|Home | About | Journals | Submit | Contact Us | Français|
The use of gemcitabine as an adjuvant modality for cholangiocarcinoma (CC) is increasing, but limited data are available on predictive biomarkers of response. Human equilibrative nucleoside transporter 1 (hENT-1) is the major transporter involved in gemcitabine intracellular uptake. This study investigated the putative predictive role of hENT-1 localization in tumor cells of CC patients undergoing treatment with adjuvant gemcitabine.
Seventy-one consecutive patients with resected CC receiving adjuvant gemcitabine at our center were retrospectively analyzed by immunohistochemistry for hENT-1 localization in tumor cells. The main outcome measure was disease-free survival (DFS). Hazard ratios (HRs) of relapse and associated 95% confidence intervals (CIs) were obtained from proportional hazards regression models stratified on quintiles of propensity score.
Twenty-three (32.4%) cases were negative for hENT-1, 22 (31.0%) were positive in the cytoplasm only, and 26 (36.6%) showed concomitant cytoplasm/membrane staining. Patients with membrane hENT-1 had a longer DFS (HR 0.49, 95% CI 0.24–0.99, p = .046) than those who were negative or positive only in the cytoplasm of tumor cells. Notably, the association between DFS and membrane hENT-1 was dependent on the number of gemcitabine cycles (one to two cycles: HR 0.96, 95% CI 0.34–2.68; three to four cycles: HR 0.99, 95% CI 0.34–2.90; five to six cycles: HR 0.27, 95% CI 0.10–0.77).
hENT-1 localization on tumor cell membrane may predict response to adjuvant gemcitabine in CC patients receiving more than four cycles of chemotherapy. Further prospective randomized trials on larger populations are required to confirm these preliminary results, so that optimal gemcitabine-based chemotherapy may be tailored for CC patients in the adjuvant setting.
Gemcitabine is becoming an increasingly used adjuvant modality in cholangiocarcinoma (CC), but limited data are available on predictive biomarkers of response. In this study, patients receiving more than four cycles of adjuvant gemcitabine and harboring Human equilibrative nucleoside transporter 1 (hENT-1, the major transporter involved in gemcitabine intracellular uptake) on tumor cell membrane had a longer disease-free survival compared with patients negative or positive for hENT-1 only in the cytoplasm of tumor cells. Overall these results may lay the basis for further prospective randomized trials based on a larger population of patients and may prove useful for tailoring appropriate gemcitabine-based chemotherapy for CC patients in the adjuvant setting.
Cholangiocarcinoma (CC) encompasses a heterogeneous group of malignancies affecting the biliary epithelium. Anatomically, the tumor is classified as intrahepatic (ICC), developing along the biliary tree within the liver, or extrahepatic (ECC), developing along the biliary tree outside the liver. ECC also includes hepatic hilar (or Klatskin) tumors .
Currently, the prognosis for both ICC and ECC is poor, with a 5-year survival rate of less than 15% [1, 2]. Radical surgery offers the only chance of long-term survival, but it is rarely accomplished because of frequent distant metastases or extensive local tumor involvement at diagnosis. Moreover, even after surgery, the prognosis remains unsatisfactory (5-year survival rate ≤40% in the best surgical series), as the risk of early recurrence is high [3–5]. An effective adjuvant therapy is therefore needed to improve the cure rate for CC patients.
To date, the survival benefit of adjuvant therapy in CC has not been definitively proven in randomized controlled trials. Despite this lack of evidence, adjuvant strategies are recommended in the National Comprehensive Cancer Network guidelines . This recommendation is based on a pooled analysis of 20 retrospective studies on adjuvant treatment in biliary tract cancer patients that reported a clinical benefit in the subgroup with lymph node-positive or margin-positive disease .
In Western countries, gemcitabine-based chemotherapy is increasingly used as an adjuvant modality, also in light of the encouraging results obtained in advanced disease, where this regimen represents the standard of care . Moreover, a recent meta-analysis of 12 studies involving 1,361 patients with biliary tract cancer who were undergoing different adjuvant treatments found gemcitabine chemotherapy to be the optimum adjuvant treatment, able to prolong patient survival with a balanced benefit-toxicity ratio .
