Fax: (404) 778-4255
An analysis of human equilibrative nucleoside transporter-1, ribonucleoside reductase subunit M1, ribonucleoside reductase subunit M2, and excision repair cross-complementing gene-1 expression in patients with resected pancreas adenocarcinoma
Implications for adjuvant treatment
Article first published online: 8 MAY 2012
Copyright © 2012 American Cancer Society
Volume 119, Issue 2, pages 445–453, 15 January 2013
How to Cite
Fisher, S. B., Patel, S. H., Bagci, P., Kooby, D. A., El-Rayes, B. F., Staley, C. A., Adsay, N. V. and Maithel, S. K. (2013), An analysis of human equilibrative nucleoside transporter-1, ribonucleoside reductase subunit M1, ribonucleoside reductase subunit M2, and excision repair cross-complementing gene-1 expression in patients with resected pancreas adenocarcinoma. Cancer, 119: 445–453. doi: 10.1002/cncr.27619
- Issue published online: 4 JAN 2013
- Article first published online: 8 MAY 2012
- Manuscript Accepted: 26 MAR 2012
- Manuscript Received: 6 MAR 2012
- excision cross-complementing gene-1;
- ribonucleotide reductase subunit 1;
- ribonucleotide reductase subunit 2;
- human equilibrative nucleoside transporter-1;
- pancreas cancer;
Tumor overexpression of excision repair cross-complementing gene-1 (ERCC1) may be associated with decreased survival in patients with pancreas adenocarcinoma (PAC). Human equilibrative nucleoside transporter-1 (hENT1) and ribonucleoside reductase subunits M1 and M2 (RRM1 and RRM2) are integral to cellular transport and DNA synthesis and are implicated as poor prognostic factors in other malignancies. To the authors's knowledge, their role in PAC is not defined.
A prospective database was used to randomly select 95 patients who underwent pancreaticoduodenectomy for PAC between January 2000 and October 2008. Immunohistochemical analysis was performed on tumor samples for hENT1, RRM1 and RRM2, and ERCC1. Main outcomes were recurrence-free survival (RFS) and overall survival (OS).
The median follow-up, RFS, and OS were 49 months, 10.6 months, and 15.5 months, respectively. The median tumor size was 3 cm. Approximately 26% of patients had positive microscopic margins, 61% had lymph node involvement, and 88% and 45% had perineural and lymphovascular invasion, respectively. High tumor expression of hENT1, RRM1, RRM2, and ERCC1 was present in 85%, 40%, 17%, and 16%, respectively, of patients. High hENT1 expression was associated with reduced RFS (9.5 months vs 44.5 months; P = .029), but not with OS. RRM1 expression was not associated with survival. High RRM2 expression was associated with reduced RFS (6.9 months vs 16.0 months; P < .0001) and decreased OS (9.1 months vs 18.4 months; P < .0001). High ERCC1 expression was associated with reduced RFS (6.1 months vs 15 months; P = .04) and decreased OS (8.9 months vs 18.1 months; P = .03). After accounting for known adverse tumor factors, high expression of RRM2 and ERCC1 persisted as negative prognostic factors for RFS and OS. A subset analysis of patients who received adjuvant therapy (n = 74) revealed the same negative effect of high RRM2 and ERCC1 expression on RFS and OS.
High tumor expression of RRM2 and ERCC1 are associated with reduced RFS and OS after resection of pancreas cancer. These biomarkers may help to personalize adjuvant therapy. Cancer 2013. © 2012 American Cancer Society.
Efforts to improve survival in patients with early stage pancreas adenocarcinoma (PAC) have focused on adjuvant therapy. Prospective randomized trials such as Radiation Therapy Oncology Group (RTOG) 9704 and Charité Onkologie Clinical (CONKO)-001 have demonstrated only a modest advantage with gemcitabine-based adjuvant chemotherapy; the ideal drug combination remains to be determined.1-3 The addition of platinum compounds such as cisplatin to gemcitabine-based regimens has improved outcomes somewhat, with survival benefits limited to patients with good performance status.4 Recently, Conroy et al demonstrated a distinct survival advantage for patients with advanced stage pancreatic adenocarcinoma who were treated with FOLFIRINOX, a chemotherapy regimen comprised of oxaliplatin, irinotecan, 5-fluorouracil, and leucovorin, compared with the standard gemcitabine monotherapy regimen.5 Although this study supports the premise that combination therapy using a platinum compound is beneficial for patients with advanced cancer of the pancreas, the value of its application to patients with early stage disease remains unknown. Thus, gemcitabine remains the mainstay of adjuvant therapy.
