The clinical application of 18F-fluorodeoxyglucose positron emission tomography to predict survival in patients with operable esophageal cancer

Authors

  • Hiroyuki Kato MD, PhD,

    Corresponding author
    1. Department of General Surgical Science (Surgery I), Gunma University, Graduate School of Medicine, Maebashi, Japan
    2. Department of Surgical Oncology (Surgery I), Dokkyo Medical University, Tochigi, Japan
    • Department of Surgical Oncology (Surgery I), Dokkyo Medical University, 880, Kitakobayashi, Mibu-machi, Tochigi, 321-0293, Japan===

    Search for more papers by this author
    • Fax: (011) 81-282-86-6213

  • Masanobu Nakajima MD, PhD,

    1. Department of General Surgical Science (Surgery I), Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author
  • Makoto Sohda MD, PhD,

    1. Department of General Surgical Science (Surgery I), Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author
  • Naritaka Tanaka MD,

    1. Department of General Surgical Science (Surgery I), Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author
  • Takanori Inose MD,

    1. Department of General Surgical Science (Surgery I), Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author
  • Tatsuya Miyazaki MD, PhD,

    1. Department of General Surgical Science (Surgery I), Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author
  • Minoru Fukuchi MD, PhD,

    1. Department of General Surgical Science (Surgery I), Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author
  • Noboru Oriuchi MD, PhD,

    1. Department of Diagnostic Radiology and Nuclear Medicine, Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author
  • Keigo Endo MD, PhD,

    1. Department of Diagnostic Radiology and Nuclear Medicine, Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author
  • Hiroyuki Kuwano MD, PhD

    1. Department of General Surgical Science (Surgery I), Gunma University, Graduate School of Medicine, Maebashi, Japan
    Search for more papers by this author

Abstract

BACKGROUND:

Metabolic tumor activity using 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) was believed to have a predictive value for patient outcome in malignancies. The objective of the current study was to assess the prognostic effectiveness of the highest standardized uptake value (SUV) in the primary or regional area (peak SUV) and the number of PET-positive lymph nodes in esophageal cancer.

METHODS:

The authors retrospectively reviewed their experience with 184 consecutive esophageal cancer patients imaged preoperatively using FDG-PET scanning.

RESULTS:

The median peak SUV was 4.5 (range, 1.4-21.9). The survival curve was analyzed using the median peak SUV as the cutoff value. Comparison of each group and clinicopathologic characteristics revealed significant associations between peak SUV and each of the following factors: tumor status (P < .001), lymph node status (P < .001), metastatic status (P < .05), stage of disease (P < .001), number of PET-positive lymph nodes (P < .001), and the number of histologically positive lymph nodes (P < .001). The 5-year overall survival (OS) rate for patients having FDG uptake with a peak SUV ≥4.5 was 47% and that for patients with a peak SUV <4.5 was 76% (P < .0001). On multivariate survival analysis using the Cox proportional hazards model, peak SUV and the number of PET-positive lymph nodes were found to be independent predictive factors for OS. The number of PET-positive lymph nodes was a single prognostic factor predicting both disease-free survival and OS.

CONCLUSIONS:

Pretreatment PET cannot only potentially diagnose the extent of disease, but also may be predictive of patient survival after esophageal cancer resection. Cancer 2009. © 2009 American Cancer Society.

Esophageal cancer has proven to be one of the most difficult malignancies to cure, despite improvement in surgical techniques, reduced perioperative mortality, and the introduction of multimodality therapies.1, 2 Accurate tumor staging, particularly with regard to the depth of tumor invasion, involvement of lymph nodes, and distant metastasis, is essential for optimal treatment selection and delivery, facilitating individually tailored patient management.1

