Metastasis to Regional Lymph Nodes in Patients with Esophageal ...

Radiology

Nuclear Medicine Young Cheol Yoon, MD Kyung Soo Lee, MD Young Mog Shim, MD Byung-Tae Kim, MD Kwhanmien Kim, MD Tae Sung Kim, MD

Index terms: Esophagus, neoplasms, 71.321 Lymphatic system, CT, 993.12912, 995.12912, 996.12912, 997.12912 Neoplasms, metastases, 993.33, 995.33, 996.33, 997.33 Positron emission tomography (PET) Positron emission tomography (PET), comparative studies Published online 10.1148/radiol.2281020423 Radiology 2003; 227:764 –770 Abbreviation: FDG ⫽ fluorodeoxyglucose 1

From the Departments of Radiology (Y.C.Y., K.S.L., T.S.K.), Thoracic Surgery (Y.M.S., K.K.), and Nuclear Medicine (B.T.K.), Samsung Medical Center, Sungkyunkwan University School of Medicine, 50, Ilwon-Dong, KangnamKu, Seoul 135-710, Korea. Received April 15, 2002; revision requested June 19; final revision received October 17; accepted November 5. Address correspondence to K.S.L. (e-mail: kslee@smc .samsung.co.kr).

Metastasis to Regional Lymph Nodes in Patients with Esophageal Squamous Cell Carcinoma: CT versus FDG PET for Presurgical Detection— Prospective Study1 PURPOSE: To prospectively compare the accuracy of fluorine 18 fluorodeoxyglucose (FDG) positron emission tomography (PET) and computed tomography (CT) for detection of primary tumor and metastasis to individual lymph node groups and for nodal staging. MATERIALS AND METHODS: From February 2000 to July 2001, 81 patients with squamous cell carcinoma of the esophagus (78 men and three women; age range, 31–90 years; mean age, 63 years) underwent CT and FDG PET before esophagectomy and lymph node dissection. During surgery, all visible and palpable lymph nodes in the surgical fields were removed. The accuracies of CT and FDG PET for depiction of metastasis to lymph nodes were compared. RESULTS: For depiction of malignant nodal groups in each lymph node group, the sensitivity, specificity, and accuracy, respectively, of CT were 11% (11 of 96 nodal groups), 95% (553 of 581), and 83% (564 of 677), whereas those of FDG PET were 30% (29 of 96), 90% (525 of 581), and 82% (554 of 677) (P values: ⬍.001, .009, and .382, respectively). Twenty-eight false-positive interpretations were rendered at CT in evaluations of 11 mediastinal, four hilar, and 13 abdominal nodal groups, and 56 false-positive interpretations were rendered at FDG PET in evaluations of 23 mediastinal, 32 hilar, and one abdominal nodal group. CONCLUSION: FDG PET is more sensitive than CT for depicting nodal metastases in patients with squamous cell carcinoma of the esophagus. FDG PET is slightly less specific than CT for depicting metastases, but the difference in specificity between the two modalities is statistically significant. Both FDG PET and CT have low sensitivity for depicting nodal metastasis. The relatively low specificity of FDG PET for depiction of nodal metastasis compared with that of CT is caused mainly by a high rate of false-positive hilar node interpretations. ©

Author contributions: Guarantors of integrity of entire study, K.S.L., B.T.K.; study concepts, K.S.L.; study design, K.S.L.; literature research, Y.C.Y.; clinical studies, Y.C.Y., Y.M.S., B.T.K.; data acquisition, Y.C.Y., Y.M.S.; data analysis/interpretation, Y.C.Y., K.S.L.; statistical analysis, Y.C.Y., K.S.L.; manuscript preparation, Y.C.Y.; manuscript definition of intellectual content, K.S.L.; manuscript editing, Y.C.Y.; manuscript revision/review and final version approval, all authors. ©

RSNA, 2003

764

RSNA, 2003

Patients with esophageal carcinoma have a relatively poor prognosis, with a 5-year survival rate of 6%–11% (1). However, patients do not have a uniformly poor prognosis. In general, patients with early-stage squamous cell carcinoma have a relatively good prognosis; excellent 5-year survival rates in the range of 57%–78% have been reported in such patients who undergo surgical treatment (1). Patients with advanced but potentially resectable esophageal cancer are generally treated with surgery and some form of neoadjuvant chemotherapy, radiation therapy, or both, with a resultant 5-year survival rate of 20%–30% (2). As with any other malignancy, optimal treatment of patients with esophageal cancer depends on accurate staging (1,3,4). Patients with limited disease progression and early-

