Breast Cancer Research and Treatment

, Volume 131, Issue 3, pp 809–818

Deoxycytidine kinase is overexpressed in poor outcome breast cancer and determines responsiveness to nucleoside analogs

Authors

  • Ernst-Jan Geutjes
    • Division of Molecular Carcinogenesis, Center for Biomedical Genetics and Cancer Genomics CenterThe Netherlands Cancer Institute
  • Sun Tian
    • Agendia BV
  • Paul Roepman
    • Agendia BV
    • Division of Molecular Carcinogenesis, Center for Biomedical Genetics and Cancer Genomics CenterThe Netherlands Cancer Institute
    • Agendia BV
Preclinical Study

DOI: 10.1007/s10549-011-1477-3

Cite this article as:
Geutjes, E., Tian, S., Roepman, P. et al. Breast Cancer Res Treat (2012) 131: 809. doi:10.1007/s10549-011-1477-3

Abstract

Only a minority of breast cancer patients responds to chemotherapy and we lack predictive biomarkers that help to select a patient-tailored therapy that takes into consideration the molecular heterogeneity of the cancer type. Responsiveness to the clinically important nucleoside analogs gemcitabine and decitabine may be critically determined by Deoxycytidine kinase (DCK) expression as this enzyme is required to convert the inactive prodrugs into their pharmacologically active forms. Here, we examined whether DCK is differentially expressed in breast cancer and evaluated whether DCK expression levels control responsiveness to these nucleoside analogs in vitro by experimentally modulating DCK expression levels. We examined DCK expression in gene expression data sets of breast tumors including the series of 295 consecutive patients that have been classified into low or high risk for recurrence using the MammaPrint 70 gene profile. We found that DCK is expressed at higher levels in patients having poor clinical outcome as judged by the MammaPrint assay. As such, patients that have a poor prognosis may thus be susceptible to treatment with nucleoside analogs. In support of this, we found a causal relationship between DCK levels and sensitivity to these nucleoside analogs in breast cancer cell lines. The data indicate that breast cancers that are at high risk of recurrence express higher levels of DCK, which we find to be strongly correlated to a favorable response to nucleoside analogs. The data suggest that DCK expression in breast cancer could be exploited to select patients that are likely to respond to treatment with nucleoside analogs.

Keywords

DCKGemcitabineNucleoside analogsBreast cancerBiomarker

Introduction

One in eight women will develop breast cancer over a lifetime. Although mortality rates have been declining due to improved diagnostics and treatment, 500,000 women died of breast cancer worldwide in 2004 (1% of all deaths) [1]. A major impediment to the effective treatment of breast cancer is the difficulty to predict which patients will respond to a given therapy. Most patients are subjected to a combination of surgery, irradiation and adjuvant systemic therapy with chemotherapeutic and/or targeted agents. The choice of therapy is not based on any real consensus, which is underscored by the fact that treatment regimens can differ dramatically for women with tumors that score identical for clinical parameters such as tumor size, grade, and lymph node (micro) metastases. Moreover, treatment is often surprisingly ineffective, for example radiotherapy and adjuvant polychemotherapy each individually increases 10 years survival by only 5% [2, 3]. This clearly illustrates the need for patient-tailored cancer therapy. Moreover, we need to identify the right therapy at an early stage before more malignant clones that have acquired traits that allow metastasis arise within the tumor mass [4, 5]. As metastatic breast cancer is practically incurable, it is therefore of utmost importance to choose the most effective therapy at an early stage. Selection of the right therapeutic strategy requires the availability of predictive biomarkers that help foretell responses to specific therapies [6].

The difficulty in predicting responsiveness to therapy is most readily explained by tumor heterogeneity and a poor ability to identify distinct molecular subtypes by appropriate diagnostics [6, 7]. The sequential acquisition of epigenetic and genetic alterations results in deregulation of specific signaling pathways in the tumor that can critically alter drug responsiveness. As such, patients with (epi-)genetically distinct tumors may have similar tumors on macromolecular level but often respond differently to identical treatment regimens. For example, HER2 positive breast cancer patients (~15%) that are treated with the HER2/NEU targeting antibody trastuzumab are unlikely to respond to trastuzumab-based treatment when they have additional downstream mutations that activate the PI3K/AKT pathway [8]. These and other findings suggest that characterization of signaling pathways deregulated in the tumor may help in selecting the optimal therapeutic regimen. Indeed, breast tumors with BRCA1 or BRCA2 mutations (~10%) appear to be exquisitely sensitive to treatment with PARP-inhibitors [9]. However, for most of the conventional chemotherapeutic anticancer drugs used in the clinic the critical determinants of drug responsiveness are unknown [6, 10]. Thus, besides developing agents that target signaling pathways frequently deregulated in breast cancer, there should be an equal investment in finding biomarkers that predict chemotherapy responsiveness as this type of intervention is still the mainstay of cancer treatment in the clinic today.