Since 2002, our referral center for the treatment of hepatobiliary tumors has offered adjuvant gemcitabine to patients with resected CC who are fit for therapy. Gemcitabine is a 2′,2′-difluoro-2-deoxycytidine analog whose intracellular uptake occurs via specialized membrane nucleoside transporters, namely human equilibrative nucleoside transporter 1 (hENT-1) [10, 11]. Recently, some studies reported on the predictive role of hENT-1 intratumoral expression in CC patients treated with gemcitabine. In particular, low hENT1 expression was found to be associated with a poor response to the drug in different settings [12–16]. However, because efficient intracellular uptake of nucleosides and their analogs requires specific hENT-1 localization on the cell membrane [17, 18], we hypothesized that an aberrant localization of this transporter in tumor cells could also result in a decreased or lack of response to gemcitabine in patients treated with this drug.
Therefore, the aim of the present study was to evaluate the putative predictive role of hENT-1 localization in the tumor cells of CC patients who had adjuvant gemcitabine after surgical resection.
All consecutive, unselected CC patients who received adjuvant gemcitabine after radical surgery at S. Orsola-Malpighi University Hospital of Bologna, Italy, from January 2002 to December 2011 were assessed for eligibility.
The type of surgery varied according to tumor location. Patients with distal CC underwent pancreatoduodenectomy, patients with ICC underwent anatomic liver resection, and patients with hilar ECC underwent major hepatectomy including caudate lobe and bile duct resection. In most cases, resection of the fourth and fifth segments was sufficient, whereas major hepatectomies were required in patients with more extensive tumors deeply invading the liver parenchyma (or right branch of portal vein or hepatic artery). Whenever jaundice was present or the cystic duct stump was positive for tumor invasion, resection of the bile duct and hepatojejunostomy were added to the hepatectomy. Regional lymphadenectomy was usually performed at the end of procedures.
The patients were considered suitable for gemcitabine adjuvant chemotherapy if ≥18 years old with Eastern Cooperative Oncology Group Performance Status Scale (ECOG PS) ≤2. Additional criteria included adequate white blood cell count, neutrophil count, platelet count, and serum creatinine concentration and normal liver function values. Patients with both R0 and R1 resection were considered. Patients older than 77 years, low ECOG PS (>2), impaired left ventricular ejection fraction, or abnormal respiratory patterns were unfit for adjuvant chemotherapy after surgery. Patients with bone marrow dysfunction (white blood cell count <3,500/μL or platelet count <100,000/μL), renal dysfunction (creatinine concentration ≥1.5× normal), and hepatic dysfunction (bilirubin and aspartate aminotransferase levels more than twice the upper normal value) were also considered unfit for adjuvant treatment. In the absence of any conclusive data on the efficacy of adjuvant chemotherapy after resection in CC, the decision on further chemotherapy after surgery was left to the patient. All patients were informed of possible risks and benefits of adjuvant chemotherapy and gave written informed consent in accordance with institutional guidelines, including the Declaration of Helsinki.
Data on clinical variables, including sex, age, and tumor/node/metastasis classification, were gathered from patients’ records.
The adjuvant chemotherapy schedule consisted of gemcitabine monotherapy (30-min intravenous infusion at 1,000 mg/m2 on days 1, 8, and 15 of a 28-day cycle for 6 cycles). The overall duration of chemotherapy (6 cycles for 6 months) has been empirically chosen borrowing the experiences in other malignancies (breast, colon, lung, and pancreas) treated with cytotoxic drugs. All patients started the first infusion between days 30 and 60 after surgery.
The follow-up program consisted of history and physical examination; laboratory tests (complete blood count, serum chemistry, carcinoembryonic antigen, and carbohydrate antigen 19-9); and chest x-ray plus abdominal ultrasonography or contrast-enhanced computed tomography scan of chest, abdomen, and pelvis every 4 months for the first 2 years and every 6 months thereafter for at least 5 years. Supplementary imaging techniques were performed whenever a local or distant recurrence was suspected.
Formalin-fixed, paraffin-embedded tumor tissue from resected specimens was collected from eligible patients. Immunohistochemistry (IHC) was carried out using Novolink Polymer Detection System (Leica Microsystems, Wetzlar, Germany, http://www.leica-microsystems.com/home) according to the manufacturer’s instructions. Briefly, 3-µm-thick sections were cut from formalin-fixed, paraffin-embedded blocks, deparaffinized, rehydrated, and subjected to antigen retrieval by heating for 30 min at 99°C in citrate buffer, pH 6.0. Endogenous peroxidase activity and nonspecific binding sites were blocked by 5-min incubations in peroxidase block and protein block, respectively. Tumor sections were then incubated overnight at 4°C with hENT-1 rabbit polyclonal antibody (SLC29A1; GeneTex, San Antonio, TX, http://www.genetex.com) (dilution 1:100). At the end of the incubation, immune complexes were incubated in postprimary antibody for 30 min at room temperature and then in Novolink polymer tertiary antibody for a further 30 min at room temperature. Finally, sections were developed in 3,3′-diaminobenzidine and counterstained with hematoxylin.