Radiology

stage tumors benefit from complete surgical resection, whereas those with locally advanced disease have a poor prognosis in spite of aggressive attempts at resection, and those with distant metastatic disease are considered to have incurable cancer (3,5). Therefore, the determination of the resectability and the stage of disease is important in selecting an appropriate treatment and evaluating a patient’s response to initial treatment (5). Imaging modalities used in esophageal cancer staging include computed tomography (CT), endoscopic ultrasonography (US), fluorine 18 fluorodeoxyglucose (FDG) positron emission tomography (PET), and techniques that involve minimally invasive surgery, such as laparoscopy and thoracoscopy (6). CT has been widely used for preoperative evaluation, but it is known to be nonsensitive for the identification of transmural spread and the detection of metastases to lymph nodes (1,7–12). The accuracy of endoscopic US for estimating the depth of penetration of the primary tumor has been validated, but endoscopic US has been shown to be inaccurate in the evaluation of nodal status (1,13,14). Recently, a combination of thoracoscopy and laparoscopy has been introduced for detection of regional and distant metastases, and high accuracy rates (93% for thoracoscopy and 94% for laparoscopy) have been achieved. However, these procedures are invasive (1,2). FDG PET has been reported to be more sensitive than CT in the detection of primary tumor and distant metastases (12,15,16). The functional images of FDG PET are not only complementary to the images obtained with more traditional modalities but may be more sensitive because alterations in tissue metabolism generally precede an anatomic change (3). There have been several clinical reports regarding the efficacy of FDG PET in the preoperative evaluation of esophageal cancer (5,9,11,13,16). However, the role of PET in the detection of nodal metastasis is still controversial (11–13,16 – 18). The purpose of our study was to prospectively compare the accuracy of FDG PET and CT for the detection of primary tumor and metastases to individual lymph node groups and for nodal staging.

MATERIALS AND METHODS Patient Population From February 2000 to July 2001, 136 consecutive patients with biopsy-proved Volume 227

䡠

Number 3

squamous cell carcinoma of the esophagus were seen in our tertiary referral hospital. For prospective investigation, both chest CT and whole-body FDG PET examinations were routinely performed. This study protocol received approval from the institutional review board of our institution. In addition, informed consent was obtained from each patient before he or she was enrolled in this study. Of 136 patients with squamous cell carcinoma, 36 did not undergo esophagectomy. Twelve of the 36 patients had surgically resectable disease but refused surgery. The remaining 24 patients could not undergo surgery. Nineteen patients had distant metastatic lesions—in the liver in six patients, in nonregional lymph nodes (classified as M1b according to the TNM system [19]) in six patients, in the lung in three, in the adrenal glands in two, in the kidney in one, and in bone in one. Three patients had evidence of direct invasion to adjacent organs (trachea, left main bronchus, or left atrium), and two patients had other advanced primary cancers (gastric or pancreatic cancer). These patients underwent chemotherapy (n ⫽ 21), radiation therapy (n ⫽ 6), or both (n ⫽ 9). One hundred patients underwent surgery. Of these patients, 12 who had been treated with preoperative neoadjuvant chemotherapy (n ⫽ 3) or concurrent chemoradiation therapy (n ⫽ 9) before surgery were excluded. At CT, seven of these 12 patients were observed to have extensive metastasis to regional lymph nodes with extracapsular invasion, three had evidence of direct invasion to the adjacent organs (tracheal invasion in two and aortic invasion in one), and two had nonregional M1b lymph node metastasis. We also excluded seven patients who underwent esophagectomy without mediastinal lymph node dissection. Therefore, this study included 81 patients. There were 78 men and three women, who ranged in age from 31 to 90 years (mean age, 63 years). The interval between CT and surgery ranged from 4 to 45 days, with a mean of 17 days. The interval between PET and surgery ranged from 1 to 40 days, with a mean of 12 days. The interval between CT and PET ranged from 0 to 29 days, with a mean of 6 days. In our institution, all patients without metastasis to distant organs or definite direct tumor invasion of adjacent organs at imaging are routinely scheduled for esophageal resection and extensive regional lymph node dissection. We did not consider N1 or M1a (ie,

metastasis to cervical lymph nodes in patients with upper esophageal cancer and metastasis to celiac lymph nodes in patients with lower esophageal cancer) cancer to be a contraindication to surgery.

