Abstract
Background/Aim: A possible predictive impact of ribonucleotide-reductase subunit-1 (RRM1) on vinorelbine efficacy in non-small cell lung cancer (NSCLC) has been previously reported. The present study aimed to further explore this finding in malignant pleural mesothelioma (MPM). Materials and Methods: Seventy-one patients with MPM receiving first-line chemotherapy with cisplatin-vinorelbine (CiV group, n=54) or carboplatin-pemetrexed (CaP group, n=17) were included. Formalin-fixed paraffin-embedded tumor specimens were analyzed by immunohistochemistry (IHC) for RRM1 expression using an H-score. Results: In 66 patients eligible for IHC, the H-score was ≥upper quartile in 21 (RRM1-positive) and <upper quartile (RRM1-negative) in 45 cases. The long-term (2-year) survival rate for patients with RRM1-negative MPM in the CiV-treated group was significantly superior to that for patients with RRM1-positive MPM (47% versus 13%, p=0.002). No difference occurred in the CaP-treated group. Conclusion: Our findings suggest a possible role of RRM1 in predicting efficacy of cisplatin–vinorelbine in MPM, supporting previous findings in NSCLC.
Malignant pleural mesothelioma (MPM) arises from the mesothelium that lines the pleural cavity and development of this type of cancer is often associated with previous asbestos exposure (1, 2). Due to the long latency from asbestos exposure to diagnosis, the incidence of MPM is rising, despite the European ban on asbestos use being issued three decades ago (3). Advanced MPM has a poor prognosis and only a small minority of patients are diagnosed at an early stage at which curative multimodality treatment is possible (4).
Personalized treatment, i.e. choosing the right drug for the right patient, has become an area of intense research in cancer therapy in order to improve results. The choice of pharmacological cancer treatment is no longer mainly directed by the histological appearance of the tumor but increasingly also by the molecular characteristics of the disease. A well-known example is the implementation of first-line treatment with tyrosine-kinase inhibitors (TKI) against non-small cell lung cancer (NSCLC) harboring epidermal growth factor (EGFR)-activating mutations (5). These drugs can produce dramatic and long-lasting responses when given to the right patients but only a minority of patients have the appropriate molecular profile for this therapy. Mezzapelle et al. recently investigated the incidence of activating EGFR mutations in a series of 77 patients with MPM and observed that none of the cases harbored a mutation that warranted use of TKI treatment (6). Similarly, novel targeted next-generation sequencing analysis of cancer genes in advanced MPM showed no targetable EGFR mutations (7).
Although different targeted therapies have been on trial in MPM, none has so far produced convincing responses (8). Several novel molecular and immunological therapeutic strategies are currently being explored with or without combined chemotherapy and hold promise for future treatment of MPM (9). However, so long as patients with MPM do not show clear benefit from such new therapeutic options, the cornerstone of MPM treatment will remain classical chemotherapy.
Today the most commonly accepted first-line treatment for MPM is combination chemotherapy with a platinum compound and pemetrexed, although several other regimes have shown similar activity in terms of response and survival rates (10, 11).
Some retrospective studies, including ours, have shown that a possible means of improving prognosis of this dismal disease might lie in the use of biomarkers to predict response to the classical chemotherapeutics and thereby direct treatment choices (12-14). Since the predictive value of the proposed biomarkers in certain cases has not been confirmed by other research groups (15, 16) and is still awaiting validation through prospective studies, identification of other predictive biomarkers is required.
Ribonucleotide reductase is an enzyme essential for DNA synthesis, converting ribonucleoside diphosphates into deoxyribonucleoside diphosphates. The enzyme consists of a large subunit (RRM1) and a small subunit (RRM2). The protein is located in the cytoplasm and is present throughout the cell division cycle and only down-regulated in senescent cells (17, 18). Overexpression of RRM1 has conventionally been associated with resistance towards the chemotherapeutic agent gemcitabine, that is a potent inhibitor of ribonucleotide reductase (19).