Preclinical studies that have improved our understanding of the mechanisms of gemcitabine and cisplatin have led to the identification of key molecular biomarkers that may be of prognostic value, as well as potential therapeutic targets. With regard to the combination of gemcitabine and cisplatin, potential biomarkers include human equilibrative nucleoside transporter-1 (hENT1), ribonucleotide reductase subunits 1 and 2 (RRM1 and RRM2), and excision repair cross-complementing gene-1 (ERCC1). hENT1 is integral to the intracellular transport of gemcitabine, whereas RRM1, RRM2, and ERCC1 are involved in DNA synthesis and repair. Depending on their expression, all have been implicated as poor prognostic factors in various malignancies.
Gemcitabine is a prodrug that requires active transport into the cell by hENT1.6 In vitro studies using pancreatic and other cancer cell lines have directly correlated expression of hENT1 with gemcitabine sensitivity.7-10 Once inside the cell, gemcitabine exerts its cytotoxic activity by several mechanisms, including inhibition of ribonucleotide reductase (RR), a dimeric enzyme that supplies deoxynucleotides essential for DNA replication and cell growth.6 RRM1, the regulatory subunit of RR, has been identified as a potential tumor suppressor gene, and high levels of RRM1 have correlated with improved overall survival (OS) in patients with cancers of the bladder and lung as well as those with advanced stage PAC.11, 12 High expression levels of RRM1, however, also are associated with gemcitabine resistance, thus demonstrating a predictive value as well.13 RRM2, the catalytic subunit of RR, is less well studied, but its overexpression has been associated with gemcitabine resistance and increased cellular invasiveness in vitro.14, 15 ERCC1 is a key player in the nucleotide excision repair pathway involved in platinum compound resistance and has been associated with reduced survival in patients with colorectal and gastric cancers.16, 17 We recently demonstrated that high ERCC1 expression was associated with reduced recurrence-free survival (RFS) and OS in individuals with early stage PAC.18
The prognostic value of hENT1, RRM1, and RRM2 expression in early stage PAC (that which is amenable to complete resection) is to our knowledge unknown. In the current study, we attempted to characterize the expression profiles of hENT1, RRM1, and RRM2 in patients with early stage PAC and evaluated these biomarkers for their prognostic value in terms of disease recurrence and OS. Another objective was to confirm that high ERCC1 expression was a negative prognostic marker for patients with resected PAC using updated patient follow-up and by assessing its independent effect in the context of hENT1, RRM1, and RRM2 expression.
MATERIALS AND METHODS
We have previously reported on the prognostic value of increased tumor ERCC1 expression in 95 patients who underwent pancreaticoduodenectomy between January 2000 and October 2008 for early stage PAC.18 These patients were randomly selected from a prospectively maintained pancreas surgery database in which tissue was available for analysis. This same population of 95 patients was used to assess the prognostic value of tumor expression of hENT1, RRM1, and RRM2. ERCC1 expression data were reanalyzed in the current study to include longer patient follow-up as well as to confirm its independent prognostic value for disease recurrence and survival in the context of hENT1, RRM1, and RRM2 expression. Permission from the Institutional Review Board was obtained and Health Insurance Portability and Accountability Act of 1996 (HIPAA) compliance was maintained.
Surveillance imaging studies obtained after surgical resection were used to assess RFS. Survival outcomes were measured from the date of surgery. Pathology reports were reviewed for known adverse tumor characteristics, including tumor size, resection margin status, lymph node (LN) involvement, and the presence of either lymphovascular invasion (LVI) or perineural invasion (PNI).
