Metabolic Characteristics of Advanced Biliary Tract Cancer Using ¹⁸F-Fluorodeoxyglucose Positron Emission Tomography and Their Clinical Implications

This study evaluated ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography findings in patients with advanced biliary tract cancer (BTC) prior to palliative chemotherapy, such as SUV_max, the number of lesions and organs with FDG uptake, pathologic findings, and clinical outcomes. The metabolic characteristics of advanced BTC differ according to primary origin and histology, which could be prognostic factors for overall survival and progression-free survival in advanced BTC.

Abstract

Background.

In advanced biliary tract cancer (BTC), the metabolic landscape has not been evaluated by ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography (PET) yet. Furthermore, reports of the clinical implications of these metabolic features are limited. We aimed to evaluate the metabolic features and their clinical relevance in advanced BTC using ¹⁸F-FDG PET.

Patients and Methods.

We consecutively enrolled patients with advanced BTC who underwent ¹⁸F-FDG PET prior to palliative chemotherapy between 2003 and 2013. We evaluated the findings of PET, such as SUV_max, the number of lesions and organs with FDG uptake, pathologic findings, and clinical outcomes.

Results.

A total of 106 patients were enrolled: (53 intrahepatic cholangiocarcinoma [ICC], 7 extrahepatic BTC, 30 gallbladder cancer [GB Ca], and 16 ampulla of Vater cancer [AoV Ca]). The median SUV_max differed according to the primary origin (ICC, 9.10; extrahepatic BTC, 5.90; GB Ca, 9.10; and AoV Ca, 6.37; p = .008) and histologic differentiation (well differentiated, 4.95; moderately differentiated, 6.60; poorly differentiated, 14.50; p = .004). Patients in the high metabolic group (SUV_max of ≥7.5) had more poorly differentiated histology and more organs and lesions with FDG uptake than did those in the low metabolic group (SUV_max of <7.5). The low metabolic group had a significantly longer OS (11.4 vs. 7.4 months, p = .007) and PFS (6.6 vs. 4.3 months, p = .024) than high metabolic group. In multivariate analysis, SUV_max was a significant prognostic factor for overall survival (OS; p = .047) and progression-free survival (PFS; p = .039).

Conclusion.

Metabolic characteristics of advanced BTC differ according to primary origin and histology. These metabolic features could be prognostic factors for OS and PFS in advanced BTC.

Implications for Practice:

The metabolic landscape of advanced biliary tract cancer and its clinical meanings have not yet been comprehensively studied. This study shows that metabolic characteristics of advanced biliary tract cancer differ significantly according to primary origin and histology. Moreover, this metabolic activity is associated with patients’ outcomes, including overall survival and progression-free survival. This study supports tumor heterogeneity in terms of cancer metabolism in biliary tract cancer.

Introduction

Biliary tract cancers (BTCs) are relatively rare tumors with poor prognosis. BTCs account for less than 2% of all malignancies in the West [1], but they are more common in Korea and Japan, accounting for approximately 4% of malignancies [2].

BTC consists of intrahepatic cholangiocarcinoma (ICC), gallbladder cancer (GB Ca), extrahepatic biliary tract cancer (extrahepatic BTC), and ampulla of Vater cancer (AoV Ca) [3]. There are several reports of biologic differences according to the site of origin of these tumors [4, 5]. Among many biologic characteristics, the metabolic landscape of biliary tract cancer has not been reported yet. Currently, interest in tumor heterogeneity prevails, regarding diverse characteristics, including cancer metabolism. It is not known whether tumor metabolism differs by the origin of BTCs. In fact, the metabolic landscape of cancer differs depending on cancer cell characteristics. In some tumor types, the metabolic landscape can be evaluated by ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography (PET) and shows metabolic heterogeneity [6–8].

PET, using the radiolabeled glucose analog ¹⁸F-FDG, can detect the enhanced glycolysis of cancer cells and has proven valuable for diagnosing and staging of cancer, detecting recurrences and assessing response to therapy in numerous malignant disorders [9]. The extent of uptake of ¹⁸F-FDG by a cancer, which usually correlates with the degree of tumor proliferation and metabolism, is evaluated by a semiquantitative measurement, the standardized uptake value (SUV) [10]. The prognostic significance of ¹⁸F-FDG uptake, measured using the maximal standardized uptake value (SUV_max), has been evaluated in lung, endometrial, cervical, and renal cell carcinomas [11–18]. Studies of the clinical significance of metabolic activity in advanced biliary tract cancers are limited. The aim of this study was to describe the metabolic landscape of biliary tract cancers and to determine the clinical importance of cancer metabolic activity for prognosis of an advanced biliary cancer.