CT Scanning CT scanning was performed with a HiSpeed Advantage scanner (GE Medical Systems, Milwaukee, Wis). Helical CT scans were obtained from the neck to the middle portion of both kidneys with 7-mm collimation and a pitch of 1.3. The scans were obtained after intravenous injection of a total of 100 mL of a contrast medium (iopamidol, Iopamiron 300; Bracco, Milan, Italy) at a rate of 2 mL/sec with a power injector (MCT Plus; Medrad, Pittsburgh, Pa). Scanning parameters were 120 kVp and 250 mA; scanning time was 1 second. The image data were reconstructed with a bone algorithm and 7-mm collimation. The reconstructed image data were directly interfaced and sent to 2-k picture archiving and communication system monitors (GE Medical Systems Integrated Imaging Solutions, Mount Prospect, Ill), at which both lung (window width, 1,500 HU; window level, ⫺700 HU) and mediastinal (window width, 400 HU; window level, 20 HU) window images could be displayed. One chest radiologist (K.S.L.) with 12 years of experience who was blinded to the results of FDG PET prospectively interpreted the CT images. The radiologist recorded the presence, size, and location of primary tumors. Cervical, mediastinal, and upper abdominal lymph nodes were considered to be positive for malignancy when they were larger than 10 mm in short-axis diameter. Hilar lymph nodes were considered to be positive for malignancy when they were greater than 10 mm in diameter in any axis (20). Lymph nodes were classified into the following 11 groups according to a modified version of the lymph node mapping system for esophageal cancer proposed by Korst et al (21): group Cx, cervical; group Pt, paratracheal; group 5, aortopulmonary; group 7, subcarinal; group 8, paraesophageal; group 9, inferior pulmonary ligament; group 10, hilar; group 15, diaphragmatic; group 17, left gastric; group 18, common hepatic; and group 20, celiac.

PET Scanning FDG PET scans were obtained with an Advanced PET scanner (GE Medical Systems). All patients fasted for at least 6

FDG PET versus CT for Detection of Nodal Metastases

䡠

765

Radiology Figure 1. Esophageal cancer in a 58-year-old man. (a) Mediastinal window of transverse contrast material– enhanced (7.0-mm collimation) CT scan obtained at the level of the suprahepatic inferior vena cava shows no demonstrable esophageal wall thickening or mass. (b) Transaxial FDG PET scan obtained at the same level as a shows an area of abnormally increased uptake (arrow) in the esophagus that proved to represent primary stage T2 squamous cell carcinoma.

hours before the PET examination. Emission scans were obtained from head to thigh for 5 minutes per frame 45 minutes after the intravenous injection of 370 MBq (10 mCi) of FDG. Tomographic images were reconstructed without attenuation correction by using filtered backprojection with a Hanning filter (cutoff frequency, 8.0 mm) and were displayed in a 128 ⫻ 128 matrix (pixel size, 4.29 ⫻ 4.29 mm with a section thickness of 4.25 mm). In-plane and transverse resolution of the reconstructed images, respectively, were 9.8-mm and 10.1-mm full width at half maximum. In addition, attenuationcorrected images were acquired in the thorax and sometimes at the level of the upper abdomen by reconstruction of 10minute postemission transmission or preinjection transmission images with germanium 68 rods. A nuclear medicine physician (B.T.K.) with 8 years of experience who was blinded to the results of CT scanning prospectively interpreted the PET images. This physician recorded the presence and location of presumed primary tumors and lymph nodes with metastases. When an area of presumed primary tumor or a lymph node showed focally prominent FDG uptake compared with background mediastinal activity, the area or node was considered to be positive for malignancy.

Surgery and Specimen Analysis Patients in whom the primary tumor was located in the upper thoracic esophagus (n ⫽ 7) underwent transthoracic 766

䡠

Radiology

䡠

June 2003

esophagectomy (involving laparotomy, right thoracotomy, and cervical anastomosis) with lymph node dissection in three fields (thoracic, abdominal, and cervical [including supraclavicular nodes]). Patients in whom the primary tumor was located in the middle portion of the esophagus (n ⫽ 42) or the lower thoracic esophagus or gastroesophageal junction (n ⫽ 32) underwent transthoracic esophagectomy (involving laparotomy, right thoracotomy, and high thoracic anastomosis) with two-field (thoracic and abdominal) lymph node dissection. During surgery, a thoracic surgeon (Y.M.S.) with 15 years of experience dissected all visible and palpable lymph nodes in the surgical field, taking into consideration all results from the preoperative imaging examinations, including CT and FDG PET. The esophagectomy specimen was opened longitudinally in the fresh state. After periesophageal fat was dissected, lymph nodes were sought. Thereafter, the specimen was fixed overnight in 10% neutral buffered formalin. Descriptions of the tumor (ie, appearance, depth of invasion, distance from both lines of resection and from cardia), the mucosal appearance, the wall thickness, and the lymph nodes (the location, size, and number of nodes seen) were recorded. Sections were obtained for the histopathologic evaluation of the tumor, the nonneoplastic mucosa, the proximal and distal lines of resection, and the lymph nodes. The specimens were stained with a standard hematoxylin-eo-