Our group recently published a study on the predictive value of immunohistochemically (IHC)-detected RRM1 in a cohort of patients with NSCLC from a randomized phase III trial, comparing triplet chemotherapy (paclitaxel, cisplatin, gemcitabine) to standard doublet therapy (cisplatin, vinorelbine). Surprisingly, we found that increased expression of RRM1 was associated with significantly reduced progression-free survival (PFS) and overall survival (OS) only in patients receiving the regimen of cisplatin and vinorelbine (20). This finding was unexpected, since vinorelbine is a spindle poison and resistance towards this agent has not been associated with RRM1 overexpression before.
In order to further investigate this finding, we tested the association between RRM1 expression assessed by IHC and outcome in a cohort of cisplatin- and vinorelbine-treated patients with MPM.
Materials and Methods
Patient population. The study population consisted of 54 consecutive patients with inoperable MPM enrolled into a phase II trial conducted at our Institution between February 2003 and September 2006 (trial registration number NCT00272558).
Treatment was 25 mg/m2 vinorelbine i.v. weekly and 100 mg/m2 cisplatin i.v. every 4 weeks (CiV) for a maximum of six cycles (21). Additionally, a separate group of 17 patients with inoperable MPM treated between 2008 and 2010 was included in the study. These patients were not part of a clinical trial and they were treated according to the standard protocol at our Institution with carboplatin area under the curve (AUC) 5 and pemetrexed at 500 mg/m2 day 1 every 3 weeks for six courses (CaP).
All patients had histologically-verified MPM and were considered inoperable because of advanced age, anatomical extension of disease (N2/N3 nodal status), non-epithelioid (biphasic or sarcomatoid) histology, or physiological inoperability (poor cardiac or pulmonary function etc.). Patients had measurable disease, Eastern Cooperative Oncology Group (ECOG) performance status 0-2, an estimated survival expectancy of 3 months or more, age 18 years or more, and all gave their informed written consent.
The biomarker study was conducted in accordance with the Declaration of Helsinki and was approved by the The Research Ethics Committee of the Capital Region of Denmark (approval number: H-3-2011-135) and the Danish Data Protection Agency (approval number: 2011-41-6825). Staging was carried-out according to the system defined by the International Mesothelioma Interest Group (IMIG) (22).
Patients were assessed with spiral computerized tomography (CT) scans at baseline and every 9 weeks during treatment. After completion of treatment, CT scans were performed every 2 months until progression or death. Patients were followed up until 2011; patients still alive were censored at the end of 2011.
Tissue samples. Formalin-fixed paraffin-embedded (FFPE) bioptic tumor-specimens from the enrolled patients were obtained from Departments of Pathology throughout Denmark. The corresponding pathology reports from the Danish National Pathology Data-Base (Patobank) and hospital records were retrospectively interrogated to obtain the diagnoses. Hematoxylin and eosin-stained sections from each specimen were reviewed by two independent observers (ZGZ and ES-R), in order to confirm the diagnosis and evaluate the suitability of the samples for IHC analysis.
Immunohistochemistry. Thin (2 μm) sections were cut from each biopsy specimen and mounted on coated glass slides. Sections were de-paraffinized and heat-induced antigen retrieval was performed in DAKO-PT-link module, using target retrieval solution pH 9 for 20 min at 97°C (DAKO, Glostrup, Denmark). Endogenous peroxidase was blocked and sections were incubated at room temperature for 20 min with rabbit polyclonal anti-human RRM1 (Protein Tech Europe, Manchester, UK) diluted 1:50. Staining was visualized with EnVision Flex+ kit (DAKO) with diaminobenzidine as chromogen. Sections were counterstained with Meyer's hematoxylin for 1 min. The immunostaining was carried out using a DakoAutostainer PLUS.
Background staining was assessed by omitting the primary antibody, in which case no staining was observed. Normal human liver and normal human lymph node served as positive controls (23). The staining pattern was predominantly cytoplasmic, but weaker nuclear staining was also observed. The fibrous connective tissue surrounding the larger vessels served as negative controls and here no staining was seen (18). A positive and negative control are shown in Figure 1. The reactivity in non-neoplastic mesothelium was evaluated by staining 34 cases of mesothelial hyperplasia from patients with relapsing spontaneous pneumothorax.
