Excision repair cross-complementing gene-1, ribonucleotide reductase subunit M1, ribonucleotide reductase subunit M2, and human equilibrative nucleoside transporter-1 expression and prognostic value in biliary tract malignancy
Sarah B. Fisher MD,
Department of Surgery, Division of Surgical Oncology, Winship Cancer Institute, Emory University, Atlanta, Georgia
Tumor expression of excision cross-complementing gene-1 (ERCC1), human equilibrative nucleoside transporter 1 (hENT1), ribonucleotide reductase subunit M1 (RRM1), and ribonucleotide reductase subunit M2 (RRM2), is associated with the efficacy of platinum and gemcitabine chemotherapy. The authors of this report recently demonstrated that high ERCC1 and RRM2 expression levels are independent negative prognostic markers for survival in early stage pancreas cancer. The differential expression and prognostic value of these biomarkers in biliary tract malignancy (BTM) is unknown.
In total, 63 patients who had tissue available for analysis were selected from a prospective database of all patients (n = 104) who underwent resection of BTM (intrahepatic, hilar, or distal cholangiocarcinoma; gallbladder carcinoma) between January 2000 and December 2008. Immunohistochemistry for ERCC1, hENT1, RRM1, and RRM2 expression was performed. Staining was scored by a single pathologist who was blinded to patient outcomes.
The median patient age was 67 years. The median overall survival (OS) was 16.2 months, and the median follow-up was 32.7 months. Only 3 BTMs (4.8%) had high ERCC1 expression, and 92.1% and 81% of BTMs exhibited high hENT1 and RRM1 expression, respectively. RRM2 expression varied, and 32% of tumors demonstrated high RRM2 expression. ERCC1 and RRM1 were not associated with OS. High RRM2 expression was associated with a trend toward improved OS (30.8 months vs 16.2 months; P = .06), and high hENT1 expression was associated with improved OS (17.7 months vs 9.5 months; P = .04).
Biliary tract malignancies (BTMs), which include cholangiocarcinoma and gallbladder carcinoma, are uncommon and often lethal cancers. Approximately 9000 new cases are diagnosed in the United States each year.1 Surgical resection offers the only opportunity for cure for all BTMs, yet recurrence is common. Although select patients with early stage (T1a) gallbladder carcinoma have a favorable prognosis, the long-term survival of most patients with BTM remains low after they undergo resection with curative intent.2-5 Evidence supporting the use of adjuvant therapy after resection is lacking.2, 6-10 In the setting of advanced disease, Valle and colleagues11 have reported promising results with a doublet chemotherapy regimen consisting of gemcitabine (Gem) and cisplatin (Cis), which extended survival by almost 4 months compared with Gem monotherapy. Whether this advantage will translate to the adjuvant setting remains to be determined, but patient selection will be important to maximize therapeutic benefit and minimize toxicity.
Molecular biomarkers are used successfully to stratify patient populations according to risk and to predict response to therapy in other cancer types (for example, human epidermal growth factor receptor 2 [HER-2/neu] amplification in breast cancer12). Although several biomarkers have been proposed in BTM, their clinical utility is limited by lack of specificity,5; and, currently, no clinically useful biomarkers are known. Preclinical studies focused on the metabolism and efficacy of Gem and Cis chemotherapy have identified several candidate biomarkers, including human equilibrative nucleoside transporter-1 (hENT1), ribonucleotide reductase subunits M1 and M2 (RRM1 and RRM2), and excision repair cross-complementing gene-1 (ERCC1).
