant ways in which PET and CT differ in terms of their methods of tumor detection. Because the evaluation of metastatic infiltration of lymph nodes is based on CT imaging measurements of lymph node size, normal-sized lymph nodes are not detected by CT. In contrast, PET can detect metastases in normal-sized lymph nodes through functional imaging and can verify the presence of malignant disease in enlarged lymph nodes. Therefore, the diagnostic accuracy of PET was expected to be significantly better than that of CT. In the current study, 32 false-negative lymph node groups identified on PET were located in the immediate vicinity of the primary tumor, and the remaining 46 lymph node groups went undetected by PET because of small metastatic tumor size, artifacts of gastric peristalsis, or physiologic FDG uptake. In summary, PET cannot distinguish adjacent lymph node metastases from primary tumors because of poor spatial resolution and cannot detect metastatic tumors measuring < 5 mm in diameter because of a partial volume effect.20
The sensitivity of CT was lower for well differentiated carcinomas than for poorly differentiated tumors, as cancer nests of lymph node metastases were smaller in patients with the former than in patients with the latter. Furthermore, in patients with poorly differentiated carcinomas, tumors may be more metabolically active, resulting in better detection.
The median sensitivity with regard to diagnostic accuracy for distant organ metastases was 69% (range, 35–100%), whereas the median specificity was 93% (range, 87–100%; Table 4). When the results were subjected to metaanalysis using the Mantel–Haenszel method, the synthesized OR was 23.18 (95% CI, 9.19–58.49). These results cannot be compared with the findings of the current study, due to differences in the methods used and particularly due to the large number of distant metastases lacking histopathologic confirmation. Luketich et al.21 reported that PET yielded significantly better results compared with CT in the diagnosis of distant metastases, and they also found that all false-negative metastatic tumors measured < 1 cm in diameter. Flanagan et al.8 reported a similar trend with regard to the detection of liver metastases. Nonetheless, liver metastases measuring < 1 cm in diameter have proven to be undetectable.38, 39
The additional value of PET as a noninvasive imaging modality for esophageal carcinoma has been investigated extensively in recent years. In the current study, PET was 14% more informative in the staging of esophageal carcinoma compared with CT. Flanagan et al.8 found that PET findings changed the management strategy significantly in 17% of all patients with esophageal carcinoma (6 of 36). Furthermore, FDG-PET provided additional information compared with CT, affecting management in 14% of patients.12 Jager et al.32 reported that FDG-PET led to the upstaging of disease in 17% of patients (3 of 18) and might have helped to prevent the performance of unnecessary surgical procedures. In other studies, the use of preoperative FDG-PET has led to the incidental detection of unsuspected distant metastases in > 20% of patients with esophageal carcinoma.21, 22 In addition, FDG-PET has been found to detect radiographically occult distant metastases in 10–20% of patients with esophageal carcinoma.9, 21
There are some disadvantages associated with PET imaging. For example, small, early-stage tumors may go undetected, because partial-volume effects result in a falsely low measurement of true FDG activity, as was indicated in our previous report.20 In addition, PET scanners have poorer spatial resolution compared with CT scanners.40 Therefore, PET cannot distinguish between adjacent lymph node metastases and primary tumors. A third drawback of PET is that FDG often accumulates in areas of inflammation. Combined PET-CT, which eliminates the former two disadvantages associated with PET imaging, is widely available and is being used with increasing frequency in clinical practice.41
In conclusion, FDG-PET provides incremental value over CT in the initial staging of esophageal carcinomas. Combined PET-CT may be the most effective method currently available for the preoperative staging of esophageal tumors.
The authors thank Akie Nakabayashi, Hideko Emura, Miwako Ohnuma, Sachiko Ueno, Tomoko Ogasawara, and Yukie Saitoh for their excellent secretarial assistance and Midori Ohno for her assistance with data management and biostatistical analysis.
- 25Japanese Society for Esophageal Diseases. [Guidelines for the clinical and pathological studies on carcinoma of the esophagus (9th edition)]. Tokyo: Kanehara, 1999.