IHC evaluation. The staining was separately evaluated by two independent observers (ZGZ and ES-R), blinded to the clinical data. In discordant cases, consensus was reached by the use of a two-headed microscope. Staining was assessed using a semiquantitative H-score as originally proposed by Olaussen and as we previously had used for RRM1 in NSCLC and other predictive biomarkers in MPM (12, 13, 20, 24). Briefly, the staining intensity was evaluated against a positive control and assigned a value of between 0 and 3 (0=no staining, 1=weaker than the control, 2=same intensity as the control, 3=stronger than the control). Then the percentage of positively stained tumor cells was evaluated based on at least 500 cells per sample, and categorized into three classes (0.1 <10%, 0.5=10-50%, 1 >50%). This proportion score was then multiplied by the staining intensity to obtain the final H-score with values ranging from 0-3.
Both nuclear and cytoplasmic staining was assessed and a liver sample stained along with the tumor specimens served as an external control with staining intensity 2 (Figure 1A). Infiltrating tumor lymphocytes served as internal controls on tissue sections.
The upper quartile value of the H-score was a priori chosen to dichotomize the samples into positive (H-score ≥upper quartile) and negative (H-score <upper quartile) categories, as previously described (13).
Statistics. The primary end-point of the study was OS, defined as the time from the onset of treatment until death from any cause. Estimates of OS were obtained with the Kaplan–Meier method. Multivariate Cox regression was used to investigate the association of RRM1 status with OS adjusting for the following known risk factors: patient's age and gender, tumor histology (epithelioid versus non-epithelioid), performance status (ECOG performance status score 0, 1 or 2) and disease stage (Ia/Ib/II, III or IV). Results are presented as hazard ratios (HR) with 95% confidence interval (CI).
Results
Patients' characteristics. Seventy-one patients were originally included in the study: 54 were treated with CiV as described by Sørensen et al. (21). Seventeen patients were treated with CaP. The FFPE diagnostic tumor tissue was available for 93% of the patients, 49/54 in the CiV group and 17/17 in the CaP group. The final study population thus consisted of 66 patients.
Patients were predominantly males (86%), with MPM of epithelioid histology (71%), PS 0-1 (90%) and IMIG stages III or IV (76%). The median age at diagnosis was 64 years (range=31-78 years). Patients' characteristics are summarized in Table I. The two patient populations (CiV and CaP) were comparable with respect to gender, PS, and stage distribution. The mean age was significantly higher in the CaP group than the CiV group (67 and 62 years, respectively, p=0.04). The CaP group also numerically included more non-epithelioid cases although this difference was not statistically significant. These differences in patient characteristics and tumor demographics might account for the statistically significant difference in survival seen between the two groups, corresponding to a median survival of 17 months and 12 months in the CiV- and CaP-treated groups, respectively.
Upon disease progression, 21 of the CiV patients received second-line monotherapy with pemetrexed, among these patients 13 and 8 had RRM1-negative and RRM1-positive tumor samples, respectively (p=0.4).
IHC evaluation. The majority of the 34 cases of non-neoplastic mesothelium from patients with post-pneumothorax mesothelial hyperplasia displayed strong predominantly cytoplasmic immunoreactivity, as illustrated in Figure 1B. Indeed, 24 out of 34 samples had the maximum H-score of 3 (median H-score 3), an example of which is shown in Figure 1B.
Patient demographics and tumor characteristics according to treatment.
Suitable FFPE tumor tissue samples were available from 66 MPM cases. The immunoreactivity in the tumor samples was more heterogeneous, with a larger range of H-scores (0.5-3). We beforehand chose the upper quartile value of the tumor sample H-scores as the cut-off point to dichotomize the samples (13). The upper quartile H-score was 3, yielding 21 RRM1-positive cases and 45 RRM1-negative cases. Figure 1C and D show an RRM1-positive and -negative case, respectively.
Survival according to RRM1 status. The distribution of patient and tumor characteristics were well-balanced between the RRM1-positive and -negative cases, both in the collective cohort and separately in the CiV and CaP groups, as shown in Table II. In univariate analysis, there were no differences in OS according to RRM1 status (log-rank p=0.2), although a significantly increased 2-year survival rate was observed among the RRM1-negative cases, among whom 38% of patients survived more than 2 years compared to 9.5% in the RRM1-positive group (p=0.02). When looking at the association between RRM1 status and 2-year survival rate in the two different treatment groups, we observed that this survival advantage was mainly observed in the CiV cohort, with 2-year survival rates of 47% vs. 13% (p=0.03) in the RRM1-negative and -positive groups, respectively. In the CaP group, the RRM1-negative cases had a 2-year survival of 9%, as opposed to 0% in the RRM1-positive cases, however, this difference was not significant (p=1.00).