The role and potential value of 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) as a noninvasive imaging modality in the management of esophageal cancer has been widely investigated over the past decade.3-15 FDG-PET imaging may facilitate diagnosis in patients with malignant diseases by enabling the differentiation between benign and malignant tumors, assessment of the extension of disease,3-11 detection of tumor recurrence12-14 and monitoring responses to therapy.15-18 We have previously reported the usefulness of FDG-PET for the staging of esophageal squamous cell carcinoma (SCC); it offers higher sensitivity, specificity, and accuracy for lymph node detection compared with computed tomography (CT), particularly in the neck and upper thoracic region.9 With regard to staging accuracy, the incremental diagnostic information provided by FDG-PET led to a change of staging in 14% of the patients with esophageal cancer, compared with conventional imaging.19, 20 Similar results have been reported by other groups.5, 6, 8, 19, 21, 22 In general, a change in International Union Against Cancer (UICC) stage occurs in approximately 20% of the patients with esophageal cancer using FDG-PET compared with conventional imaging modalities.23 Furthermore, changes in tumor metabolic activity during preoperative treatment have been shown to correlate with histopathologic response and patient survival.24-26

Some researchers have suggested that FDG may be a predictive tool for patient outcome in esophageal cancer, as has already been demonstrated for other types of malignancies including lung cancer and head and neck cancer.27, 28 However, to our knowledge, only limited data are available with regard to esophageal cancer. The degree of FDG uptake in the primary tumor, presenting as the standardized uptake value (SUV), is also associated with patient survival in esophageal cancer.29-37 By contrast, the relation between tumor FDG uptake and patient survival was not found to be statistically significant.22 We have previously reported a study of 25 patients with resected esophageal SCC that demonstrated a correlation between pretreatment maximum (max) SUV and survival, in addition to diagnostic accuracy for staging.9 However, data regarding the clinical application of SUV measurements in such patients are scarce, because to our knowledge only a small number of patients have been described with a selection of disease stages.

In the current study, we retrospectively reviewed our experience with esophageal cancer patients imaged preoperatively by FDG-PET scanning to determine the prognostic value of the highest SUV in the primary or regional area and the number of PET-positive lymph nodes. We also investigated whether the above parameters were independent prognostic predictors above other clinical variables in patients with esophageal cancer who were undergoing curative surgery.

MATERIALS AND METHODS

Patients

We studied 184 consecutive patients with thoracic esophageal SCC who had radical esophagectomy performed at the Department of General Surgical Science at the Graduate School of Medicine of Gunma University between January, 1999 and April, 2007. None of the patients had received prior treatment. The median age of the patients was 63.9 years (range, 41 years-79 years). Tumor stage and disease grade were classified according to the sixth edition of the TNM classification of the UICC.38 We determined the tumor stage using CT scans of the neck, chest, and abdomen; bone scans; endoscopic ultrasound (EUS); endoscopy; and esophagography. Resectability was determined by staging, which included CT scans of the neck, chest, and abdomen; bone scans; magnetic resonance imaging; EUS; and esophagography. In the case of proximal thoracic tumors, bronchoscopy was performed to exclude major airway invasion. The blood sugar levels in all patients were <120 mg/dL at the time of the PET scan.

Treatment and Clinical Outcome

Two different procedures were used. In suitable patients, a standard esophagectomy was performed by the McKeown method (right thoracotomy followed by laparotomy and neck incision with a cervical anastomosis) and 3-field (thoracoabdominal and cervical) lymph node dissection was also performed if indicated. In other patients, the Ivor-Lewis esophagectomy was used (right thoracotomy and laparotomy with anastomosis in the chest) and a 2-field (thoracoabdominal) lymph node dissection was performed. All patients underwent curative thoracic esophagectomy that included the esophagogastric junction.19 After surgery, the lymph nodes were separated from the resected esophagus and the adjacent tissue and were assigned specific numbers indicating the localization of the lymph node according to the guidelines of the Japanese Society for Esophageal Diseases.39 The surgical specimens were fixed, embedded, stained with hematoxylin and eosin, and examined microscopically by 2 pathologists.

PET Imaging

We obtained PET images using a SET 2400 W (Shimadzu Corporation, Kyoto, Japan) or a hybrid PET/CT system (GE Discovery ST8; GE, Milwaukee, Wis). A whole-body image was initiated 40 minutes after the injection of FDG using the simultaneous emission/transmission method.20 Between 4 and 5 sections were imaged from head to thigh for a period of 8 minutes per section. Patients had fasted for at least 4 hours before FDG-PET. Our institutional review board approved the imaging protocols,20 and all patients provided informed consent before undergoing the examination. Two experienced nuclear medicine physicians qualitatively evaluated all PET images. SUV was defined as the concentration of radioactivity in the tissue or lesion (megabecquerel [MBq]/mL) × patient body weight (g)/injected dose (MBq). For semiquantitative analysis, we used max SUV. Regional lymph nodes, which were evaluated using PET, were assigned specific numbers to indicate localization, in accordance with the guidelines of the Japanese Society for Esophageal Diseases.39