Negative controls were made by omitting the primary antibody. According to instructions of the hENT-1 antibody manufacturer, cortical renal parenchyma was chosen as external control. In every IHC run, hENT-1 positivity was observed in the glomerular capillaries and arteriolar walls (supplemental online Fig. 1A). Internal controls for IHC included vascular endothelial cells and lymphocytes (where present) in all cases analyzed (supplemental online Fig. 1B, 1C). IHC analysis was performed by one dedicated pathologist who was blinded to patients’ clinical characteristics and outcome; hENT-1 localization in tumor cells was recorded in all samples.
Summary statistics of continuous variables were expressed as mean (SD) or median (interquartile range [IQR]) and compared with Student’s t test or Mann-Whitney U test. Categorical variables were reported as numbers (percentages) and compared with Pearson’s χ2 test or Fisher’s exact test, according to Cochran’s rule.
The main outcome measure was disease-free survival (DFS), defined as the time from the date of CC surgery to the date of relapse, death from any cause, or last follow-up. Patients’ survival curves were plotted using the Kaplan-Meier method. The association between hENT-1 localization in tumor cells and DFS was characterized by hazard ratios (HRs) and associated 95% confidence intervals (CIs). The limited number of subjects (n = 71) did not allow the direct inclusion of all covariates in multivariate Cox proportional hazards regression models. To deal with the small number of events per parameter, we performed two separate sets of analyses. First, we fitted Cox proportional hazards regression models stratified on quintiles of propensity score, i.e., the probability of a patient expressing membrane hENT-1 given a set of possible confounders . To estimate the propensity score, we fitted a probit regression model including age, gender, anatomical site, tumor size and extent (T), regional lymph nodes (N), distant metastases (M), histological grade, microscopic resection margins, and days elapsed between surgery and start of adjuvant chemotherapy. For the continuous predictors, possible nonlinearity with DFS was tested by applying fractional polynomials . Second, we conducted a second set of regression models including covariates selected based on the change-in-estimates method, using a threshold for inclusion of a 10% change in the HR of interest . After preliminary analysis, regional lymph node involvement and days elapsed between surgery and chemotherapy were retained in the multivariable models as actual confounders of the association between hENT-1 expression and DFS.
To test the a priori hypothesis that the effect of hENT-1 localization could be reduced by short, poorly effective chemotherapy, we fitted Cox proportional hazards regression models including interaction terms between hENT-1 localization on tumor cell membrane and number of cycles of adjuvant gemcitabine chemotherapy.
Statistical analysis was carried out using Stata 12.1 SE (Stata Corp., College Station, TX, http://www.stata.com/company). All tests were two-sided, and a p value < .05 was considered statistically significant.
Of the consecutive, unselected CC patients who underwent adjuvant gemcitabine after radical surgery at our center from January 2002 to December 2011, 21 were excluded because of lack of follow-up. The remaining 71 patients were eligible for the study (see Table 1 for baseline characteristics).
Among the 71 cases, 44 (62.0%) were ICC and 27 (38.0%) ECC (including 19 Klatskin tumors and 8 cancers of the extrahepatic bile ducts). Tumors were graded according to the World Health Organization : grade 1 in 6 (8.4%) cases, grade 2 in 33 (46.5%) cases, and grade 3 in 32 (45.1%) cases. Forty-five patients (63%) received five or six cycles of adjuvant gemcitabine, 20 (28%) three or four cycles, and 6 (9%) one or two cycles. No treatment-related deaths were observed in our case series.
At IHC analysis for hENT-1 localization in tumor cells, 23 (32.4%) cases were completely negative for the transporter, 22 (31.0%) showed only cytoplasm positivity, and 26 (36.6%) had concomitant membrane/cytoplasm immunoreactivity (Fig. 1). In all cases analyzed, hENT-1 staining (when present) was homogeneously distributed within the tumor tissue.
Data cutoff for DFS analysis was December 31, 2012, when median duration of follow-up was 18.1 months (IQR 9.1–36.2): 16.4 months (8.7–28.9) in the negative hENT-1 group and 23.4 months (9.2–44.7) in the positive hENT-1 groups. Forty-nine relapses were observed in 1,678 person-years. The median DFS in the entire study population was 19.9 months (IQR 9.7–53.5).