Ribonucleotide-reductase subunit-1 (RRM1) staining. A: Human liver specimen (external control). B: Non-neoplastic mesothelial hyperplasia, H-score 3 (staining intensity 3, proportion score 1). C: RRM1-positive epithelioid malignant pleural mesothelioma, H-score 3 (staining intensity 3, proportion score 1). D: RRM1-negative epithelioid malignant pleural mesothelioma, H-score 1 (staining intensity 1, proportion score 1).
The Kaplan–Meier plot presented in Figure 2 depicts OS according to RRM1 status in the different treatment groups. By this method, we observed that RRM1 negativity distinguishes a group of patients who experience long-term survival after treatment with the CiV regime (p=0.002).
A multivariate Cox regression analysis showed that RRM1-positive patients in the CiV group had a significantly shorter survival than RRM1-negative patients receiving the same treatment (HR=2.26, 95% CI=1.12-4.54, p=0.02). The only other factor with a significant influence on OS was the PS at the onset of treatment (HR=2.38, 95% CI=1.08-5.25, p=0.03). The Cox regression analysis for OS in the CiV group is shown in Table III.
To account for the possible influence of second-line pemetrexed treatment, we fitted a Cox proportional hazard model, which included an RRM1 by second-line pemetrexed interaction term. This term was not significant (p=0.6).
In the CaP group, there was no significant association between RRM1 status and OS upon Cox regression analysis. Here, only PS emerged as a borderline significant factor influencing OS (p=0.07).
Overall survival according to ribonucleotide-reductase subunit-1 (RRM1) status. CiV: Cisplatin and vinorelbine, CaP: carboplatin and pemetrexed.
Patient demographics and tumor characteristics according to ribonucleotide-reductase subunit-1 (RRM1) status.
Cox-regression analysis for overall survival according to ribonucleotide-reductase subunit-1 (RRM1) status in the cisplatin plus vinorelbine-treated group (CiV group).
Discussion
This study aimed at further exploring the possible role of RRM1 protein expression in predicting the efficacy of vinorelbine-containing treatment regimens, as recently shown in a cohort of patients with NSCLC (20). In the latter study, bioptic tumor specimens from a phase III trial in advanced NSCLC were evaluated for RRM1 protein expression by IHC and overexpression of RRM1 was found to be a negative prognostic factor only in those patients assigned to cisplatin–vinorelbine therapy and not to the gemcitabine-containing treatment regimen, as would have been expected.
The results of the present study on MPM support our previous observations in NSCLC (20). Herein we found that low RRM1 expression is associated with a significantly increased survival rate in patients with MPM treated with cisplatin and vinorelbine, an association that is not seen in patients treated with carboplatin and pemetrexed. On multivariate analysis, RRM1 status emerged as an independent prognostic factor with a negative influence on survival in patients treated with cisplatin–vinorelbine (HR=2.26, 95% CI=1.12-4.54, p=0.02).
These results are in accordance with recently published observations by Szulkin et al., who used primary MPM cell cultures from pleural effusions and an ex vivo drug sensitivity assay to test the response of the samples to 32 different drugs (25). Immunocytochemical analysis on the primary cell cultures showed that high RRM1 expression correlated with general drug resistance, especially resistance towards the anti-microtubule agents, and with the patients' survival time.
The mechanistic explanation for these and our findings is not straightforward, as ribonucleotide reductase and consequently RRM1 are essential for DNA synthesis and part of the DNA repair system (19, 26). Vinorelbine is a semi-synthetic vinca alkaloid, which is believed to exert its antitumor effects by inhibiting microtubule assembly and thereby possibly interfering with microtubule dynamics, blocking mitosis, and leading to apoptosis (27).