Patients and Methods

Patients

We retrospectively enrolled consecutive patients with advanced biliary tract cancer who underwent ¹⁸F-FDG PET imaging at the time of diagnosis of an advanced biliary tract cancer at the Seoul National University Hospital (Seoul, Republic of Korea) between 2003 and 2013. All patients fulfilled the following criteria: (a) biliary tract cancer was histologically confirmed; (b) biliary tract cancer was unresectable, metastatic, and recurrent; (c) the patient had received at least one cycle of palliative chemotherapy; and (d) the patient underwent PET prior to palliative chemotherapy.

The medical records of each patient were reviewed, and the following data were abstracted: sex, age, anatomical origin of tumor, tumor localization, pathology, progression-free survival (PFS), and overall survival (OS).

¹⁸F-FDG PET

The patients were instructed to fast for at least 8 hours to ensure a serum glucose level of less than 7.8 mmol/L (140 mg/dL). Then ¹⁸F-FDG (5.18 MBq/kg of body weight) was intravenously injected. One hour after the injection, all scans were performed with one of two PET systems (Biograph from Siemens Medical Solutions, Hoffmann Estates, IL, http://www.medical.siemens.com; or Gemini from Philips Medical Systems, Cleveland, OH, http://www.healthcare.philips.com) in three-dimensional mode from the base of the skull to the proximal thigh. For quantitative analysis of the ¹⁸F -FDG uptake, a region of interest (ROI) was placed over the most intense area of ¹⁸F-FDG accumulation. The activity concentration within the ROI was determined and expressed as the SUV calculated as follows:

$$\text{SUV}=\text{Region’s}\hspace{0.17em}\text{radioactivity}\hspace{0.17em}\text{concentration}\left(\text{Bq}/\text{mL}\right)/\left[\text{Injected}\hspace{0.17em}\text{dose}\left(\text{Bq}\right)/\text{Patient’s}\hspace{0.17em}\text{weight}\left(\text{g}\right)\right]\mathrm{.}$$

The SUV_max, defined as the highest SUV in the pixel with the maximal SUV within the ROI, was measured and recorded for the focal areas of uptake. SUV_max values were standardized for the injected dose and the patient’s weight. For the purpose of this study, we evaluated whether the SUV_max of both primary and metastatic lesions was associated with the number of organs and lesions with FDG uptake, pathologic findings, and other clinical outcomes, including response to chemotherapy.

Statistical Analyses

Nonparametric approaches, such as Wilcoxon rank-sum test and Kruskal-Wallis test, were used to evaluate the relationship between the SUV_max distribution and pathological factors and the primary tumor origin. The log rank test was used to select the most discriminative SUV_max cutoff value. The χ² test was used to compare baseline characteristics between two groups of patients: those with SUV_max values less than the cutoff value and those with SUV_max greater than or equal to the cutoff value.

The OS was calculated from the first day of palliative chemotherapy to the day of death, and the PFS was calculated from the first day of palliative chemotherapy to the first day of progression or last follow-up. The OS and PFS were estimated using the Kaplan-Meier method. The log rank test was used to assess the differences in OS and PFS across groups. The effect of patients’ characteristics and other prognostic factors, such as PET SUV_max, on survival was evaluated using Cox proportional hazard models (univariate and multivariate analysis) and the resultant hazard ratio (HR) and 95% confidence intervals (95% CIs). A p value of .05 or less was considered statistically significant. All analyses were performed using SPSS software for Windows (version 20; SPSS software, IBM Corp., Armonk, NY, http://www-01.ibm.com/software/analytics/spss/).

Ethics

The study protocol was reviewed and approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H-1306-109-500). The study was conducted according to guidelines (Declaration of Helsinki) for biomedical research.