A preliminary DFS analysis among subgroups (hENT-1-negative, hENT-1-positive only for cytoplasm, and hENT-1-positive for both membrane and cytoplasm) showed an overlap between the DFS curves of hENT-1-negative patients and those positive for hENT-1 in only the cytoplasm of tumor cells (Fig. 2). Therefore, in subsequent analyses, these two subpopulations were grouped in the same subpopulation of CC patients, hereafter referred to as “membrane hENT-1 negative.” Accordingly, the hENT-1-positive membrane and cytoplasm subpopulation is hereafter referred to as “membrane hENT-1 positive.”
Kaplan-Meier survival analysis revealed that membrane hENT-1-positive patients had a longer DFS (median 39 months, IQR 10 to not reached) than membrane hENT-1-negative patients (median 16 months, IQR 9–34) (Fig. 3). This pattern of estimates was similar to the analysis stratified by cancer site, despite the absence of statistical significance because of the small number of patients analyzed (supplemental online Fig. 2A, 2B). However, ICC showed a shorter DFS (median 16 months [IQR 8–34] in membrane hENT-1-negative versus median 24 months [IQR 10 to not reached] in membrane hENT-1-positive patients) than ECC (median 23 months [IQR 11–38] in membrane hENT-1-negative versus median 41 months [IQR 20 to not reached] in membrane hENT-1-positive patients).
Estimates for the association between membrane hENT-1 localization and DFS are shown in Table 2. Unadjusted estimates were close to adjusted estimates. Also, estimates adjusted via propensity score were almost equal to those obtained from multivariable models selected with the change-in-estimates method. Membrane-positive hENT-1 was associated with a longer DFS (adjusted HR via propensity score 0.49, 95% CI 0.24–0.99) compared with membrane-negative hENT-1. Notably, the association between membrane hENT-1 and DFS was found to depend on the number of cycles of gemcitabine chemotherapy (one to two cycles: HR 0.96, 95% CI 0.34–2.68; three to four cycles: HR 0.99, 95% CI 0.34–2.90; and five to six cycles: HR 0.27, 95% CI 0.10–0.77).
The prognosis of CC remains poor, and an effective adjuvant chemotherapy is urgently needed together with surgery to improve the outcome of CC patients. Although the efficacy of adjuvant therapy in CC has yet to be proven in prospective randomized trials (because of the limited number of patients eligible for surgery and the high morbidity and mortality rates associated with this disease), a recent pooled analysis of 20 retrospective studies suggests that gemcitabine-based chemotherapy may represent an interesting therapeutic option to improve the survival rate in CC patients .
To date, predictive biomarkers of response to adjuvant gemcitabine have been hard to pinpoint in CC; therefore a considerable number of patients undergo this chemotherapeutic regimen without the prospect of clinical benefit. Recently, some studies reported the predictive role of hENT-1 intratumoral expression in CC patients treated with gemcitabine [12–16]. To our knowledge, the present study is the first to explore the putative predictive role of hENT-1 localization in tumor cells of cancer patients who had adjuvant gemcitabine chemotherapy. A significant association between membrane hENT-1 localization and DFS was observed in our CC case series. In particular, membrane hENT-1-positive patients were found to have a longer DFS than membrane hENT-1-negative ones. This finding is clinically plausible in light of the biological role of hENT-1 within cells. hENT-1 is an integral membrane protein that mediates the transport of nucleosides and their analogs across the cell membrane. As such, efficient intracellular uptake of these molecules requires specific localization of the transporter on the cell membrane [17, 18]. Accordingly, disruption of hENT-1 membrane localization has been shown to result in impaired gemcitabine uptake and induction of chemoresistance in in vitro studies .
It can be hypothesized that a similar condition may also occur in vivo. Aberrant localization of hENT-1 in tumor cells of cancer patients receiving adjuvant gemcitabine may therefore lead to a decreased response or lack of response to chemotherapy because of inefficient intracellular uptake of the drug. This hypothesis is supported by the observation that, in the subgroup of membrane hENT-1-negative patients, DFS curves overlapped between patients completely negative for the transporter and those positive for hENT-1 in the cytoplasm only.
These findings go beyond the current literature data, as they suggest for the first time that localization of hENT-1 on the tumor cell membrane is required to achieve better survival in patients undergoing gemcitabine chemotherapy, whereas localization of this transporter in the cytoplasm alone is not a sufficient condition to obtain a clinical response to this treatment. It should be pointed out, however, that in our CC case series, membrane hENT-1 positivity invariably occurred together with cytoplasmic positivity. This observation is consistent with the trafficking of hENT-1 within cells, as the transporter is translocated to the cell membrane in association with a variety of cytoplasmic vesicles . It is therefore possible that in the previous studies reporting the predictive role of hENT-1 expression in tumor cells by IHC, many of the CC cases with high hENT-1 cytoplasmic staining also had concomitant hENT-1 membrane localization but, because of important variables in the analysis (time of tissue fixation, type of fixation, long-term stability of the epitope, and specificity of the antibody used), this detail could not be detected in the samples.