However, Rabbani-Chadegani et al. showed that vinorelbine also binds with high affinity to histone proteins in hepatocytes (28), thereby interfering with DNA and chromatin assembly, while Fukuoka et al. reported that vinorelbine can inhibit radiation-induced DNA repair in SCLC cell lines (29). Therefore, the existence of alternative vinorelbine targets might explain why overexpression of a protein of the DNA synthesis and repair pathway, such as RRM1, can predict resistance towards this spindle poison.
Although these notions are intriguing, caution should be taken in interpreting our results, as the current study presents several limitations. One of them is the small number of patients. The group treated with CaP comprised only 17 patients, thus increasing the risk of type 2 errors, and rejecting an effect of RRM1 status in this group also. We did see a numerical trend towards increased 2-year survival among the RRM1-negative patients in the CaP group but this difference was not statistically significant. Had this group contained more patients, however, it might have reached statistical significance, implying that RRM1 may be a prognostic rather than a predictive biomarker. Moreover, the platinum compound used was different in the two treatment groups, one being cisplatin and the other carboplatin, both known to be DNA-damaging agents. This difference might somehow influence the observed effect of RRM1 expression level on survival, as one might speculate that the latter may also be due to the platinum part of the therapies. In this case, RRM1 could possibly be more involved in the repair of DNA damage induced by cisplatin than by carboplatin. Obviously, further investigations are required to verify this possibility.
Another explanation for the results might rely on the methodology used in the study. We chose IHC because it is a well-established, cheap and efficient method for determining protein expression in archival FFPE tissue samples. The primary antibody was chosen because this was the antibody used in previous publications on the predictive value of RRM1 (20, 30-32). This antibody is polyclonal, that may raise concern on its specificity, since it is a heterogeneous mixture of different antibody clones recognizing different epitopes of the RRM1 protein. This in general poses a greater risk of crossreactivity with other intracellular structures. Thus, the antibody might react with an intracellular antigen, unrelated to RRM1, which is involved in vinorelbine metabolism, a hypothesis that may require further investigations.
An additional concern in IHC biomarker research is reproducibility. This issue has recently been addressed for excision repair cross-complementation group 1 enzyme (ERCC1), another potential biomarker in both MPM and NSCLC (12, 24, 33). By re-staining the biopsies used in their original International Adjuvant Lung Cancer Trial study (24), which had proposed ERCC1, as assessed by the widely used monoclonal antibody clone 8F1, as a predictive marker for platinum resistance in NSCLC, Friboulet et al. were unable to reproduce their original results using new batches of the 8F1 antibody (34). This suggests that the antibody reactivity had changed over time and questions the specificity of clone 8F1, making it a possible source of false prognostic and predictive value of ERCC1 expression. Additionally, only one out of four functionally distinct ERCC1 isoforms was linked to DNA repair activity and platinum resistance and thereby could be considered a potentially useful predictive biomarker in clinical samples for platinum-based chemotherapy. None of the commercially available ERCC1 antibodies was able to specifically detect this unique functional ERCC1 isoform because of cross-reactivity with the other isoforms (34). Finally, the 8F1 antibody reacts not only with ERCC1 but also with other independent antigens (35). Therefore, the usefulness of 8F1-based IHC for ERCC1 in guiding therapeutic decisions is doubtful at the moment, because the 8F1 antibody seems to lack reliable sensitivity and specificity.
The results of our current study are, however, in favor of the antibody used. Indeed, we were able to reproduce in the current cohort of MPMs the predictive impact of RRM1 previously observed in a cohort of NSCLC (20), using the same antibody and the same evaluation scheme. As mentioned above, our results are also consistent with the correlation between RRM1 expression and resistance to vinca alkaloids in primary MPM cells from pleural effusions reported by Szulkin et al., who used a similar RRM1 antibody in their study (25).
Although firm conclusions cannot be drawn from this small study, the results suggest that high RRM1 expression is associated with adverse outcomes in patients with MPM treated with cisplatin and vinorelbine, thereby implying a possible predictive value of the marker for this drug combination.
Acknowledgements
The work was supported by the Danish Cancer Society and the Danish Lung Association. We thank Lone Svendstrup and Anne Jørgensen at the Department of Pathology of Rigshospitalet for expert technical assistance.
- Received July 3, 2015.
- Revision received September 22, 2015.
- Accepted October 19, 2015.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