Results

Patients and Treatments

The study included 106 patients with advanced BTC. Patient characteristics are shown in Table 1. The study cohort consisted of 75 (70.8%) males and 31 (29.2%) females. The median age was 61 years (range, 27–81 years). There were 53 patients with ICC (50%), 30 patients with GB Ca (28.3%), 7 patients with extrahepatic BTC (6.6%), and 16 patients with AoV Ca (15.1%). When advanced BTC was diagnosed, the median (range) of laboratory values were: carcinoembryonic antigen (CEA), 2.6 ng/mL (0–2,135 ng/mL), carbohydrate antigen 19-9 (CA 19-9) level, 80 U/mL (1–240,000 U/mL); total bilirubin, 0.7 mg/dL (0.1–9.6 mg/dL); and albumin, 3.9 mg/dL (2.2–4.8 mg/dL).

Table 1.

Baseline characteristics of patients

In total, 54 (51.9%) patients received gemcitabine plus platinum chemotherapy, and 45 (42.3%) patients received fluoropyrimidine plus platinum chemotherapy. The median follow-up duration was 7.8 month (95% CI: 5.8–9.8). The overall survival of all patients was 8.3 months (95% CI: 5.7–10.8), and progression-free survival was 4.9 months (95% CI: 3.4–6.3).

SUV_max Distribution

The median SUV_max of primary and metastatic lesions combined was 7.8 (range, 0–44) (Table 2). Among all 106 patients, only one patient with ICC showed no FDG uptake (PET negative) in the cancer lesion seen on CT. The OS of this patient was 9.2 months.

Table 2.

Distribution of SUV_max and FDG uptake among lesions and organs

Of the 106 patients, 48 (45.3%) showed a higher SUV in the primary tumor site than in metastatic lesions, and 57 (53.8%) showed a higher SUV in metastatic sites than the primary tumor origin. The median number of organs with FDG uptake, including primary and metastatic organs, was 3 (range, 1–6), and the median number of lesions with FDG uptake was 4 (range, 0–19).

SUV_max According to Patient Characteristics

SUV_max did not differ by age, sex, ECOG PS, CA 19-9, total bilirubin, albumin, and objective response rate (Table 3). However, SUV_max differed significantly by the primary tumor origin, histologic differentiation, and CEA level (CEA <10 vs. ≥10).

Table 3.

Comparison of SUV_max according to patient characteristics

The SUV_max was higher in patients with ICC and GB Ca than in patients with extrahepatic BTC and AoV Ca (median SUV_max in ICC, 9.10; GB Ca, 9.10; extrahepatic BTC, 5.90; and AoV Ca, 6.37; p = .008). The distributions of SUV_max in each subtype of BTCs are shown in supplemental online Figure 1. The SUV_max of BTCs with poorly differentiated histology was higher than in BTCs with well or moderate differentiated histology (median SUV_max of well differentiated BTCs, 4.95; moderately differentiated, 6.6; and poorly differentiated, 14.50; p = .004).

Comparisons Between High- and Low-Metabolism Groups

Based on the analysis using the log-rank test to select the most discriminative SUV_max value in relation to survival, the cutoff value 7.5 was used to categorize high/low metabolism of the BTCs (supplemental online Figure 2). When we further analyzed this SUV_max cutoff in each primary tumor subtype, this value of 7.5 was also defined in GB Ca and AoV Ca except ICC (supplemental online Table 1).

Patients with a SUV_max of <7.5 (low metabolic group) had a significantly longer OS than those with a SUV_max of ≥7.5 (high metabolic group) (11.4 vs. 7.4 months, p = .007) (Fig. 1). The frequency of patients with ICC and GB Ca was greater in the high metabolism than in the low-metabolism group, whereas the frequency of patients with AoV Ca was higher in the low-metabolism group (Table 4). Similarly, the patients with poorly differentiated pathology were more frequent in the high-metabolism group. Those patients with more organs (≥3) and more lesions (≥4) with FDG uptake were more frequent in the high-metabolism group. The metabolism groups did not differ by age, sex, ECOG performance status (PS), CA 19-9 level, total bilirubin, albumin, and objective response rate. However, patients with a high CEA level (≥10 ng/mL) were relatively more frequent in the high-metabolism group (p = .024). When we further analyzed the association of SUV_max and OS in each subtype of tumors, the GB Ca and AoV Ca showed the significant difference in OS according to SUV_max (p = .044 and p = .036, respectively). However, in ICC, the difference of OS according to SUV_max was not observed (p = .69). In the case of extrahepatic BTC, this analysis was difficult because of a limited number of patients (supplemental online Table 1).

Figure 1.

Overall survival (OS) according to metabolic activity. Patients with low metabolism (SUV_max < 7.5) had a significantly longer OS than those with high metabolism (SUV_max > 7.5) (11.4 vs. 7.4 months, p = .007).