Statistical Analysis

All analyses were based on assessment of the highest SUV in the primary or regional area (peak SUV) and the number of PET-positive lymph nodes. The relations between each group and clinicopathologic parameters were determined using the chi-square method and the Fisher exact test. The correlations between each group and patient age and the number of histologically positive lymph nodes were determined using analysis of variance (ANOVA) or the Student t test. For survival analysis, the study group was divided into 2 subgroups, a low group and a high group, using the median peak SUV value as the cutoff value. Survival rates were calculated by the Kaplan-Meier method for analysis of censored data. The significance of differences in survival was analyzed by means of the log-rank test. To discover independent prognostic factors, we performed univariate and multivariate analyses of survival using the Cox proportional hazards model. All P values were 2-tailed and a P < .05 was considered to be statistically significant.

RESULTS

Patient Characteristics

Characteristics of the 184 patients (169 males [92%] and 15 females), with a mean age 63.9 years with a standard deviation of 8.2 years, are shown in Table 1. The locations of tumors were 24 in the upper esophagus, 87 in the middle esophagus, and 73 in the lower esophagus. SCC was present in 168 patients, adenocarcinoma were noted in 8 patients, and other types of cancer were present in the remaining 8 patients. All metastatic cancers were found in the cervical lymph nodes. The pathologic tumor stages were stage I in 56 patients, stage II in 51 patients, stage III in 39 patients, and stage IV in 38 patients. A McKeown esophagectomy was performed in 114 (62%) patients and an Ivor-Lewis esophagectomy in 70 (38%) patients.

Table 1. Patient Characteristics and Peak Standardized Uptake Value
ParameterNo. of PatientsPeak SUV <4.5Peak SUV ≥4.5P
n = 184n = 92n = 92
  1. SUV indicates standardized uptake value; SD, standard deviation; G1, well-differentiated tumor; G2, moderately differentiated tumor; G3, poorly differentiated tumor; G4, undifferentiated tumor; PET, positron emission tomography.

Gender
 Male1698386.4192
 Female1596 
Age, y
 Average63.963.464.5.3322
 SD8.28.18.2 
Location
 Upper241410.0765
 Middle874938 
 Lower732944 
Tumor status
 T1827111.0001
 T219316 
 T3751461 
 T4844 
Lymph node status
 N0845727.0001
 N11003565 
Metastatic status
 M01467967.0289
 M1381325 
Stage
 I56515.0001
 II512328 
 III39534 
 IV381325 
Histology
 G1372017.4041
 G2974948 
 G3442222 
 G4615 
No. of PET-positive lymph nodes
 01097237.0001
 1471235 
 220713 
 3817 
No. of histologic positive lymph nodes
 Average2.61.33.9.0003
 SD4.92.86.1 

Peak SUV and Clinicopathologic Parameters

Primary tumors were detected in 148 (81%) of the 184 patients by PET imaging. The median of peak SUV, 4.5 (range, 1.4-21.9), was used as the cutoff between the high (n = 92) and low (n = 92) peak SUV groups.

Comparison of each group and clinicopathologic characteristics revealed significant associations between peak SUV and each of the following factors (Table 1): tumor status (P < .001), lymph node status (P < .001), metastatic status (P < .05), stage (P < .001), number of PET-positive lymph nodes (P < .001), and the number of histologically positive lymph nodes (P < .001). The number of histologically positive lymph nodes in the low peak SUV group was 1.3 ± 2.8, but that in the high peak SUV group was 3.9 ± 6.1 (P = .0003).

