The Role of Positron Emission Tomography in Esophageal Cancer

Abstract

The most common methods for staging esophageal cancer are endoscopic ultrasonography (EUS) and computed tomography (CT). EUS is well established in differentiating between early tumor stages and more advanced primary lesions. When combined with fine needle aspiration, EUS has become an important tool in assessing the regional lymph nodes, as well. EUS has its limitations, esophageal obstruction makes passage of the endoscope beyond the tumor nearly impossible and with a narrow field of evaluation, it is not useful for detecting metastatic disease. CT allows for assessment of local tumor invasion while simultaneously providing information regarding distant disease. Its usefulness locally, however, is limited. CT and EUS yield anatomic visualization. Fluorodeoxyglucose (FDG)-positron emission tomography (PET) can provide functional information and is an effective diagnostic modality in esophageal cancer. Its role in the management of esophageal cancer includes staging as well as potential utility in the evaluation of neoadjuvant therapy response and in follow-up after definitive therapy. FDG-PET will likely be more readily used in combination with anatomical imaging like CT to provide additional diagnostic information to aid radiation oncologists in target delineation and planning. In addition, FDG-PET has also been shown to have prognostic value that can be applied to patient management and aid in development of emerging therapies.

Approximately 14,550 people will be diagnosed with esophageal cancer in the United States in 2006. Of these patients, about 13,770 will ultimately die of the disease.¹ While the national incidence has been slowly increasing over the past 25 years at roughly 4.5 cases per 100,000 people per year, the death-to-incidence ratio continues to be relatively high, at over 0.90.² The main reason for the poor prognosis is that esophageal cancer is largely asymptomatic in its early stages. Thus, most cases are diagnosed when the disease has become either locally advanced, with disease spreading to local draining lymph nodes, or distantly metastatic.³

Historically, it has been difficult to distinguish potentially curable from likely incurable disease. This inaccuracy in staging makes both local and distant recurrences relatively common, even after aggressive multimodality treatment using combinations of surgery, chemotherapy, and radiation therapy. While improving curative therapy continues to be an objective, accurate staging is crucial, both to guide therapy and to predict prognosis. Proper staging allows selection of patients who can potentially benefit from local treatment, while sparing those patients with metastatic disease from undergoing intensive multimodality local therapy.³ In addition, the prognostic information obtained through accurate staging aids both the physician and the patient regarding treatment decisions. Finally, the results of clinical trials of therapy for esophageal cancer have been severely confounded by stage migration as a direct result of inaccurate techniques in staging.

In esophageal cancer, the most common methods for staging are endoscopic ultrasonography (EUS) and computed tomography (CT) of the chest and abdomen. EUS is well established in differentiating between early tumor stages and more advanced primary lesions.⁴ When combined with fine needle aspiration, EUS has become an important tool in assessing the regional lymph nodes, as well.⁵ The accuracy of EUS for evaluating primary tumor and nodal status has been reported to be 85% and 75%, respectively while the sensitivity has been reported to be in the range of 85% to 95% for primary tumor evaluation and 70% to 80% for nodal evaluation.^6,7 EUS has its limitations, esophageal obstruction makes passage of the endoscope beyond the tumor nearly impossible and with a narrow field of evaluation, it is not useful for detecting metastatic disease.³ CT allows for assessment of local tumor invasion while simultaneously providing information regarding distant disease. Its usefulness locally, however, is limited, with sensitivity for staging primary tumor and nodal disease reported to be about 50% and 60% to 87%, respectively.⁷

While CT and EUS yield anatomic visualization, positron emission tomography (PET) is a nuclear medicine imaging modality that allows for measurement of physiologic and biochemical processes.⁸ PET can potentially determine quantitative information regarding blood flow, receptor status, and metabolic processes, depending on the radiopharmaceutical selected.⁸ Regarding the role of PET in esophageal cancer imaging, many esophageal tumors exhibit increased cellular metabolism and are associated with increased glycolysis, as well as increased glucose transporter proteins. ⁹