An experienced pancreatic pathologist (P.B.) identified a representative section of both tumor and normal tissue on formalin-fixed, paraffin-embedded slides. The tissue was reacted with anti-hENT1 polyclonal antibody (Sigma-Aldrich Corporation, St. Louis, Mo), anti-RRM1 polyclonal antibody (Proteintech Group, Chicago, Ill), and anti-RRM2 monoclonal antibody (clone 1E1; Sigma-Aldrich Corporation) to determine levels of tumoral expression of hENT1, RRM1, and RRM2, respectively. Hematoxylin and eosin and immunohistochemistry (IHC) slides were graded by the same pancreatic pathologist, who was blinded to patient outcomes.
Membranous, fine granular cytoplasmic and nuclear staining for hENT1 was regarded as positive. Fine granular cytoplasmic and nuclear staining for RRM1 and RRM2 was regarded as positive. The methodology for quantification of IHC stains for the percentage of cells and intensity of staining is based on a previously described scale of 0 to 3.18 A scoring system18 that incorporates both the percentage of cells stained and the intensity of staining was used to derive a quantitative overall score to allow for comparison. The overall score was divided into a low- or high-expression group if the score was ≤ 2 or > 2, respectively.
Statistical Package for the Social Sciences software (version 19.0 for Windows; SPSS, Inc, Chicago, Ill) was used to analyze the data. The association between the expression of each gene with disease recurrence and survival was assessed using the Kaplan-Meier log-rank survival analysis. Univariate and multivariate Cox regression analyses were used to determine the prognostic value of the expression of each biomarker (hENT1, RRM1, and RRM2) for RFS and shortened OS. Known adverse pathologic factors and ERCC1 expression status also were included in the regression analysis. Factors that were significant to a level of P < .1 on univariate analysis for either RFS or OS were included in the multivariate model. A subset analysis using the same multivariate Cox regression model was also performed only for those patients who were administered adjuvant therapy.
The clinical and pathologic characteristics of the current study population have been reported previously18 and are summarized in Table 1. At the time of last follow-up, 75 (79%) patients had died; the median follow-up for survivors was 49.3 months. The median RFS and OS for all patients were 10.6 months and 15.5 months, respectively.
|Age, years||63 (37-84)|
|Tumor size, cm||3 (0.7-6)|
|Positive microscopic margin||25 (26%)|
|LN-positive disease||58 (61%)|
|Poorly differentiated||32 (34%)|
A total of 74 patients (78%) were administered adjuvant therapy after surgical resection; only 2 patients received preoperative therapy. Gemcitabine was the most commonly administered adjuvant regimen. Increased tumor size, a positive microscopic resection margin, LN involvement, PNI, and LVI were all found to be associated with reduced RFS and OS on univariate analysis (P ≤ .1). Grade was not found to be associated with either RFS or OS (P = .183 and P = .144, respectively).
There was differential expression of hENT1, RRM1, and RRM2 in PAC as assessed by both the percentage of cells stained and the intensity of staining. When our definition of low and high expression for each category was applied, 95 (100%), 77 (81%), and 41 (43%) patients had tumors that demonstrated a high percentage of cells staining for hENT1, RRM1, and RRM2, respectively. When the intensity of expression was scored, 84 (88%), 52 (55%), and 36 (38%) patients displayed high-intensity staining tumors for hENT1, RRM1, and RRM2, respectively. Based on our formula for assessing overall expression, the majority of patients (85% n = 81) had tumors that exhibited high hENT1 expression. RRM1 expression was more evenly split, with 40% (n = 38) of patients exhibiting high expression. RRM2 was overexpressed in 17% (n = 16) of patients. Figure 1 depicts sample IHC stains for patients with low- and high-expression profiles of hENT1, RRM1, and RRM2. The expression of ERCC1 in this patient population has been reported previously.18
Survival Analysis: All Patients
High hENT1 expression was associated with a reduced RFS (9.5 months vs 44.5 months; P = .029) (Fig. 2a), but this association did not maintain significance with regard to OS (15.2 months vs 19.5 months; P = .175) (Fig. 2b). High RRM1 expression was not associated with either RFS or OS (Figs. 2c-2d). High RRM2 expression was associated with reduced RFS (6.9 months vs 16.0 months; P ≤ .0001) (Fig. 2e) and decreased OS (9.1 months vs 18.4 months; P ≤ .0001) (Fig. 2f). Survival analysis for ERCC1 expression was previously reported with a median follow-up of 25 months.18 When a median follow-up of 49.3 months was examined, the same relations were observed: high ERCC1 expression was associated with reduced RFS (6.1 months vs 14.9 months; P = .037) (Fig. 2g) and decreased OS (8.9 months vs 18.1 months; P = .032) (Fig. 2h). The subset of patients who had tumors with a low expression of both RRM2 and ERCC1 (n = 66) had the best RFS, with a median time to disease recurrence of 16.3 months compared with 7.3 months for those with high expression of 1 biomarker versus 0.6 months for those with tumors that demonstrated high expression of both RRM2 and ERCC1 (P ≤ .0001) (Fig. 3a). Similarly, the median OS for patients with low expression of both RRM2 and ERCC1 was 21.6 months compared with 9.1 months for those patients with high expression of 1 biomarker and 8.5 months for those with high expression of both biomarkers (P ≤ .0001) (Fig. 3b).