PET-related Parameters, Disease-Free Survival, and Overall Survival

The 5-year disease-free survival (DFS) and overall survival (OS) rates were calculated by the Kaplan-Meier method (Table 2). With regard to 5-year DFS and OS, there were significant correlations between survival and each of the following factors: tumor status, lymph node status, metastatic status, stage, peak SUV, and number of PET-positive lymph nodes. However, sex, age, location, and histology were not found to be significantly correlated with DFS and OS. The survival curve was analyzed using a cutoff value of 4.5, which was the median peak SUV (Fig. 1). The 5-year OS of the patients with an FDG uptake with a peak SUV ≥4.5 was 47%, and that for patients with a peak SUV <4.5 was 76% (P < .0001). In these analyses of peak SUV, there were 19 patients with an SUV of a metastatic lesion that was higher than that of the primary tumor. The 5-year OS of these patients was 18%.

Figure 1.

Overall survival (OS) according to the median peak standardized uptake value (SUV) is shown. The 5-year OS of patients having 18F-fluorodeoxyglucose (FDG) uptake with a peak SUV ≥4.5 was 47%, and that for patients with a peak SUV <4.5 was 76% (P < .0001).

Table 2. Survival Rates of 5-Year DFS and OS Using the Kaplan–Meier Method
ParameterNo. of PatientsDFS, %POS, %P
n = 184
  1. DFS indicates disease-free survival; OS, overall survival; G1, well-differentiated tumor; G2, moderately differentiated tumor; G3, poorly differentiated tumor; G4, undifferentiated tumor; SUV, standardized uptake value; PET, positron emission tomography.

Gender
 Male16963.626360.1415
 Female1571 84 
Age, y
 <648659.086761.7142
 >649869 63 
Location
 Upper2448.058745.0521
 Middle8770 71 
 Lower7361 57 
Tumor status
 T18287.000184.0001
 T21961 55 
 T37546 47 
 T4813 13 
Lymph node status
 N08485.000182.0001
 N110045 44 
Metastatic status
 M014671.000168.0001
 M13835 36 
Stage
 I5695.000192.0001
 II5169 65 
 III3940 39 
 IV3835 36 
Histology
 G13770.771361.9418
 G29764 64 
 G34459 58 
 G4666 42 
Peak SUV
 Peak SUV <4.59278.000176.0001
 Peak SUV ≥4.59249 47 
No. of PET-positive lymph nodes
 010979.000178.0001
 14742 41 
 22034 29 
 >3820 18 

With regard to DFS and OS in univariate survival analysis using a Cox proportional hazards model, there were significant correlations noted between survival and each of the following factors: tumor status, lymph node status, metastatic status, stage, peak SUV, and the number of PET-positive lymph nodes. With regard to DFS on multivariate survival analysis using the Cox proportional hazards model, there were significant correlations noted between DFS and tumor status (P < .001), lymph node status (P < .01), and the number of PET-positive lymph nodes (P < .05). However, there were no significant correlations noted between DFS and metastatic status and peak SUV. With regard to OS on multivariate survival analysis, there were significant correlations noted between OS and tumor status (P < .01), lymph node status (P < .01), peak SUV (P < .05), and number of PET-positive lymph nodes (P < .05). No significant correlations were observed between OS and metastatic status.

DISCUSSION

To the best of our knowledge, the value of baseline PET in patients with localized esophageal cancer with regard to prognostic outcome has not been established to date. We hypothesized that baseline PET would correlate with DFS and OS in patients with operable esophageal cancer. Thus, we retrospectively analyzed whether the highest SUV in the primary or the regional area (peak SUV) and the number of PET-positive lymph nodes were associated with patient survival.

The analysis of peak SUV was considered because the max SUV of metastatic lymph nodes has been reported to often be higher than that in primary tumors.36 It is known that such patients have a poor prognosis. In the current study, the survival of patients having SUVs of metastatic lesions that are higher than those of the primary tumors was extremely poor. However, peak SUV was found to be a significant prognostic factor for OS, but not DFS on multivariate survival analysis using the Cox proportional hazards model. The number of PET-positive lymph nodes was a significant prognostic factor for both DFS and OS. This may be because regional lymph node metastasis is one of the most important prognostic factors in esophageal cancer. We speculated that peak SUV might be mainly associated with tumor status, and that the number of PET-positive lymph nodes might be associated with lymph node status. In short, these PET-related parameters might be correlated considerably with tumor aggressiveness and TNM staging.