A positron emitting radiotracer, 2-[¹⁸F]-fluoro-2-deoxyglucose (FDG) is a glucose analogue and is transported intracellularly and phosphorylated to 2-[¹⁸F]fluoro- 2-deoxyglucose-6-phosphate (FDG-6-phosphate) via the same pathways as glucose (Figure 1). Due to insufficient amounts of intracellular glucose phosphatase, there is a lack of further metabolism, meaning that the FDG-6-phosphate cannot be further broken down. Since FDG-6-phosphate is a highly polar molecule, it remains trapped in the cancer cell.⁸ When used in nuclear medicine, FDG-6-phosphate accumulates in tumors following intravenous injection and provides a signal of high glycolytic tissue activity throughout the whole body.¹⁰ In most organ systems with the exception of the brain and urinary tract, malignant tumors, including both primary and metastatic esophageal tumors, are readily detected using FDG-PET.¹¹

Figure 1.

2-[¹⁸F]fluoro-2-deoxyglucose (FDG) is transported intracellularly and phosphorylated to 2-[¹⁸F]fluoro-2-deoxyglucose-6-phosphate (FDG-6- phosphate) via the same pathways as glucose. Secondary to insufficient amounts of intracellular glucose phosphatase, there is a lack of further metabolism preventing FDG-6-phosphate from breaking down further. FDG-6-phosphate is a highly polar molecule, trapping it in the cancer cell.

FDG-PET AS A STAGING TOOL

Primary Tumor

Both primary squamous and adenocarcinomas of the esophagus have been extensively evaluated by FDG-PET. With squamous cell carcinomas (SCC), most studies have found a high accumulation of FDG at the primary tumor site, with the majority of the false-negatives found in small volume tumors.¹² The primary cause for inability to detect smaller primary SCC was the limited spatial resolution of the PET imaging devices (approximately 5–8 mm). In primary evaluation of adenocarcinoma, however, insufficient or absent FDG accumulation is more frequently encountered. In adenocarcinoma, FDG avidity was found to be related to tumor growth type, differentiation degree, and mucus content.¹³ Non-avid tumors were often poorly differentiated, showing a diffuse, nonintestinal growth type and mucus-containing tumor type.¹³ Reasons for non-avidity include large amounts of inert mucus which does not accumulate FDG and lack of the glucose transporter, GLUT-1, expression on the cell surface of these tumors.^12,14 While primary tumor detection is a potential use for FDG-PET, EUS remains the standard for local tumor evaluation.

Lymph Nodes

Metastasis to regional lymph nodes is a very important factor in esophageal cancer prognosis. Those patients without involvement of the regional lymph nodes have a clearly better overall survival than those patients with nodal involvement, with survival decreasing as the number of involved lymph nodes increases.¹⁵ Anatomical imaging with CT and/or EUS are commonly used to evaluate for lymph node metastasis, with the combined accuracy of both CT and EUS reported as being greater than that of either modality alone.³

Limitations of anatomical imaging in nodal evaluation include the inability to detect tumor involvement in normal-sized lymph nodes, as well as difficulty in differentiating metastatic from inflammatory disease in enlarged lymph nodes. Also, uptake within the primary tumor may obscure evaluation of the regional lymph nodes directly adjacent to the tumor mass. Due to the inaccuracy of anatomic imaging in these situations, invasive procedures like mediastinoscopy are often necessary to evaluate for lymph node involvement properly and, thus, choose the most ideal therapy for the individual patient. These invasive procedures are associated with potential morbidity and relatively high cost, and ideally would be reserved only for those patients for whom questions remain after comprehensive noninvasive imaging. FDG-PET has been used in this situation as another noninvasive tool to stage patients for regional disease and potentially limit the need for invasive staging.

In a study by Kim et al, FDG-PET and CT were compared using tissue from esophagectomy and lymph node dissection specimens.¹⁶ Forty-seven patients were evaluated preoperatively by FDG-PET and CT scan. Regarding FDG-PET, the sensitivity, specificity, and accuracy for metastatic lymph node detection were 52%, 94%, and 84%, respectively, compared to 15%, 97%, and 77%, respectively, for CT. The authors of the study concluded that FDG-PET has similar specificity, but significantly greater sensitivity and accuracy than CT for the detection of lymph node metastases. While superior to CT alone, comparisons of FDG-PET with EUS appear to be somewhat mixed. Flamen et al prospectively examined 74 patients with esophageal cancer, showing EUS to have 81% sensitivity and 67% specificity for nodal metastasis compared to 33% sensitivity and 89% specificity for FDG-PET.¹⁷ Similar results have been found in other studies comparing EUS to FDG-PET and the combination of EUS and CT with FDG-PET, with results showing a greater sensitivity for EUS.^3,18