sin technique and examined with light microscopy. In each patient, the pathologic stage of primary esophageal cancer was recorded. A total of 633 nodal groups (24 cervical, 391 mediastinal, 19 hilar, and 199 abdominal) were dissected in the 81 patients. Twenty-five patients were observed to have positive lymph nodes (representing 44 lymph node groups—29 hilar, 13 mediastinal, and two abdominal groups) at either preoperative CT or FDG PET, but these nodes could not be dissected surgically and were considered to be negative for malignancy when they did not show any change in size on follow-up CT scans (range of follow-up period, 3–19 months; mean, 9.6 months). These nodes were mainly of the hilar and mediastinal nodal groups because bilateral hilar or mediastinal nodes could not be assessed simultaneously with one surgical approach. These nodes were also included in this study. Therefore, our study included a total of 677 nodal groups (24 cervical, 404 mediastinal, 48 hilar, and 201 abdominal).

Data Analysis and Statistics Because there were long time intervals between CT and surgery and between FDG PET and surgery, we tested how time intervals may have affected diagnostic sensitivity and sensitivity. For this evaluation, we analyzed whether the difference in sensitivity or the difference in specificity between CT and PET was affected by the time intervals between imaging and surgery by using a generalized estimating equation. The accuracy of CT and FDG PET in the detection of a primary tumor and metastasis in each nodal group was determined, with results of pathologic examination as the standard of reference; the accuracy values of the two modalities were then compared. For purposes of comparison, all lymph node groups were subdivided into cervical, mediastinal, hilar, and abdominal groups; detection rates for metastasis in these cervical, mediastinal, hilar, and abdominal nodes at CT and FDG PET were then compared. In addition, the nodal stage (N and M stages) of cancer in each patient, as determined at CT and FDG PET, was assessed, and the results of each modality were compared. In this study, N1 disease was defined as involvement of any regional lymph node groups, including thoracic and upper abdominal nodes (except cervical and celiac lymph nodes; involvement of these nodes was regarded as indicating M1a or M1b cancer accordYoon et al

TABLE 1 Detection of Lymph Node Groups with Metastasis at CT and FDG PET in Patients with Esophageal Cancer

Radiology

Lymph Node Group* Modality and Nodal Groups CT† FDG PET† Number of truly positive nodal groups/number of nodal groups evaluated

Cx

Pt

5

7

8

9

10

15

17

18

20

Total

0 1

1 11

0 0

0 0

0 6

1 0

0 1

0 1

8 7

0 0

1 2

11 29

3/24

26/140

2/30

5/74

22/99

3/58

3/48

1/3

24/80

3/46

4/75

96/677

* Cx ⫽ cervical, Pt ⫽ paratracheal, 5 ⫽ aortopulmonary, 7 ⫽ subcarinal, 8 ⫽ paraesophageal, 9 ⫽ inferior pulmonary ligament, 10 ⫽ hilar, 15 ⫽ diaphragmatic, 17 ⫽ left gastric, 18 ⫽ common hepatic, 20 ⫽ celiac. † Data are numbers of nodal groups considered positive for metastasis.

ing to the TNM staging system). The McNemar test was used to compare the accuracy of CT and FDG PET in the detection of primary tumor and nodal stage. In our comparison of the differences in diagnostic accuracy between CT and PET in terms of the detection of metastasis in an individual nodal group, intrapatient and intranodal correlations were taken into account, because multiple nodes with metastasis might appear in a single patient at imaging or surgery, and multiplicity could affect the interpretation of the image and increase the likelihood of a particular diagnosis. For this comparison study, we estimated the variances of the sensitivity and specificity values of CT and PET and the covariances of the sensitivity and specificity values of CT and PET, and then compared the accuracy of the two modalities (22). A P value of less than .05 was considered to represent a statistically significant difference.

RESULTS Results of application of the generalized estimating equation to determine whether the difference in sensitivity or the difference in specificity between CT and PET was affected by the time interval between imaging and surgery indicated that neither difference was affected by the time interval (P ⫽ .641 for sensitivity, P ⫽ .200 for specificity).