Another open question in CC treatment is the optimal duration of adjuvant chemotherapy. A significant association between membrane hENT-1 localization and DFS, depending on the number of gemcitabine cycles, was observed in our case series. Indeed, whereas the response to gemcitabine was negligible with up to four cycles, five or six cycles of chemotherapy were associated with a threefold decrease in relapse risk in membrane hENT-1-positive patients compared with membrane hENT-1-negative cases. Although the small sample size and retrospective design of this study do not allow any conclusions on the optimal number of adjuvant chemotherapy cycles in CC, this finding is in line with the results recently reported in the European Study Group for Pancreatic Cancer 3 (ESPAC-3) study  and may deserve further clinical investigation in a prospective trial that, in light of the trend we observed, could also include an additional subgroup of CC patients receiving adjuvant gemcitabine for more than 6 months.
The development of a prospective trial including a control group of not-treated patients and a group of patients receiving a different number of adjuvant gemcitabine cycles (both stratified according to membrane hENT-1 localization) may also be useful to clarify whether membrane hENT-1 may represent a prognostic, rather than predictive, factor in CC. Indeed, because of the design of the present study, a prognostic role of membrane hENT-1 cannot be excluded in these patients. However, this hypothesis seems unlikely in light of the functional role of hENT-1 in cells, i.e., the transport across the cell membrane of nucleosides required for DNA synthesis. In these terms, in fact, hENT-1 would represent one of the key factors driving cell proliferation, being therefore more associated with a worse, rather than a better, prognosis, as already reported in other malignancies [26, 27].
Overall, our findings suggest that localization of hENT-1 on tumor cell membrane may predict response to adjuvant gemcitabine in resected CC patients receiving more than four cycles of chemotherapy. However, because of the retrospective design of this study, prospective randomized clinical trials based on a larger population of patients are required to confirm these preliminary results, so that appropriate gemcitabine-based chemotherapy may be tailored for CC patients in the adjuvant setting.
See http://www.TheOncologist.com for supplemental material available online.
This work was supported by a grant from Fundamental Oriented Research (RFO 2013) and Fondazione Cassa di Risparmio di Bologna (CARISBO) to G. Brandi. The Gruppo Italiano Colangiocarcinoma members are Guiseppe Aprile, Stefano Cereda, Lorenzo Fornaro, Francesco Leone, Sara Lonardi, Daniele Santini, Nicola Silvestris, and Enrico Vasile.
For Further Reading: Lipika Goyal, Aparna Govindan, Rahul A. Sheth et al. Prognosis and Clinicopathologic Features of Patients With Advanced Stage Isocitrate Dehydrogenase (IDH) Mutant and IDH Wild-Type Intrahepatic Cholangiocarcinoma. The Oncologist 2015; 20:1019–1027.
Implications for Practice: Previous studies assessing the prognostic impact of the isocitrate dehydrogenase (IDH) gene mutation in intrahepatic cholangiocarcinoma (ICC) mainly focused on patients with early-stage disease who have undergone resection. These studies offer conflicting results. The target population for clinical trials of IDH inhibitors is patients with unresectable or metastatic disease, and the current study is the first to focus on the prognosis and clinical phenotype of this population and reports on the largest cohort of patients with advanced IDH mutant ICC to date. The finding that the IDH mutation lacks prognostic significance in advanced ICC is preliminary and needs to be confirmed prospectively in a larger study.
Conception/Design: Giovanni Brandi, Simona Tavolari
Provision of study material or patients: Giovanni Brandi, Francesco Vasuri, Alessio Degiovanni, Mariacristina Di Marco, Antonio D. Pinna, Matteo Cescon, Alessandro Cucchetti, Giorgio Ercolani, Antonietta D’Errico-Grigioni, Maria A. Pantaleo, Guido Biasco
Collection and/or assembly of data: Marzia Deserti, Andrea Palloni, Stefania de Lorenzo, Giorgio Frega, Maria A. Barbera, Ingrid Garajova, Simona Tavolari
Data analysis and interpretation: Giovanni Brandi, Francesco Vasuri, Andrea Farioli, Simona Tavolari
Manuscript writing: Giovanni Brandi, Francesco Vasuri, Andrea Farioli, Simona Tavolari
Final approval of manuscript: Giovanni Brandi, Simona Tavolari
Antonio D. Pinna: Novartis (RF). 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