A recently published meta-analysis of FDG-PET in detecting regional metastases in twelve studies revealed a pooled sensitivity and specificity of 51% (95% confidence interval [CI], 34% to 69%) and 84% (95% CI, 76% to 91%), respectively.¹⁹ Similar to primary tumor evaluation, false-negative results with FDG-PET are mainly the result of low tumor burden. Another limitation of FDG-PET is difficulty in detecting nodes close to the primary tumor (3 cm or less).²⁰ In these instances, FDG uptake by the primary tumor can obscure the detection of associated nodal metastasis. EUS is a more accurate diagnostic alternative for detecting lymph node metastases in that situation.¹⁷ Of note, false positives can be the result of inflammatory disease causing increased FDG uptake, as well as from heterogeneous tracer uptake of the primary tumor mimicking nodal disease.³ Also, conditions like sarcoidosis may appear as extensive mediastinal metastatic adenopathy when imaged by FDG-PET.⁸ Nodal assessment continues to be relatively insensitive regardless of the imaging modality used, thus nodal sampling is still performed for potential surgical candidates. ¹²

Distant Metastases

FDG-PET has been found to be of greater value in detecting distant metastases. In esophageal cancer, the most common sites of distant metastases are the lungs, liver, and distant lymph nodes, all of which are evaluable sites by FDG-PET. FDG-PET has been shown to have a higher sensitivity for detection of metastatic disease compared to CT in combination with EUS.¹⁷ Flamen et al reported both higher accuracy (82% vs. 64%, P = .004) and sensitivity (74% vs. 47%) for FDG-PET when compared to CT and EUS for the detection of distant disease. The use of FDG-PET in this study resulted in upstaging 15% of the patients from M0 to M1 disease and in the downstaging of 7% of the patients from M1 to M0 disease.¹⁷ In patients upstaged by FDG-PET, the majority had advanced primary tumors (clinical T3 or greater). Metaanalysis of FDG-PET in detecting distant metastases, revealed a pooled sensitivity and specificity of 67% (95% CI, 58% to 76%) and 97% (95% CI, 90% to 100%), respectively.¹⁹

It is notable that the relatively good sensitivity associated with FDG-PET in detecting metastases diminishes with decreasing size of the lesion, with an approximate minimum threshold of 1 cm.⁸ There is some variability, however, in that very small lesions can be detected, assuming they have relatively large tracer uptake, and larger lesions with low or no uptake can be missed. There is also the potential for false-positive detection of metastatic disease using FDG-PET. The main causes are benign tumors, inflammatory processes, and increased uptake in normal tissue like muscle, bowel, and brown adipose. While false positives using FDG-PET are relatively rare, each positive should be confirmed by biopsy or another imaging treatment before the treatment plan is altered.

RESTAGING

Assessing Response to Neoadjuvant Treatment

In many cases, treatment of localized esophageal cancer relies on a multimodality approach. Routinely, in more locally advanced cases, use of chemotherapy in combination with radiation is used in a neoadjuvant fashion. The aims of neoadjuvant treatment are to improve the control of both local and distant disease, while allowing for a more complete resection. Assessing response to chemoradiation is important in providing information for planning further treatment. It has been shown that in patients with locally advanced disease undergoing esophagectomy, only those with completely negative surgical margins have any chance of long-term survival.¹⁵ Additionally, there are many patients with more advanced disease or comorbid conditions for whom chemoradiotherapy is definitive management. Tumor response with chemoradiotherapy is highly variable and nonresponsiveness predicts for a poor prognosis.²¹ Therefore, it is important to identify those patients not responding to therapy early in their course of treatment, so as to potentially spare them the cost and toxicity of avoidable therapy.