Figure 2. Esophageal cancer in a 60-year-old man. (a) Mediastinal window of transverse contrast-enhanced (7.0-mm collimation) CT scan obtained at the level of azygos arch shows no markedly enlarged mediastinal lymph node. (b) Transaxial FDG PET scan obtained at same level as a shows an area of abnormally increased uptake (arrow) at the right paratracheal area that proved to represent a lymph node with metastasis at pathologic examination.

tients. Primary tumor was correctly detected in 65 patients (80%) at CT and in 74 patients (91%) at FDG PET. FDG PET was significantly more sensitive than CT for the depiction of primary tumor (P ⫽ .006) (Fig 1). Two cases of carcinoma in situ, nine cases of T1 cancer, three cases of T2 cancer, and two cases of T3 cancer were missed at CT, while one case of carcinoma in situ and six cases of T1 cancer were missed at FDG PET. Five cases of T1 cancer and one case of carcinoma in situ were missed at both CT and FDG PET.

Detection of Primary Tumors The distribution of postoperative T stages in the 81 patients was as follows: carcinoma in situ in two patients, T1 cancer (tumor invades lamina propria or mucosa) in 13 patients, T2 cancer (tumor invades muscularis propria) in 22 patients, T3 cancer (tumor invades adventitia) in 42 patients, and T4 cancer (tumor invades adjacent structures) in two paVolume 227

䡠

Number 3

Detection of Malignant Lymph Nodes in Each Nodal Group A total of 677 nodal groups (24 cervical, 404 mediastinal, 48 hilar, and 201 abdominal) were evaluated, and 96 nodal groups (three cervical, 59 mediastinal, three hilar, and 31 abdominal) in 46 patients proved to be positive for malignancy at pathologic examination.

With CT, 11 (11%; variance, 0.0011) of 96 nodal groups that proved at pathologic examination to be metastatic were detected (none of three cervical, two of 59 mediastinal, none of three hilar, and nine of 31 abdominal nodal groups). With FDG PET, 29 (30%; variance, 0.0026) of the 96 metastatic nodal groups were detected (one of three cervical, 18 of 59 mediastinal, one of three hilar, and nine of 31 abdominal nodal groups) (P ⬍ .001; covariance for CT and FDG PET, 0.0007) (Table 1; Figs 2, 3). There were 28 false-positive interpretations (in 11 mediastinal, four hilar, and 13 abdominal nodal groups) at CT, for a specificity of 95% (553 of 581 groups; variance, 0.0002), and 56 false-positive interpretations (in 23 mediastinal, 32 hilar, and one abdominal nodal group) at FDG PET, for a specificity of 90% (525 of 581 groups; variance, 0.0002) (Table 2). CT was significantly more specific for the de-


䡠

767

Radiology

tively (P ⫽ .25). The specificities of CT and FDG PET for depicting M1 disease were 95% (70 of 74 patients) and 99% (73 of 74 patients), respectively (P ⫽ .375). The accuracies of CT and FDG PET for nodal staging were 49% (27 of 35 patients with N0 disease, 12 of 39 patients with N1 disease, and one of seven patients with M1 disease) and 62% (22 of 35 patients with N0 disease, 25 of 39 patients with N1 disease, and three of seven patients with M1 disease), respectively (P ⫽ .066).

DISCUSSION

Figure 3. Esophageal cancer in a 61-year-old man. Mediastinal windows of transverse contrastenhanced (7.0-mm collimation) CT scans obtained at the level of (a) the distal trachea and (b) the bronchus intermedius show no evidence of markedly enlarged mediastinal or hilar lymph nodes. The wall (arrow in b) of the subcarinal midesophagus is slightly thickened; this was regarded as the location of the primary tumor. (c, d) Transaxial FDG PET images obtained at the same levels as a and b, respectively, show areas of abnormally increased uptake at the right hilum (standardized uptake value, 5.5) (arrow in d), left hilum (standardized uptake value, 3.8) (arrow in c), and subcarinal esophagus (arrowhead in d). Bilateral hilar lymph node groups were dissected and proved to be negative for malignancy at pathologic examination (results not shown). Stage T2 squamous cell carcinoma was found in the midesophagus in the surgical specimen (not shown).

piction of metastases to lymph nodes (P ⫽ .009; covariance for CT and FDG PET, 0.00001). The accuracies of CT and FDG PET for the depiction of malignant lymph node groups were 83% (564 of 677 groups) and 82% (554 of 677 groups), respectively (P ⫽ .382).