Previous reports have shown that SUV is correlated with the immunohistochemistry of Ki-67 (proliferation index marker) in lung cancer.40 In vitro and clinical studies have indicated a strong association between elevated FDG uptake and tumor viability, aggressiveness, hypoxia, and metabolic activation.41-43 For these reasons, SUV measured by PET may be associated with patient survival.

To our knowledge, 2 types of reports on the correlation between baseline PET and patient survival have been published to date: those limited to primary tumors22, 29-34 and others that have included both primary tumors and PET-related parameters.35-37 A study by van Westreenen et al. reported that the max SUV of the primary tumor was associated with survival on univariate analysis.29 However, multivariate analysis demonstrated that max SUV did not have a significant additional impact on estimating patient survival. These results are similar to the analyses of 125 patients with surgically resected esophageal cancer performed by Omloo et al.30 in 2008.These studies indicated that, when analysis also included pathologic and clinical stage, SUV was not found to be independently predictive of survival, despite the large number of patients studied. This may be because many of the primary tumors were classified as T3 and T4. In the former study, there were T3 and T4 tumors in 30 of 40 (75%) patients and in the latter 92 of 125 (74%) patients.

The study by Rizk et al.31 indicated that a high max SUV in patients with esophageal adenocarcinoma is predictive of clinical and pathologic stage as well as OS after resection. In addition, a high max SUV identifies a subset of patients who have clinical and pathologic early stage disease, but who have a poor prognosis. Furthermore, an earlier study by Cerfolio and Bryant32 demonstrated that max SUV as determined by an FDG-PET scan of a patient with a primary esophageal tumor is an independent predictor of survival. They concluded that FDG-PET is a better predictor of survival than the current TNM staging system. Thus, many studies have indicated that pretreatment max SUV of the primary tumor is correlated with patient survival. By contrast, a study by Stahl et al.22 reported that the intensity of FDG uptake in adenocarcinoma did not demonstrate any significant relation with patient survival and concluded that this observation may indicate a truly different significance of the metabolic signal in SCC and adenocarcinoma in terms of tumor aggressiveness and patient survival. However, this study included just a small number of patients and was considerably different from the above reports in terms of cancer treatment and the homogeneity of disease stage.

To our knowledge, there have been 3 previous studies related to both primary tumor and FDG uptake.35-37 Choi et al.,35 in a prospectively studied group of 69 patients with surgically resected SCC of the esophagus, indicated that the number of PET-positive lymph nodes was the greatest prognostic factor predicting both DFS and OS in patients with esophageal cancer. Furthermore, Hong et al. indicated that pretreatment PET scan results were correlated with OS.36 More specifically, the number of PET abnormalities was found to be an independent prognostic variable for OS. However, the other PET parameters such as SUV of the primary tumor, peak SUV, and total SUV did not correlate with OS and DFS. In these studies, the number of PET-positive lymph nodes was found to be an independent prognostic factor, which is consistent with the findings of the current study. This may be because the number of pathologically positive lymph nodes is an independent prognostic factor in esophageal cancer.1, 37 This clinical reality may be consistent with the clinical value of baseline PET in patients with operable esophageal cancer. In this study, all patients with >3 PET-positive lymph nodes had >5 histologically positive lymph nodes, highlighting that there is a distinct possibility that patients with PET-positive lymph nodes already have multiple histologically positive lymph nodes.

In esophageal cancer, some researchers have noted a correlation between PET and survival, but it has been thought uncertain whether intensity was an independent factor for prognosis. To our knowledge, the current study has enrolled the largest number of patients with a homogeneous distribution of disease stage and who were receiving the same treatment. According to our data, peak SUV and the number of PET-positive lymph nodes were independent predictive factors for OS. The number of PET-positive lymph nodes was the single prognostic factor predicting both DFS and OS.

In conclusion, pretreatment PET could not only diagnose the extent of disease but could also predict patient survival after esophageal cancer resection. In the future, there is the possibility that PET will be able to predict the time and location of recurrence in patients undergoing esophagectomy.

Acknowledgements

We thank T. Yoshida, M. Ohno, T. Ogasawara, and Y. Saitoh for secretarial assistance. We also thank H. Emura for help with data management and biostatistical analysis during the preparation of this article.

Conflict of Interest Disclosures

The authors made no disclosures.

Ancillary