Prospective Evaluation of the Clinical Implications of the Tumor Metabolism and Chemotherapy-Related Changes in Advanced Biliary Tract Cancer

Patients

Seventy-five 75 patients were enrolled, and their characteristics are shown in Table 1. The median age was 64 y (range, 46–83 y), and 43 (57.3%) patients were men. Twenty-eight (37.3%) patients had gallbladder cancer, 22 (29.3%) had IHCC, 19 (25.3%) had EHCC, and 6 (8.0%) had AoV cancer. The Eastern Cooperative Oncology Group performance status was 0 in 20 (26.7%) patients. Moderately differentiated adenocarcinoma was identified to be the most common pathology (38 patients, 50.7%). Immunohistochemistry showed positive expression of c-Myc in 12 (30.0%) patients among the 40 patients who were analyzed. Thirty-five (46.7%) patients had initially unresectable diseases, and the remaining (53.3%) had recurrent disease. All patients received gemcitabine/cisplatin as first-line palliative chemotherapy. The median follow-up duration was 6.8 mo (range, 1.0–27.2 mo). The median PFS was 5.6 mo (95% confidence interval [CI], 4.4–6.8), and the median OS was 13.2 mo (95% CI, 7.1–19.3). There were 4 cases with discrepancies between 2 readers. All cases were evaluated as stable disease by the first reader and progressive disease by the second reader. Three cases were finally determined as stable disease (based on tumor sum) and 1 case as progressive disease (based on a new lesion) after repeated evaluation and discussion.

SUV_max Distribution at Baseline and Its Changes During Chemotherapy

The distribution of the median SUV_max at baseline among all lesions (combined primary and metastatic lesions), primary lesions, and metastatic lesions were 8.6 (range, 1.0–20.5), 3.9 (range, 1.0–20.5), and 5.8 (range, 1.0–15.2), respectively. The median SUV_max reductions among all lesions at the best metabolic response and during the initial evaluation were 9.5% (range, −162.5%–88.8%) and 5.2% (range, −162.5%–85.3%), respectively. The median number of organs and lesions with ¹⁸F-FDG uptake were 2 (range, 0–5) and 2 (range, 0–41), respectively (Supplemental Table 1 [supplemental materials are available at http://jnm.snmjournals.org]; Figs. 1A and 1B). Seventy patients had ¹⁸F-FDG–avid tumors. In terms of primary tumors, the median SUV_max values at baseline among all lesions were 9.9, 7.5, 5.4, and 9.5 in gallbladder cancer, IHCC, EHCC, and AoV cancer, respectively (Fig. 1C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

(A) Distribution of initial SUV_max. (B) SUV_max reduction at best metabolic response and at first ¹⁸F-FDG PET evaluation. (C) Distribution of initial SUV_max according to primary tumor origin.

Cutoff Value of Initial SUV_max and Degree of Metabolic Reduction During Chemotherapy

The most optimal SUV_max cutoff values for predicting PFS and OS were determined by the log-rank test to be 9.0 and 10.0, respectively (Supplemental Table 2). On the basis of these results, we selected the SUV_max values as the discriminating values, respectively.

All cutoff values for SUV_max reduction at the best metabolic response were associated with PFS, and 20.0% were optimal cutoff values for predicting OS (Supplemental Table 2). On the basis of these results, we selected a SUV_max reduction of 20% as the discriminating value.

Comparison of Patient Characteristics Between High- and Low-Metabolism Groups

We divided the patients into high- and low-metabolism groups using the SUV_max cutoff value of 9.0 (Table 2). Gallbladder cancer was more common in the high-metabolism group, and EHCC was more common in the low-metabolism group. Poorly differentiated carcinoma and c-Myc–positive tumors were more frequently observed in the high-metabolism group. Initial metastatic disease was more frequent than recurrent disease in the high-metabolism group. The high-metabolism group showed high leukocytes and had more lesions and organs with ¹⁸F-FDG uptake. Age, sex, performance status, body mass index, carcinoembryonic antigen, carbohydrate antigen 19-9 levels, total bilirubin, albumin level, and treatment response did not differ between the 2 groups. The evaluation of metabolic activities according to patient characteristics showed similar findings (Supplemental Table 3).

Prognostic Implications of Initial SUV_max and Degree of Metabolic Reduction During Chemotherapy

The PFS was significantly shorter in patients of the high-metabolism group (3.8 vs. 7.0 mo; P = 0.002, respectively; Fig. 2A) and in the lesser SUV_max reduction group at the best metabolic response (3.9 vs. 8.8 mo, P < 0.001; Fig. 2B). Primary tumor origin, initial SUV_max, and the degree of SUV_max reduction were identified as independent prognostic factors for PFS in multivariate analysis (Table 3). Patients in the high-metabolism group (hazard ratio [HR], 4.09; 95% CI, 1.73–9.66; P = 0.001) and those with lesser reduction of SUV_max had worse outcomes (HR, 3.35; 95% CI, 1.55–7.20; P = 0.002).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

PFS according to initial SUV_max (A) and SUV_max reduction at best metabolic response (B). OS according to initial SUV_max (C) and SUV_max reduction at best metabolic response (D).

Patients with high metabolic activity had significantly worse OS (10.9 vs. 19.1 mo, P = 0.003; Fig. 2C). Patients with a lesser reduction of SUV_max at the best metabolic response showed a trend of worse OS (13.2 vs. 20.7 mo, respectively, P = 0.074; Fig. 2D). The initial SUV_max was identified as an independent prognostic factor for OS in multivariate analysis. Although statistically insignificant, SUV_max reduction and organs with ¹⁸F-FDP uptake were potentially associated with clinical outcome. Patients with high metabolic activity (HR, 2.61; 95% CI, 1.18–5.81; P = 0.019) and lesser SUV_max reduction (HR 1.96, 95% CI 0.91–4.20, P = 0.082) had worse OS (Table 4).

After the patients were divided into 4 groups depending on the initial SUV_max values and their changes at the best metabolic response, patients having high metabolic tumors who achieved lesser SUV_max reduction showed the worst survival outcomes, whereas those having low metabolic tumors who achieved greater SUV_max reduction showed the best survival outcomes (2.8 vs. 11.5 mo, P < 0.001 for PFS; 6.2 vs. 20.7 mo, P = 0.013 for OS; Fig. 3).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

PFS (A) and OS (B) after patients were divided into 4 groups by initial SUV_max and its response.

The analysis of the relationship between metabolic changes in SUV_max and their tumor response according to RECIST 1.1 showed that all patients who achieved partial response had reduced SUV_max values (Supplemental Fig. 1). However, the reduction of SUV_max was also observed in many of the patients who achieved stable disease status.

Prognostic Value of Initial SUV_max and Degree of Metabolic Reduction During Chemotherapy in Patients Who Achieved Disease Control According to RECIST 1.1

In patients who achieved disease control (complete response + partial response + stable disease) according to RECIST 1.1, patients in the high-metabolism and lesser SUV_max reduction groups had worse PFS than those in the low-metabolism (4.7 vs. 8.8 mo, P = 0.003; Supplemental Fig. 2A) and higher SUV_max reduction (5.3 vs. 9.2 mo, P = 0.013; Supplemental Fig. 2B) groups. Patients in the high-metabolism group had significantly worse OS (10.9 vs. 19.1 mo, P = 0.01; Supplemental Fig. 2C), and patients in the group with lesser metabolic rate reduction potentially showed worse OS (13.2 vs. 20.7 mo, P = 0.156; Supplemental Fig. 2D).