Assessment of Nodal Stage by Patient The distribution of postoperative nodal stages in the 81 patients was as follows: N0 in 35 patients, N1 in 39 patients, and M1 (indicating metastasis to 768

䡠

Radiology

䡠

June 2003

cervical or celiac lymph nodes) in seven patients. For preoperative nodal staging, the sensitivities of CT and FDG PET for depicting N1 disease were 31% (12 of 39 patients) and 64% (25 of 39 patients), respectively (P ⫽ .005). The specificities of CT and FDG PET for depicting N1 disease were 86% (36 of 42 patients) and 69% (29 of 42 patients), respectively (P ⫽ .118). The sensitivities of CT and FDG PET for depicting M1 disease (ie, metastasis to cervical or celiac lymph nodes) were 14% (one of seven patients) and 43% (three of seven patients), respec-

Metastasis to lymph nodes is the most important prognostic factor in esophageal carcinoma (23). Both the number and the location of involved lymph nodes have been considered to be important prognostic factors (11). The main cause of death in patients with esophageal carcinoma is local recurrence, the rate of which is reported to be up to 80% (1). About one-half of these recurrences seem to be attributable to metastases in lymph nodes left after surgery (1). Therefore, once the esophageal carcinoma is found and treatment is considered, the next step is to assess for metastasis to lymph nodes (1). CT and FDG PET have been described as not being accurate enough for evaluation of metastases to lymph nodes in esophageal carcinoma; widely variable ranges of sensitivity, specificity, and accuracy have been reported for both modalities. The sensitivity, specificity, and accuracy, respectively, of CT have been reported to range from 8% to 75%, 60% to 99%, and 45% to 96%; those of FDG PET have been reported to range from 22% to 57%, 90% to 100%, and 37% to 90% (1–5,8,9,11,23,24). Because nodal metastasis detection is based on the size of lymph nodes, CT has been nonsensitive for depicting metastatic spread to regional lymph nodes, in which tumor cells tend to involve nodes of normal size (7,11). FDG PET also may not enable detection of microscopic metastases in lymph nodes due to its limited resolution and scatter effects (9,11,25). Carbon 11 (11C)–labeled choline PET has been reported to have better sensitivity in the detection of metastases in mediastinal lymph nodes than FDG PET and CT. However, 11C-labeled choline PET is ineffective in the detection of metastases located in the upper abdominal nodes because of the normal uptake of 11C-labeled choline in the liver (1). Because lymph Yoon et al

TABLE 2 False-Positive Detections of Lymph Node Groups with Metastasis at CT and FDG PET in Patients with Esophageal Cancer

Radiology

Lymph Node Group* Modality and Nodal Groups CT† FDG PET† Number of truly negative nodal groups/number of nodal groups evaluated

Cx

Pt

5

7

8

9

10

15

17

18

20

Total

0 0

2 5

0 3

6 7

1 8

1 0

4 32

1 0

9 0

0 0

4 1

28 56

21/24

114/140

28/30

69/74

77/99

55/58

45/48

2/3

56/80

43/46

71/75

581/677

* Cx ⫽ cervical, Pt ⫽ paratracheal, 5 ⫽ aortopulmonary, 7 ⫽ subcarinal, 8 ⫽ paraesophageal, 9 ⫽ inferior pulmonary ligament, 10 ⫽ hilar, 15 ⫽ diaphragmatic, 17 ⫽ left gastric, 18 ⫽ common hepatic, 20 ⫽ celiac. † Data are numbers of nodal groups incorrectly believed to be positive for metastasis.

node enlargement can be caused by either reactive hyperplasia or granulomatous inflammation, it is difficult to differentiate between benign lymph nodes and malignant lymph nodes at either CT or FDG PET alone (3,11,12,26). In the current study, the sensitivities of CT and FDG PET in the detection of metastases in lymph nodes were comparable to those reported by previous researchers (5,11) and were relatively low. One possible cause for these relatively low sensitivity values may be related to the inclusion criteria we used. In this study, only patients who underwent esophagectomy with lymph node dissection were included. Patients who received palliative treatment, preoperative chemotherapy, or radiation therapy were excluded. Therefore, more cases of early-stage disease, and, therefore, more patients with microscopic metastatic foci, could have been included in the current study. In the current study, the sensitivity of FDG PET in the detection of metastases in intrathoracic nodes was superior to that of CT: Two (3%) of 59 such metastases were detected at CT versus 18 (31%) at FDG PET. However, CT and FDG PET had equally poor sensitivity in the detection of abdominal lymph nodes (nine [29%] of 31 abdominal nodal groups for both modalities) (Table 1). Use of a predetermined lymph node size threshold (10 mm in diameter) as the diagnostic criterion for nodal metastasis at CT may have been problematic. Small mediastinal or upper abdominal lymph nodes that were less than 10 mm in short-axis diameter may have contained malignant cells. FDG PET had relatively low sensitivity for depicting metastases in lymph nodes in the abdomen, images of which were not routinely attenuation corrected (unlike images of the mediastinum). Because regional lymph nodes or esophageal cancers are primarily located deep Volume 227