Abbreviation: PET, positron emission tomography.

Table 4.

Comparison of patient characteristics between the high/low-metabolism groups

Prognostic Value of SUV_max

In analysis of prognostic factors for survival, the high metabolism was associated with shorter OS (HR, 1.77; 95% CI, 1.00–3.13, p = .047) (Table 5). The tumor origin, CEA level, and response to first-line chemotherapy were also independent prognostic factors for OS. Patients with the high metabolism had a significant shorter PFS than those with low metabolism (6.6 vs. 4.3 months, p = .024) (Fig. 2).

Table 5.

Analysis of prognostic factors for overall survival

Figure 2.

Progression-free survival according to metabolic activity. Patients with low metabolism had a significant longer PFS than those with high metabolism (6.6 vs. 4.3 months, p = .024).

Abbreviation: PET, positron emission tomography.

Discussion

There are several reports on the metabolic landscape of ¹⁸F-FDG PET that show the metabolic heterogeneity of cancers [6–8]. In breast cancer, FDG uptake was independently associated with subtypes of invasive breast cancer [6]. In endometrial cancer, the preoperative SUV_max was significantly correlated with tumor histological grade [7]. In lung cancer, the SUV_max values of adenocarcinoma are lower than those of squamous cell types [8]. However, there are no reports of the metabolic landscape of BTC using ¹⁸F-FDG PET. Our study is the first report using SUV_max to elucidate the metabolic characteristics of BTC. We found that the metabolic activity differs according to the origin of BTCs and cancer cell histological differentiation. Furthermore, this metabolic activity is associated with the prognosis of advanced BTC.

Various studies have assessed whether the tumor SUV can be used to predict the survival of patients with malignancies [9, 12, 14–18]. However, data are limited about the potential of PET to give the information on the prognosis of BTC. Furukawa et al. [19] reported that SUV analysis of FDG PET was one of the useful predictors of overall survival for BTC. However, most of the patients in that study had operable BTC, and PET imaging was performed before surgery. Their multivariate analysis did not prove that SUV_max was an independent prognostic factor. Kitamura et al. [20] reported that the SUV of the primary lesion is a useful prognostic factor in patients with extrahepatic BTC. However, this study was limited to extrahepatic BTC and analyzed the SUV of only the primary lesion. Therefore, our study is the first report of an investigation of SUV_max (in both primary and metastatic lesions) as a potential prognostic factor for OS and PFS in unresectable advanced BTC.

Based on SUV_max values, the metabolic activity among ICC, GB Ca, extrahepatic BTC, and AoV Ca differed significantly. This finding means that tumor metabolism varies according to the tumor origin of BTC and supports the existence of biological differences among BTCs of different tumor origin. Importantly, SUV_max was related to the differences in OS of all BTC. ICC had the shortest overall survival and the highest SUV_max, whereas AoV Ca had the longest overall survival and the lowest SUV_max. In concordance with this finding, patients with high metabolism (SUV_max of ≥7.5) had a significantly shorter OS and PFS than those with low metabolism (SUV_max of <7.5). When we further analyzed the association of SUV max and OS in each subtype of tumor, the GB Ca and AoV Ca showed a significant difference in OS according to SUV_max (p = .044 and p = .036, respectively). However, in ICC, the difference of OS according to SUV_max was not observed (p = .69). In case of extrahepatic BTC, this analysis was difficult because of a limited number of patients (supplemental online Table 1).

A previous study demonstrated a significant positive correlation with glucose transporter-1 (GLUT-1) expression and SUV_max in cancer [7]. Increased GLUT-1 expression is correlated with increased glucose metabolism and use. Increased glucose metabolism in cancer cells was shown to be related to hypoxic conditions, angiogenesis, loss of tumor-suppressor genes, and oncogene activation [21]. Our study showed that a higher baseline SUV_max was associated with more widespread FDG uptake, both in more organs (≥3) and in more lesions (≥4). This association suggests that tumors with a higher SUV_max are more likely to metastasize. In addition, SUV_max differed significantly by CEA level, an indicator of tumor burden [22]. This suggests that the greater the tumor burden, the higher the SUV_max value. This finding is consistent with reports that SUV_max is significantly related to the pathological stage [19, 23].