In the past, CT, MRI, and EUS have all been employed to assess response to chemoradiotherapy, with mixed success. As a group, these modalities are frequently unable to differentiate between residual disease and post-therapy changes, and there is a delay of many weeks after therapy before response can be properly assessed.³ Additionally, there is some question regarding the relationship of anatomical imaging and pathologic response, with data suggesting no significant correlation between the two.²²

Westerterp et al in a review of the literature, compared FDG-PET, CT, and EUS in diagnostic accuracy for assessing response to neoadjuvant therapy.²³ In this comparison, 4 studies with CT, 13 with EUS, and 7 with FDG-PET met the criteria for evaluation. The accuracy was found to be 54% for CT, 86% for EUS, and 85% for FDG-PET. The accuracy of CT was significantly lower than that of FDG-PET (P = .006) and of EUS (P = .003), while the accuracy of FDG-PET and EUS were similar (P = .839). Notably, 6% of patients were not evaluable by EUS. The authors concluded that FDG-PET is a promising noninvasive tool for assessing response to neoadjuvant therapy.²³

FDG-PET offers a functional alternative to anatomical imaging in assessing response to treatment (Figure 2). Weber et al reported using FDG-PET to assess response to neoadjuvant chemotherapy in gastroesophageal (GE) junction tumors.²⁴ In this study of 40 patients, FDG-PET was performed initially and 2 weeks after induction of chemotherapy, and the results were compared to endoscopic response measured 3 months after completion of therapy or to pathologic response in patients undergoing definitive surgery. The study found that responders had a significantly greater decrease in tumor FDG uptake compared with nonresponders to therapy. The authors found that a reduction of tracer uptake of 35% was a reasonable cutoff value for differentiating between responders and nonresponders. Using this cutoff value, FDG-PET had a sensitivity of 93% and specificity of 95% for detecting tumor response. Histopathologic complete or near complete tumor regression was found in 53% of the responders, but only in 5% of the FDG-PET nonresponders.²⁴

Figure 2.

On the left is an initial staging FDG-PET showing increased uptake in the region of a known distal esophageal adenocarcinoma. On the right is a restaging FDG-PET 4 weeks post neoadjuvant chemotherapy in combination with radiation, showing no abnormal uptake.

Similarly, a study by Brücher et al found a cutoff of 52% reduction in tracer uptake predicted response to chemoradiotherapy, with a sensitivity of 100% and specificity of 55%.²⁵ The positive predictive value for pathologic response was 72%, while the negative predictive value was found to be 100%. The relatively low specificity was felt to be related to treatment-related inflammation. In this study, post-treatment FDG-PET was performed 3 weeks after the completion of radiotherapy.

A study by Flamen et al also found that FDG-PET can predict “major response” to chemoradiation.²⁶ FDG-PET scans were performed before and 4–6 weeks after completion of therapy and were found to have a predictive accuracy for major response of 78%. Sensitivity and specificity were found to be 71% and 82%, respectively. A discrepancy between FDG-PET response at the primary site and at regional sites was found in 22% of patients in this study. Of those with an inconsistency of response between sites, 88% were nonresponders, meaning that the least responding lesion determines the overall responsiveness and, ultimately, the prognosis.²⁶ An overestimation of response was due to false-negative PET findings from the persistence of residual microscopic disease. Table 1 summarizes several trials evaluating FDG-PET in restaging following neoadjuvant therapy.

Table 1.

Timing and results of trials evaluating FDG-PET in restaging following neoadjuvant therapy for esophageal cancer.

Study	No. pts	Neoadjuvant therapy	Timing of restaging PET	Threshold reduction in uptake	Sensitivity in predicting response	Specificity in predicting response
Flamen26	36	Chemo plus RT	4–6 weeks after completion	80%	71%	82%
Brücher25	27	Chemo plus RT	3–4 weeks after completion	52%	100%	55%
Weber24	40*	Chemo alone	2 weeks after initiation	35%	93%	95%
Wieder37	27	Chemo plus RT	2 weeks after initiation	30%	93%	88%

^*Study evaluated patients with tumors of the gastroesophageal junction

Abbreviations: PET = ¹⁸F-fluoro-2-deoxyglucose positron emission tomography; Chemo = chemotherapy; RT = external beam radiation