䡠

Number 3

within the body, attenuation effects may have substantially decreased the sensitivity of FDG PET for depiction of small lymph nodes, which are more vulnerable to partial volume effects. Esophageal and stomach peristalsis, with attendant motion artifacts, may also have contributed to the decrease in the sensitivity of PET for depicting metastases in paraesophageal and abdominal lymph nodes (11). Twenty-eight benign nodal groups, including six subcarinal, four hilar, nine left gastric, and four celiac nodal groups, were falsely assessed as malignant at CT. The large size of reactive lymph nodes in the subcarinal area and a relatively low frequency of visible lymph nodes in the upper abdomen may have led to falsepositive results. Fifty-six nodal groups were falsely interpreted as positive for malignancy at FDG PET. Fifty-five of these 56 false-positive results were for nodes in the thorax, and more than half (32 nodal groups in 20 patients) of them were for hilar lymph node groups. Because granulomatous disease (tuberculosis) is endemic in our country, we attribute such false-positive results at FDG PET for the hilar node groups to the presence of reactive hyperplasia or active inflammation of normal-sized lymph nodes. Therefore, interpretation of FDG PET scans of hilar nodal groups, particularly in areas of endemic chronic inflammatory diseases such as tuberculosis and bronchiectasis, should be performed cautiously (11). Conversely, on FDG PET scans that were interpreted as truly positive for metastasis in the hilar nodes, areas of increased FDG uptake were commensurate with enlarged hilar lymph nodes on CT scans. Accuracy for nodal staging has been described to range from 45% to 62% for CT and from 48% to 90% for FDG PET (11,15,17). In the current study, the accuracies for nodal staging (49% for CT

and 62% for FDG PET) were similar to those observed in previous studies. In our study, CT had low sensitivity in the detection of malignant lymph nodes. Several drawbacks limited our study. First, because patients with advanced stages of esophageal cancer were excluded from this study, both sensitivity and accuracy in the detection of metastases to lymph nodes may have been underestimated for both modalities. Second, some lymph nodes that were initially reported to be positive for malignancy at CT or FDG PET but could not be dissected due to anatomic or technical factors (these nodes were considered to be negative for malignancy because follow-up CT studies did not show any evidence of enlargement) were classified as falsely positive; in other words, specificities in the detection of metastases in lymph nodes and in the detection of N1 disease may have been underestimated. Third, only one radiologist and only one nuclear medicine physician, respectively, interpreted the CT and PET images. This may have added some bias. In addition, surgeons were guided with preoperative CT and PET findings. This fact may have added verification bias. Fourth, the images were reconstructed by using filtered back-projection rather than an iterative method, and attenuation correction was performed only for images of the chest and by using measured attenuation values rather than a segmentation method. These imaging methods are likely to have produced somewhat noisier PET images and to have hampered more accurate interpretations. In summary, FDG PET scanning is significantly more sensitive than CT for depicting metastases to lymph nodes in patients with esophageal squamous cell carcinoma, although it has relatively lower specificity than CT because of many false-positive interpretations, espe-


䡠

769

Radiology

cially of pulmonary hilar nodes. Therefore, interpretation of FDG PET images for metastases in hilar nodes should be performed cautiously, especially when there is evidence of chronic inflammatory pulmonary diseases either at other imaging examinations or in the patient’s clinical history and in areas where chronic inflammatory diseases such as tuberculosis are endemic.

7.

Acknowledgment: We thank Seonwoo Kim, PhD, of the Biostatistics Unit of the Samsung Biomedical Research Institute of Samsung Medical Center for being a great help to us in statistical consultation.

10.