hENT1 is the predominant cellular transporter responsible for Gem entry into the cell.13 In vitro studies have directly correlated hENT1 expression with Gem sensitivity in hepatobiliary and other human cancer cell lines.14-17 After transport into the cell, Gem inhibits ribonucleotide reductase (RR), which is composed of 2 subunits, RRM1 and RRM2. During normal cellular replication, RRM1 and RRM2 dimerize and catalyze the production of deoxynucleotide triphosphates, which are necessary for DNA synthesis.18 In vitro studies have associated high expression levels of RRM1 and RRM2 with Gem resistance.19-21 ERCC1 is a key player in the nucleotide excision repair (NER) pathway, which normally functions to protect the body from carcinogens by excising bulky intrastrand adducts from DNA during cellular replication.22 Platinum compounds like Cis halt cellular replication by forming similar intrastrand adducts, and high ERCC1 expression has been associated with resistance to platinum therapy.22
hENT1, RRM1, RRM2, and ERCC1 have each been proposed as useful in predicting response to therapy in a variety of other cancer types.23-25 In addition to a predictive value for response to chemotherapy, a biomarker can also possess prognostic value for survival outcomes; the terms predictive and prognostic are not synonymous or mutually exclusive. We recently demonstrated that high tumor expression levels of either ERCC1 or RRM2 were independent negative prognostic markers for survival in patients with resected pancreas cancer.26 The expression profiles and prognostic value of ERCC1, hENT1, RRM1, and RRM2 in BTMs is unknown. The objectives of the current study were to characterize the tumor expression profiles of ERCC1, hENT1, RRM1, and RRM2 in BTM and to evaluate their utility as prognostic biomarkers for survival after resection of early stage disease.
MATERIALS AND METHODS
A prospectively maintained hepatobiliary resection registry was queried for all patients who underwent resection for BTM between January 2000 and December 2008 (n = 104). Patients who did not have tissue available for analysis were excluded (n = 41). A chart review was performed for 63 patients, and patient comorbidities, operative details, pathologic characteristics, and survival outcomes were collected. Survival outcomes were measured from the date of surgery. Perioperative mortalities, defined as death within 30 days after operation, were excluded from survival analyses. This study was approved by Emory University's Institutional Review Board, and compliance with the Health Insurance Portability and Accountability Act of 1996 was maintained.
Formalin-fixed, paraffin-embedded samples with representative areas of tumor tissue were agreed upon by 2 experienced hepatobiliary pathologists (A.B.F. and M.G.L.) who were blinded to outcomes. Tissue was reacted with anti-ERCC1 monoclonal antibody (clone 8F1; Neomarkers, Fremont, Calif), anti-RRM1 polyclonal antibody (clone AD203; ProteinTech, Chicago, Ill), anti-RRM2 monoclonal antibody (clone 1E1; Sigma-Aldrich, St. Louis, Mo), and anti-hENT1 polyclonal antibody (polyclonal SLC29A1a; Sigma-Aldrich) to determine levels of tumor expression of ERCC1, RRM1, RRM2, and hENT1, respectively. A separate pathologist (K.E.F.), who also was blinded to outcomes, graded the immunohistochemical stains based on previously established parameters for staining intensity and the percentage of stained cells.27
Fine granular nuclear staining was regarded as positive for ERCC1. Fine granular cytoplasmic and nuclear staining for RRM1 and RRM2 was regarded as positive. Membranous, fine granular cytoplasmic and nuclear staining for hENT1 was regarded as positive. We used our previously published algorithm27 to calculate an overall score for each sample. The overall score was divided into a low expression group and a high expression group.
Data were analyzed using Statistical Package for the Social Sciences 19.0 for Windows (SPSS, Inc. Chicago, Ill). Continuous clinicopathologic variables were compared using the Student t test. Categoric clinicopathologic variables were compared with Pearson chi-square and Fisher exact tests, as appropriate. Two tailed P values <.05 were considered statistically significant. The primary outcome was overall survival (OS). Kaplan-Meier log-rank survival analysis was used to assess the relation of each biomarker expression profile and survival. Univariate Cox regression was performed for OS using known adverse clinicopathologic factors and biomarker expression. A multivariate Cox regression model was conducted by including any factors that were significant at the level of P < .2 in univariate analysis.