On univariate analysis, high hENT1, high RRM2, and high ERCC1 expression as well as other known adverse tumor factors (Table 2) were found to be associated with both decreased RFS and OS. When examined in a multivariate analysis, only high RRM2 and high ERCC1 tumor expression persisted as negative prognostic factors for both RFS and OS (Table 2).
|HR (95% CI)||P||HR (95% CI)||P||HR (95% CI)||P||HR (95% CI)||P|
|Tumor size||1.2 (1-1.5)||.05a||1.1 (0.8-1.3)||.67||1.3 (1-1.6)||.02||1.1 (0.9-1.5)||.28|
|Positive margin||1.7 (1-2.8)||.05||1.6 (0.9-2.9)||.12||1.7 (1-0.9)||.04||1.5 (0.8-2.7)||.19|
|Positive LN||1.7 (1.1-2.8)||.03||1.6 (0.9-2.9)||.11||1.9 (1.2-3.1)||.01||1.7 (0.9-3)||.08|
|PNI||1.0 (0.5-2.1)||.92||1.0 (0.5-2.1)||.97||3.0 (1.1-8.2)||.03||3.3 (1.2-9.4)||.03|
|LVI||1.5 (0.9-2.4)||.10||1.3 (0.8-2.2)||.34||1.7 (1.1-2.8)||.02||1.4 (0.9-2.4)||.17|
|High hENT1||2.3 (1.1-5.1)||.04||1.5 (0.7-3.5)||.32||1.7 (0.8-3.5)||.18||0.8 (0.4-1.8)||.61|
|High RRM1||1.0 (0.7-1.7)||.87||—b||—b||0.9 (0.6-1.4)||.64||—b||—b|
|High RRM2||3.1 (1.6-6.2)||.001||3.2 (1.5-6.8)||.002||3.3 (1.8-6.2)||≤.0001||2.8 (1.4-5.6)||.003|
|High ERCC1c||1.9 (1.03-3.5)||.04||2.7 (1.4-5.4)||.004||1.9 (1.04-3.4)||.04||3.0 (1.5-5.8)||.002|
Survival Analysis: Patients Receiving Adjuvant Therapy
A total of 74 patients received some form of adjuvant therapy, with the exact regimen known in 53 patients (72%). Of those 53 patients, most regimens were gemcitabine-based (62%), whereas 45% received 5-fluorouracil and 4% of patients received a platinum agent. The exact regimen was not known in 28% of patients. In a subset analysis of only those 74 patients who received adjuvant therapy after resection, high tumor expression of hENT1 was found to be associated with decreased RFS (20.8 months vs 40.3 months; P = .04) (Fig. 4a), but not with shortened OS (27 months vs 36.1 months; P = .20) (Fig. 4b). RRM1 expression was not significantly associated with either RFS or OS (Figs. 4c-4d). High RRM2 expression was associated with both reduced RFS (8.3 months vs 15.0 months; P = .001) (Fig. 4e) and decreased OS (11.6 months vs 18.4 months; P = .001) (Fig. 4f). Survival analysis for ERCC1 expression was previously reported within this population with a median follow-up of 25 months.18 In the current study, with a median follow-up of 49.3 months, the same relations were observed: high ERCC1 tumor expression was associated with reduced RFS (4.3 months vs 15.0 months; P = .01) (Fig. 4g) and decreased OS (8.5 months vs 18.4 months; P ≤ .0001) (Fig. 4h). When the same multivariate model was applied to this subset of patients who received adjuvant therapy, high RRM2 and ERCC1 expression were the only negative prognostic factors found to be associated with both decreased RFS and OS (Table 3).