In the present study, a SUV_max cutoff value of 7.5 was selected using the lowest log-rank test P value (supplemental online Figure 2), which was the same procedure reported by other studies [11, 14]. When we further analyzed this SUV_max cutoff in each primary tumor subtype, this value of 7.5 was also defined in GB Ca and AoV Ca but not ICC (supplemental online Table 2). Several different methods are used to select cutoff values for measures that are used to predict survival in BTC and other cancers [12, 23–26]. In a PET study of BTC, Furukawa et al. [19] reported that a SUV_max cutoff of 6.3, selected using receiver operating characteristic curves, had discriminative ability in 69 patients. Kitamura et al. [20] reported that a cutoff of 5.7, selected using the median value of SUV_max, had discriminative ability in 73 patients.

Consequently, we need to recognize that SUV is a semiquantitative index, and its measurement can vary from one center to another. SUV is dependent on parameters such as time after injection, ROI size, plasma glucose/insulin concentration, body weight, instrumental factors, partial volume effect, and the resolution of the PET unit [27]. Differences in any of these parameters might account for differences in the threshold SUV among studies.

Some authors reported that differences in cutoff values of PET SUV_max are attributable to different cancer subtypes [11]. Others claim that more analyses with larger numbers of cases or independent studies are needed to determine the most appropriate cutoff point and thereby decrease the risk of false-positive results [14]. Accordingly, further investigations in larger populations or a validation set or prospective study are needed to evaluate the most discrimination SUV_max cutoff values for OS and PFS.

Recently, molecular biological analysis of BTC has shown that the epidermal growth factor receptor, human epidermal growth factor receptor 2, vascular endothelial growth factor, and c-Met are related to the biologic pathogenesis [28–30]. There are no data about metabolic differences according to molecular characteristics of BTC. Further studies on that point are needed.

The usefulness of PET should be addressed by the frequency of new lesions detected on PET not otherwise seen on conventional imaging such as CT and, more importantly, by the fact that this finding on PET could change the management of patients. In our study, among all enrolled 106 patients, 23 patients showed the PET-positive/CT-negative lesions. The management was changed based on PET finding in 5 of 23 patients. (supplemental online Table 3). Most of the lesions were lymph node and bone metastasis. This is in accordance with previous reports about detection of L.N and the occult metastatic site of PET [22, 31].

Otherwise, there were 7 patients with non-FDG avid lesions detected on CT. (supplemental online Table 4). In 1 ICC patient, there was no FDG uptake at all in all lesions noticed on CT; that is, the SUV_max was 0. In this patient, the treatment decision was made based on CT finding. In 6 patients, we found non-FDG avid lesions noticed on CT in each patient. In other words, 6 patients had a combination of FDG avid lesion and one non-FDG avid lesion simultaneously. In these 6 patients, the treatment decision was not influenced by the non-FDG avid lesions.

From another point of view, the prognostic significance of PET monitoring during treatment is also important thing to be addressed. To do that, we should check the PET at predefined time point after chemotherapy. In our study, among all 106 patients, 8 were evaluated with pre- and post-first-line chemotherapy PET. The post-first-line chemotherapy PETs were obtained at the disease progression with first-line chemotherapy. Seven patients showed the decrease of SUV_max after chemotherapy (supplemental online Table 5). It should be cautiously interpreted because of retrospective design and small patient number.

Conclusion

Metabolic characteristics of advanced BTC differ according to its primary origin and histology. Metabolic features such as SUV_max could be prognostic factors for OS and PFS in advanced BTC.

See http://www.TheOncologist.com for supplemental material available online.

Author Contributions

Conception/Design: Kyoung-Min Cho, Do-Youn Oh

Provision of study material or patients: Kyoung-Min Cho, Do-Youn Oh, Tae-Yong Kim, Kyung Hun Lee, Sae-Won Han, Seock-Ah Im, Tae-You Kim, Yung-Jue Bang

Collection and/or assembly of data: Kyoung-Min Cho, Do-Youn Oh

Data analysis and interpretation: Kyoung-Min Cho, Do-Youn Oh

Manuscript writing: Kyoung-Min Cho, Do-Youn Oh

Final approval of manuscript: Kyoung-Min Cho, Do-Youn Oh, Tae-Yong Kim, Kyung Hun Lee, Sae-Won Han, Seock-Ah Im, Tae-You Kim, Yung-Jue Bang

Disclosures

Seock-Ah Im: AstraZeneca (C/A, RF), Novartis (C/A). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board