The median patient age was 67 years (range, 32-82 years), 30 patients (47.6%) were men, and the majority of patients (68.3%; n = 43) were white. At the time of last follow-up, 47 patients (74.6%) had died. The median OS was 16.2 months, and the median follow-up was 32.7 months for survivors. Six patients died within 30 days of operation, and those patients were excluded from survival analyses. No patients received neoadjuvant therapy. Records regarding adjuvant therapy were available for 63% of patients (n = 39); and, of those 39 patients, 41% (n = 16) received chemotherapy and 43% (n = 17) underwent radiation. Regarding chemotherapy, 50% of patients received a Gem-based regimen (Gem, 8 patients; capecitabine, 5 patients; 5-fluorouracil, 4 patients; Cis, 3 patients). Intrahepatic, hilar, and distal cholangiocarcinoma were diagnosed in 15 patients (23.8%), 25 patients (39.7%), and 14 patients (22.2%), respectively. Nine patients (14.3%) had gallbladder carcinoma. For all patients, the median tumor size was 3.6 cm (range, 0.3-15 cm), 12 patients (19%) had microscopically positive resection margins, and 23 patients (36.5%) had positive lymph nodes. Most patients had a solitary lesion (95.2%; n = 60), and the remaining 3 patients had 2 lesions (4.8%). Clinicopathologic characteristics of the study cohort are summarized in Table 1.
Table 1. Clinicopathologic and Perioperative Characteristics
No. of Patients (%)
Major complications were defined as complications with a Clavien-Dindo score from III to V.
There were no patients with T2 gallbladder carcinoma.
ERCC1 expression in BTM was predominantly low, and only 3 patients (4.8%) had tumors with high ERCC1 expression. Conversely, hENT1 and RRM1 expression levels were predominantly high: Fifty-eight patients (92.1%) and 51 patients (81%) had tumors with high expression levels, respectively. RRM2 expression had greater heterogeneity, and 21 patients (33.3%) had tumors with high RRM2 expression. Representative immunohistochemical stains with low and high expression of each biomarker are provided in Figure 1.
When tumors were stratified according to individual disease type, low ERCC1 expression was observed in 100%, 92% (n = 23), 94% (n = 14), and 89% (n = 8) of intrahepatic cholangiocarcinomas, hilar cholangiocarcinomas, distal cholangiocarcinomas, and gallbladder carcinomas, respectively. High RRM1 expression was observed in 87% (n = 13), 76% (n = 19), 79% (n = 11), and 89% (n = 8) of intrahepatic cholangiocarcinomas, hilar cholangiocarcinomas, distal cholangiocarcinomas, and gallbladder carcinomas, respectively. High RRM2 expression was observed in 60% (n = 9) of intrahepatic cholangiocarcinomas; whereas low expression of RRM2 was observed in 72% (n = 18), 86% (n = 12), and 67% (n = 6) of hilar cholangiocarcinomas, distal cholangiocarcinomas, and gallbladder carcinomas, respectively. High hENT1 expression was observed in 87% (n = 13), 96% (n = 24), 100% (n = 14), and 78% (n = 7) of intrahepatic cholangiocarcinomas, hilar cholangiocarcinomas, distal cholangiocarcinomas, and gallbladder carcinomas, respectively. There were no significant differences in biomarker expression when tumors were stratified according to individual disease type, although RRM2 expression trended toward high expression in intrahepatic cholangiocarcinomas and toward low expression in the other 3 types (P = .06) (Table 2).
Table 2. Differential Biomarker Expression Stratified by Pathologic Type
Patients also were stratified into 2 groups according to anatomic disease site: intrahepatic (intrahepatic cholangiocarcinoma) or extrahepatic (hilar and distal cholangiocarcinoma and gallbladder carcinoma): The distribution of ERCC1, RRM1, and hENT1 expression did not differ significantly according to anatomic location (P = .76, P = .76, and P = .75, respectively) (Table 3). RRM2 expression was more likely to be high in intrahepatic tumors and low in extrahepatic tumors (P = .02) (Table 3).