|HR||95% CI||P||HR||95% CI||P|
In the adjuvant setting after complete resection, a molecular biomarker with prognostic value for survival outcomes can guide further treatment decisions, such as selecting patients for adjuvant therapy or stratifying patients in a randomized trial. A biomarker with predictive value may help to optimize therapy by selecting an individualized chemotherapy regimen. Such biomarkers are particularly important in patients with PAC, a majority of whom develop disease recurrence and ultimately die of their disease. Although preclinical studies have identified potential biomarkers, there is a paucity of information regarding clinically applicable biomarkers in patients with early stage disease.
hENT1 expression has been assessed by IHC in a variety of malignancies, including breast, biliary tract, and lung cancers.11, 19, 20 In patients with advanced stage PAC, those receiving gemcitabine therapy who had detectable levels of hENT1 by IHC were found to have a significantly longer median survival than their counterparts with low or undetectable hENT1 expression.21 Two studies have examined hENT1 as a predictive factor for outcomes in patients with resected PAC who were treated with gemcitabine chemotherapy.22 In an analysis of a subset of patients enrolled in the RTOG 9704 trial evaluating adjuvant gemcitabine versus 5-fluorouracil, Farrell et al reported that high expression of hENT1 was predictive of improved OS, but only in those patients treated with gemcitabine.23 In the current study, high hENT1 expression was found to be a poor prognostic factor for early disease recurrence, although this association did not persist for OS. Given the retrospective nature of this study, the predictive value of hENT1 cannot be discerned because the adjuvant therapy regimen was not standardized. Further studies are necessary to evaluate the role of hENT1 as a prognostic and/or predictive biomarker in patients with early stage PAC. It is possible that high hENT1 expression confers an independent prognostic value that is incongruous with its predictive value for gemcitabine-based therapy, as has been reported for ERCC1 in non-small cell lung cancer (NSCLC) and platinum therapy.11
The majority of research regarding RRM1 has been conducted in patients with NSCLC. The prognostic implications of the expression profile of RRM1 are dependent on the stage of NSCLC; patients with early stage disease appear to benefit from high expression, whereas those who demonstrate high expression and advanced disease while receiving platinum-based regimens have a worse prognosis.11 In vitro studies in lung, breast, colon, biliary tract, and pancreas cancer cell lines have found that high RRM1 expression is associated with gemcitabine resistance.12 In the clinical setting, the role of RRM1 expression in PAC is less well defined. Two studies have demonstrated an association similar to that observed in NSCLC: high RRM1 expression was correlated with improved OS in patients with early stage pancreas cancer,24 but for patients with advanced disease, only those with low RRM1 expression derived any benefit from gemcitabine-based chemotherapy.13, 24 Conversely, others have been unable to find a significant relation between RRM1 expression and outcome in patients with advanced stage pancreas cancer.12, 25 The results of the current study are consistent with the latter group, showing no association between survival outcomes and tumor RRM1 expression in patients with early stage disease.