Microscopic disease falls below the threshold of detection by FDG-PET, making prediction of complete response difficult.¹² Underestimation of response by FDG-PET was primarily due to radiation-induced inflammation at the primary site.³⁵ Another study found that when quantitative FDG-PET is used post therapy, it is not possible to differentiate between residual tumor and chemoradiotherapy-induced esophagitis.²⁷ Post-chemoradiotherapy inflammation and esophagitis are common findings and, in many instances, obscure response evaluation by FDG-PET, making the timing of treatment evaluation important in maximizing its prognostic value.³

Assessing Recurrent Disease

Following definitive management for esophageal cancer, anatomical diagnostics can be relatively inaccurate due to the anatomic changes associated with chemoradiotherapy and/or surgery.¹² Because of the importance of follow-up diagnostics and the shortcomings of conventional anatomic imaging, FDG-PET has been studied for use in the post-treatment setting. In one particular study, FDG-PET was shown to be a highly sensitive tool for diagnosing recurrent disease, both locally and distantly.²⁸ In this study, the addition of FDG-PET to the post-treatment regimen yielded additional information in 27% (11/41) of patients. Twelve percent (5/41) of patients with negative or unclear findings by conventional imaging were found to have recurrent disease by FDG-PET, while another 12% (5/41) of patients with local recurrence were found to have distant disease by the addition of FDG-PET. The findings of this study suggest that FDG-PET is an important tool in the follow-up period, potentially allowing for earlier disease detection and management.²⁸

The Prognostic Value of FDG-PET

FDG-PET has shown to be effective in both staging and restaging of esophageal cancer, and it appears, as alluded to earlier, to hold some prognostic value for the patient, also. In a study by Brücher et al, 27 patients with squamous cell carcinoma of the esophagus underwent an initial FDG-PET and another scan 3–4 weeks post chemoradiotherapy —tumor response was found to be a significant prognostic factor.²⁵ In this study a cutoff of 52% reduction in metabolic activity was used to classify a tumor as having responded to therapy. Responders to therapy had a median overall survival of 22.5 months compared to 8.8 months for nonresponders. A related study of 36 patients yielded similar results, with a median survival of 16.3 months for patients responding to therapy by FDG-PET, and only 6.4 months for those patients classified as nonresponders (P = .05).²⁶ Evaluation of a series of 83 patients undergoing neoadjuvant chemoradiation followed by restaging FDG-PET prior to surgery found that FDG-PET SUV (standardized uptake values) of ≥4 was predictive of decreased survival, with a 2-year survival rate of 33% vs. 60% in patients with SUV <4 (P = .01).²⁹

There is also information obtained at initial FDG-PET staging that appears to be of prognostic value. The intensity of the primary tumor at presentation has been found to be of prognostic value. A study of 89 patients found that the higher the maximum FDG-PET SUV at diagnosis, the more likely patients were to have poorly differentiated tumors and advanced stage.³⁰ In addition, using a cutoff for maximum SUV of 6.6, the 4-year survival in patients with a maximum SUV of ≤6.6 was 89% vs. 31% for patients with maximum SUV >6.6 (P < .001).

Hong et al tested the hypothesis that the number of FDG-PET abnormalities found on initial staging of patients with locoregional esophageal cancer would correlate with overall and disease-free survival.³¹ In this study, patients with stage II or III esophageal cancer underwent a baseline FDG-PET, followed by chemoradiotherapy and surgery. In the 47 patients who underwent neoadjuvant chemoradiotherapy followed by surgery, the number of abnormalities detected by FDG-PET was significantly associated with both overall survival (P = .02) and disease-free survival (P = .04). Patients with more than one lesion by FDG-PET had a hazard ratio for death of 4.49 (reference, 1 lesion by FDG-PET). In a multivariate analysis, the number of FDG-PET abnormalities was independently predictive of overall survival (P = .03), while other variables like the SUV of the primary tumor, peak SUV, and total SUV did not correlate with response, overall survival, or disease-free survival. The authors concluded that a baseline FDG-PET has prognostic value in the setting of recently diagnosed esophageal cancer.³¹