Table 3. Differential Biomarker Expression Stratified by Anatomic Location
ERCC1 and RRM1 expression levels were not associated with OS (Fig. 2a,b). High RRM2 expression demonstrated a trend toward improved OS (30.8 months vs 16.2 months; P = .06) (Fig. 2c). High hENT1 expression was associated with improved OS (17.7 months vs 9.5 months; P = .04 (Fig. 2d). In univariate Cox regression analysis, high RRM2 expression and high hENT1 expression were associated with improved OS (hazard ratio, 0.51; 95% confidence interval, 0.24-1.04 [P = .07] and hazard ratio, 0.30; 95% confidence interval, 0.99-11.55 [P = .05], respectively). Perineural invasion, tumor size >3.6 cm (dichotomized around the median), and lymph node disease also were identified as significant in univariate analysis (Table 4). Patient age, intraoperative blood loss, margin status, poor tumor grade, lymphovascular invasion, receipt of adjuvant therapy, high ERCC1 expression, and high RRM1 expression were not related significantly to OS in univariate analysis (Table 4). No factors maintained significance in multivariate Cox regression analysis (Table 4). Subset analyses with patients stratified according to pathologic type or anatomic location (intrahepatic vs extrahepatic) and biomarker expression were not possible because of small sample sizes. Similarly, a subset analysis of the patients who received adjuvant therapy, either as a group or stratified by specific regimen, also was limited by small sample size.
Table 4. Univariate and Multivariate Cox Regression Model for Overall Survival
BTM represents a rare and clinically challenging entity with limited prospective evidence available to guide management.6 Although BTMs are comprised of several distinct disease sites, they often are grouped together for purposes of study given the relative rarity of each individual malignancy.11 In the current study, we report on the expression profiles and prognostic value of 4 candidate biomarkers in cholangiocarcinoma and gallbladder carcinoma.
Biomarkers with prognostic value are particularly important in BTM, in which many patients suffer recurrence and OS is disappointingly low, even for patients with early stage disease. Given the paucity of data to support adjuvant therapy after resection, a prognostic biomarker may help to guide further treatment decisions, such as selecting patients for adjuvant therapy. Furthermore, patients who are enrolled in clinical trials evaluating adjuvant therapy may be stratified according to biomarker expression. Currently, the majority of the literature regarding ERCC1, RRM1, RRM2, and hENT1 focuses on their predictive role for response to Gem and Cis chemotherapy, and little is known regarding their prognostic value. We recently demonstrated that high tumor levels of both ERCC1 expression and RRM2 expression are independent, negative prognostic factors for early recurrence and shortened survival in resected pancreas cancer.26 The value of ERCC1, RRM1, RRM2, and hENT1 expression in BTM is not established.
In contrast to pancreatic cancer, in which 16% of tumors demonstrated high expression of ERCC1,27 almost all BTMs demonstrated low expression. In pancreas cancer, high ERCC1 expression was associated with significantly reduced survival.26 In BTM, no survival distinction was observed based on ERCC1 expression, although the uniformity of low expression may have precluded any separation. Recently, Roa and colleagues28 examined a large cohort of patients with resected gallbladder carcinoma as well as a control group of nonmalignant cholecystectomy specimens. Those authors demonstrated that nonmalignant tissue and early stage gallbladder carcinomas had higher expression of ERCC1 than advanced stage tissues, and they hypothesized that the loss of ERCC1 expression may be related to disease progression.28 A similar pattern was not identified in the limited number of gallbladder carcinoma specimens that were included in the current study (most of which were T3 tumors). It is possible that this finding of low ERCC1 expression in most BTMs partly explains the findings of the second Advanced Biliary Cancer (ABC-02) trial, in which adding a platinum agent to the standard Gem regimen for patients with advanced BTM produced a survival benefit.11 ERCC1 expression in BTM remains a potential subject of investigation, particularly in studies that assess the role of platinum chemotherapy in either the advanced or adjuvant setting.