To the best of our knowledge, only limited data exist regarding the M2 subunit of ribonucleotide reductase, RRM2, and its relation with outcomes in cancer patients. Drugs targeting RRM2 have been shown to potentiate the effects of gemcitabine in lung cancer cell lines26 and in patients with mantle cell lymphoma.27 Unfortunately, phase 2 trials with similar anti-RRM2 drugs in combination with gemcitabine in patients with advanced stage PAC have demonstrated substantial drug toxicity without any survival advantage.28, 29 In patients with advanced stage ovarian cancer, RRM2 overexpression has been associated with shorter OS.30 Duxbury et al reported that high RRM2 expression is associated with gemcitabine resistance in PAC cell lines and found that small interfering RNA (siRNA) directed toward RRM2 suppressed tumor growth in an orthotopic xenograft model.14 To our knowledge, with the exception of the current report, only 1 other study to date has assessed the value of RRM2 expression by IHC in patients with PAC. In 31 patients with advanced stage pancreas cancer, Itoi et al reported that patients with low RRM2-expressing tumors receiving gemcitabine therapy had a longer median survival and greater response to gemcitabine compared with patients with high RRM2-expressing tumors.31 To our knowledge, the study reported herein is the first to examine the role of RRM2 expression in patients with early stage PAC. High RRM2 expression was associated with a shorter time to disease recurrence and decreased OS. Furthermore, high tumor RRM2 expression was a stronger predictor of poor outcome compared with all other known adverse tumor factors, including LN involvement and a positive resection margin. This association persisted in patients who received adjuvant therapy, the majority of whom received a gemcitabine-based regimen. These results suggest that high RRM2 expression is a valuable prognostic marker for early recurrence and decreased survival in patients with early stage PAC. RRM2 expression may be useful in personalizing an adjuvant treatment regimen, but further investigation in a prospectively controlled study is needed to determine its true predictive value.
hENT1, RRM1, and RRM2 were selected for study as potential biomarkers because of their potential roles in gemcitabine metabolism and efficacy. Similarly, ERCC1 has been identified both as a valuable prognostic and predictive biomarker because of its role in platinum compound metabolism.18 We sought to delineate interactions between biomarker expression, because more than 1 biomarker may be necessary to create a clinically useful expression profile.25 Based on multivariate Cox regression analysis, high RRM2 and ERCC1 expression maintained independent prognostic value for early recurrence and decreased survival. Patients with tumors with a low expression of both RRM2 and ERCC1 appeared to fare the best (Fig. 3), suggesting that the combination may be a more clinically useful stratification tool than a single biomarker expression profile.
The IHC technique used in the current study was semiquantitative. However, each biomarker expression was scored by 1 experienced pathologist who was blinded to patient outcomes. The persistence of a negative prognostic effect of high tumor expression of RRM2 and ERCC1 on disease recurrence and survival after accounting for the 5 most common adverse tumor factors (size, margin, LN status, PNI, and LVI) supports the current study findings indicating that high tumor expression of RRM2 and ERCC1 are indeed negative prognostic factors after surgical resection.
The possibility of personalized chemotherapy based on a biomarker expression profile is an attractive one, and indeed a prospective randomized trial using combinations of biomarkers to stratify patients to treatment arms has been successfully conducted in patients with advanced NSCLC, with promising results.32 Simon et al randomized chemotherapy-naive patients with advanced stage NSCLC to doublet chemotherapy based on expression patterns of RRM1 and ERCC1, and compared their results with an aggregate of clinical trials using the same drug combinations without examining biomarkers. The patients in the personalized therapy group had statistically significant improvements in response rate, OS, and progression-free survival.32
Although gemcitabine therapy remains the cornerstone of adjuvant treatment in patients with PAC,1-3 it remains relatively ineffective for the population as a whole. It is possible that differential tumor expression of RRM2 and ERCC1 may contribute to the various results observed with gemcitabine and cisplatin combination therapy in patients with advanced disease. Stratification of patients by RRM2 and ERCC1 expression may allow for the identification of a population who will truly benefit from the doublet regimen. Furthermore, it may identify those patients who will only experience increased toxicity and morbidity without any significant survival advantage. Use of RRM2 and ERCC1 biomarker expression to predict response to the combination of gemcitabine and cisplatin needs to be tested in a prospective trial.
The results of the current study demonstrate that high tumor RRM2 expression is an independent negative prognostic factor for early recurrence and decreased survival in patients with early stage PAC. Our previous finding that ERCC1 overexpression had a similar negative prognostic value18 has been confirmed with longer patient follow-up and persists in the context of other biomarker expression. This finding lays the groundwork for the future examination of these biomarkers in prospective trials.
Supported by a Winship Cancer Institute Investigator-Initiated Trial Seed Grant.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 32Preliminary indication of survival benefit from ERCC1 and RRM1-tailored chemotherapy in patients with advanced nonsmall cell lung cancer: evidence from an individual patient analysis [published online ahead of print October 25, 2011]. Cancer. doi: 10.1002/cncr.26522., , , et al.