Based on the available data, it appears that response can be assessed either early (after the first or second cycle of treatment in the case of induction chemotherapy), or late (several weeks after completion of treatment in the case of neoadjuvant chemoradiotherapy). Depending on the timing of the scan, different information is obtained. ¹² Using FDG-PET shortly after initiation of treatment evaluates the responsiveness of a tumor to a certain treatment and can provide a method of early detection of non-responding tumors. When using FDG-PET after treatment, it detects residual viable tumor following completion of therapy and can be of prognostic value.¹² In terms of detecting responders to therapy, the choice of a cutoff value for response is dependent on the treatment strategy of the treating team. If the goal of the scan is to have a high sensitivity for detecting nonresponders, then a high response value is used, while if the goal is to have a high sensitivity for detecting responders, then the cutoff is set to a lower value.¹²

EMERGING FUNCTIONS OF FDG-PET

The Role of PET in Radiotherapy Planning

FDG-PET could play a role in defining radiation tumor volumes by providing additional information to the planning CT and diagnostic EUS.³² Mourneau-Zabotto et al published a study of 34 patients where FDG-hybrid PET was fused to planning CT for conformal planning.³³ Prior to initiation of chemoradiotherapy, each patient underwent a planning CT simulation and FDG-hybrid PET in the same treatment position, with the PET images coregistered using 5 fiducial markers. Targets were then identified by CT and the PET data were then added to modify the volumes. FDG-hybrid PET identified metastatic disease in two patients, thus changing their treatment course. Additionally, the gross tumor volume (GTV) was decreased in 35% (12/34) of patients after adding the information from PET, and was increased in 21% (7/34) of patients after modification for PET. In 17% (6/34) of patients, the GTV was considerably modified (≥25%) after incorporating additional information provided by PET. In over half the patients (18/34), the change in the GTV ultimately led to modification in the treatment volumes of the patients. The authors concluded that FDG-PET fusion with CT affected radiotherapy planning, however its effect on treatment outcomes remains unknown.³³

A similar study by Konski et al evaluated the role of FDG-PET and EUS in GTV and target delineation.³⁴ In this study, 25 patients with esophageal cancer underwent FDG-PET scan after a planning CT in the same treatment position. The mean length of the primary tumor was 5.4 cm by PET, 6.77 cm by CT, and 5.1 cm for the 22 patients undergoing EUS. The size of the primary tumor was significantly larger when measured by CT compared to FDG-PET (P = .0063).³⁴ EUS was able to detect significantly more regional adenopathy than both FDG-PET and CT. This study again demonstrated that additional information provide by FDG-PET can influence treatment planning. As PET/CT continues to become more integrated, there is likely to be even greater refinement in radiotherapy planning for locally advanced esophageal cancers.

FDG-PET/CT

While FDG-PET provides physiologic information not obtainable through traditional anatomic diagnostic methods, the anatomic data it accumulates are of limited value. Recently, FDG-PET scans have been combined with CT in a single diagnostic study with almost identical patient positioning where the FDG-PET and CT images are fused (Figure 3).³⁵ This technology is able to provide accurate functional and anatomic information, and has been shown to improve tumor staging accuracy in patients with non–small-cell lung cancer.³⁶ This improved accuracy will potentially allow for improved diagnostic accuracy in esophageal cancer, as well. In patients with questionable nodal metastases in close proximity to the primary tumor, the combination of FDG-PET/CT can theoretically better elucidate metastases than either FDG-PET or CT alone. FDG-PET/CT should have a role in restaging, as well, by simultaneously providing information about tumor size and activity.⁸ FDG-PET/CT can also be used in radiotherapy planning, aiding the radiation oncologist in identifying the extent of the primary tumor and providing additional information about indeterminate regional metastases.

Figure 3.

On the top are axial slices of a CT and FDG-PET demonstrating an abnormality. PET/CT fusion on the bottom left correlate the area of increase uptake on FDG-PET with the anatomic information provided by CT.

DISCUSSION

Functional imaging in the form of FDG-PET has shown to be an effective diagnostic modality in esophageal cancer. It now has a defined role in initial staging and continues to be studied in the evaluation of neoadjuvant therapy response and in routine follow-up after definitive therapy. Additionally, its use in combination with anatomic imaging modalities, such as CT, provides additional diagnostic information that can be used by radiation oncologists in target delineation and planning. FDG-PET has also been shown to have prognostic value as well that can be applied to patient management and aid in the development of emerging therapies.