Most research regarding RRM1 has been conducted in nonsmall cell lung cancer. The effect of high RRM1 expression seems to be stage-specific, because patients who undergo resection for early stage disease with high RRM1 expressing tumors fare better, whereas patients who have advanced disease and high RRM1 tumor expression have a worse prognosis.25 Studies examining RRM1 in patients with BTM have focused on the predictive value of RRM1 expression for response to Gem therapy in the advanced setting. In a small study (n = 12) of patients with advanced BTM who received Gem, trend toward better response rate was observed for patients who had tumors with low RRM1 expression.29 Similarly, Nakamura and colleagues30 demonstrated that, in a group of 10 patients with advanced BTM, low RRM1 expression was associated with Gem sensitivity and improved OS. With the exception of the current study, only 1 study has addressed RRM1 expression in early stage (surgically resectable) BTM. Wakai and colleagues31 observed a greater response to neoadjuvant Gem in 1 patient who had a tumor with low RRM1 expression compared with another patient who had a tumor with high RRM1 expression. To our knowledge, the current study is the first to assess the prognostic value of RRM1 in resectable BTM. Most BTMs demonstrated high RRM1 expression, and there was no discernible separation in survival between patients with low-expressing tumors and high-expressing tumors.
Compared with RRM1, RRM2 is less studied. High expression of RRM2 has been associated with shorter OS in ovarian cancer regardless of therapy type,32 and we recently demonstrated that high RRM2 expression was associated with a poor prognosis in patients with resected pancreatic cancer.26 Only 1 other study has examined RRM2 in BTMs; in that study, higher levels of RRM2 expression in vitro were associated with Gem resistance.30 The findings of the current study suggest that the prognostic value of RRM2 may be distinct from its predictive value for response to therapy, because high RRM2 expression in BTM appears to be associated with a favorable prognosis. The discordance between prognostic and predictive value is not without precedent, as described above for RRM1 expression in nonsmall cell lung cancer. Further evaluation regarding RRM2 expression in BTM is necessary.
The predictive value of hENT1 expression as the primary intracellular transporter of Gem has been assessed by immunohistochemistry in a variety of malignancies, including breast, pancreatic, and lung cancers.23, 25, 33 In BTM, Mori and colleagues16 demonstrated that expression levels of hENT1 were correlated with Gem sensitivity in vitro. Studies in advanced BTM have correlated high hENT1 expression with improved survival for patients who receive Gem therapy.34, 35 Most of the literature regarding hENT1 focuses on its predictive role for response to Gem therapy. To our knowledge, our study is the first to assess the prognostic value of hENT1 expression in resectable BTM. The findings here suggest that high hENT1 expression may have prognostic value in addition to its predictive value, particularly because only a few patients were known to have received Gem therapy. The findings are limited by small sample size and need to be validated in larger (perhaps multi-institutional) studies.
The prognostic value of RRM2 and hENT1 expression was not observed in a multivariate Cox regression analysis. In fact, no single clinicopathologic factor was associated with reduced survival when tested in the multivariate model. This finding probably is related to the number of patients included in this study. In addition, other clinicopathologic factors that were not accounted for or captured in our multivariate model may affect survival. Regardless, the observations described above regarding tumor biomarker expression and survival in patients with BTM provide an incentive for further study.
The current study is subject to several inherent limitations. Like most other immunohistochemical studies, the immunohistochemistry technique used in this study was semiquantitative. However, biomarker expression was scored by a single experienced pathologist who was blinded to outcomes in an attempt to minimize bias. The retrospective nature of this study poses limitations on the ability to retrieve data and form conclusions. In addition, at a large, tertiary care referral center, medical records are limited regarding the extent and type of adjuvant therapy administered, because many patients elect to pursue postoperative medical care locally. Furthermore, the heterogeneity of administered chemotherapy regimens in a retrospective study precludes an assessment of each biomarker's predictive value. The findings reported here warrant further study in a prospective trial design.
The ABC-02 trial demonstrated a survival benefit using a combination chemotherapy regimen consisting of Gem and Cis compared with standard Gem monotherapy for patients with advanced stage BTM.11 Further investigations, particularly studies that assess the value of this doublet regimen in the adjuvant setting after resection, should incorporate correlative studies that assess tumor expression of ERCC1, RRM1, RRM2, and hENT1. This information may help to stratify patients based on response and to select patients after resection to receive a potentially more efficacious chemotherapy doublet regimen.
This study was supported by a Winship Cancer Institute Investigator-Initiated Trial Seed Grant.