References 1. Kobori O, Kirihara Y, Kosaka N, Hara T. Positron emission tomography of esophageal carcinoma using (11)C-choline and (18)F-fluorodeoxyglucose: a novel method of preoperative lymph node staging. Cancer 1999; 86:1638 –1648. 2. Buenaventura P, Luketich JD. Surgical staging of esophageal cancer. Chest Surg Clin N Am 2000; 10:487– 497. 3. Block MI, Patterson GA, Sundaresan RS, et al. Improvement in staging of esophageal cancer with the addition of positron emission tomography. Ann Thorac Surg 1997; 64:770 –776. 4. Rice TW. Clinical staging of esophageal carcinoma: CT, EUS, and PET. Chest Surg Clin N Am 2000; 10:471– 485. 5. Kim K, Park SJ, Kim BT, Lee KS, Shim YM. Evaluation of lymph node metastases in squamous cell carcinoma of the esophagus with positron emission tomography. Ann Thorac Surg 2001; 71:290 –294. 6. Whyte RI. Advances in the staging of intrathoracic malignancies. World J Surg 2001; 25:167–173.

770

䡠

Radiology

䡠

June 2003

8.

9.

11.

12.

13.

14. 15.

16.

17.

Picus D, Balfe DM, Koehler RE, Roper CL, Owen JW. Computed tomography in the staging of esophageal carcinoma. Radiology 1983; 146:433– 438. Vilgrain V, Mompoint D, Palazzo L, et al. Staging of esophageal carcinoma: comparison of results with endoscopic sonography and CT. AJR 1990; 155:277–281. Rankin SC, Taylor H, Cook GJ, Mason R. Computed tomography and positron emission tomography in the pre-operative staging of oesophageal carcinoma. Clin Radiol 1998; 53:659 – 665. Helmberger H. CT for local staging of esophageal cancer. Dis Esophagus 1999; 12:202–204. Choi JY, Lee KH, Shim YM, et al. Improved detection of individual nodal involvement in squamous cell carcinoma of the esophagus by FDG PET. J Nucl Med 2000; 41:808 – 815. Hansen CP, Oskarsson K, Mortensen D. Computed tomography for staging of oesophageal cancer. Ann Chir Gynaecol 2000; 89:14 –18. Luketich JD, Schauer PR, Meltzer CC, et al. Role of positron emission tomography in staging esophageal cancer. Ann Thorac Surg 1997; 64:765–769. Meyenberger C, Fantin AC. Esophageal carcinoma: current staging strategies. Recent Results Cancer Res 2000; 155:63–72. Kole AC, Plukker JT, Nieweg OE, Vaalburg W. Positron emission tomography for staging of oesophageal and gastroesophageal malignancy. Br J Cancer 1998; 78: 521–527. Flamen P, Lerut A, Van Cutsem E, et al. Utility of positron emission tomography for the staging of patients with potentially operable esophageal carcinoma. J Clin Oncol 2000; 18:3202–3210. Flanagan FL, Dehdashti F, Siegel BA, et al. Staging of esophageal cancer with 18Ffluorodeoxyglucose positron emission to-

18.

19. 20.

21.

22. 23.

24.

25.

26.

mography. AJR Am J Roentgenol 1997; 168:417– 424. Meltzer CC, Luketich JD, Friedman D, et al. Whole-body FDG positron emission tomographic imaging for staging esophageal cancer comparison with computed tomography. Clin Nucl Med 2000; 25: 882– 887. Sobin LH, Wittekind C, eds. UICC TNM classification of malignant tumor. 5th ed. New York: John Wiley & Sons, 1997. Glazer GM, Gross BH, Aisen AM, Quint LE, Francis IR, Orringer MB. Imaging of the pulmonary hilum: a prospective comparative study in patients with lung cancer. AJR 1985; 145:245–248. Korst RJ, Rusch VW, Venkatraman E, et al. Proposed revision of the staging classification for esophageal cancer. J Thorac Cardiovasc Surg 1998; 115:660 – 669. Obuchowski NA. Nonparametric analysis of clustered ROC curve data. Biometrics 1997; 53:567–578. Lerut T, Flamen P, Ectors N, et al. Histopathologic validation of lymph node staging with FDG-PET scan in cancer of the esophagus and gastroesophageal junction: a prospective study based on primary surgery with extensive lymphadenectomy. Ann Surg 2000; 232:743–752. Goei R, Lamers RJS, Engelshove HA, Oei KT. Computed tomographic staging of esophageal carcinoma: a study on interobserver variation and correlation with pathologic findings. Eur J Radiol 1992; 15:40 – 44 McAteer D, Wallis F, Couper G, et al. Evaluation of 18F-FDG positron emission tomography in gastric and oesophageal carcinoma. Br J Radiol 1999; 72:525–529. Lightdale CJ. Positron emission tomography: another useful test for staging esophageal cancer. J Clin Oncol 2000; 18: 3199 –3201.

Yoon et al