In the past decade, several multi-gene diagnostic tests have been developed to help identify those early stage breast cancer patients that are at high risk for metastatic disease and hence require adjuvant chemotherapy. We previously published a 70-gene signature that classifies breast tumors as low or high risk for recurrence [11, 12]. Deoxycytidine kinase (DCK) is among the 70 genes in this MammaPrint prognostic signature. DCK is the rate-limiting enzyme for activation of deoxyribonucleoside prodrugs, which interfere with DNA synthesis and repair. Several nucleoside analogs such as 5′-aza-3′-deoxycytidine (5-azadC; decitabine), 2′-deoxy 2′,2′-difluorocytidine (dFdC; gemcitabine), and arabinofuranosyl cytidine (AraC; cytarabine) are important anticancer drugs and are substrates for DCK. The pharmacological activity of these nucleoside analogs is dependent on phosphorylation by DCK. As such, several studies have suggested that DCK activity is a critical determinant of responsiveness to nucleoside analogs [1317]. We report here that poor outcome breast cancers express DCK at higher levels than good outcome breast tumors. We therefore hypothesized that these patients might be more sensitive to treatment with these nucleoside analogs than patients having a good prognosis breast cancer. We report here that experimental modulation of DCK levels in various breast cancer cell lines critically alters sensitivity to 5-azadC, dFdC, and AraC. These results thus suggest a novel treatment option for patients with a MammaPrint 70 gene high risk profile.

Materials and methods

Compounds and culture conditions

Nucleoside analogs 5′-aza-2′deoxycytdine (5-azadC, sigma), 2′-deoxy 2′,2′-difluorocytidine (dFdC, Sigma), and arabinofuranosyl cytidine (AraC, Sigma) were dissolved in 50% acetic acid/dH2O and stored at −80°C. Stock solutions (20 mM/l) were diluted with the medium to the desired concentrations for specific experiments. MCF7 breast carcinoma cells, HCC1954, MDA-MB-231, SK-BR-3, BT-474, T47D breast carcinoma cells, and A549 lung carcinoma cells were purchased from ATCC and maintained in DMEM supplemented with 10% FBS (Greiner), 1% glutamine (Gibco), and 1% penicillin/streptomycin (Gibco). MCF10A breast epithelial cells were purchased from ATCC and maintained in DMEM/F12 (1:1, Gibco), supplemented with 10% FBS, 1% glutamine, 1% penicillin/streptomycin, 20 ng/ml EGF (BD Biosciences), 0.2% amphotericin B (Gibco), 50 μg/ml hydrocortisone (Sigma), and 10 μg/ml insulin (Sigma).

shRNA design and plasmid construction

Design of oligonucleotides was done as previously described [18]. Multiple independent synthetic oligonucleotides were designed to target the DCK transcript and purchased from Invitrogen. The oligonucleotides were cloned into the pRS retroviral vector as previously described [18]. More info and protocols on the oligo design and vector can be found at: http://www.screeninc.nl. The RNAi target sequences that were used for this study are as follows: pRS_DCK.2, 5′-GGGAAGAAATGAAGAGCAA-3′; pRS_DCK.4, 5′-GGATGTTAATGAAGACTTT-3′.

A pENTR221 vector expressing DCK was obtained from Invitrogen (Ultimate ORF clone collection). The DCK coding sequence was subcloned into a gateway compatible pQCXIP vector (Promega) using Gateway cloning according to the manufacturer’s protocol (Invitrogen).

Viral transduction

For all studies with retroviral infection presented, the subclones of each relevant line expressing the murine ecotropic receptor were generated. Retroviral transductions were performed in three rounds using Phoenix cells as producers of viral supernatants as described (http://www.stanford.edu/group/nolan/retroviral_systems/phx.html). Infected cells were selected for successful retroviral integration using 2 μg/ml of puromycin.

Cell proliferation assays

Single-cell suspensions were seeded into 6-well plates (5–10 × 103 cells/well) for colony formation assays. Cells were refreshed every 3 days with new medium in absence or presence of increasing concentrations of 5-azadC, dFdC, or AraC. At the endpoints of colony formation assays, cells were fixed with 4% formaldehyde (Merck), stained with 0.1% crystal violet (Sigma), and scanned. The growth curves were performed in triplicate according to the standard 3T3 protocol in the absence or presence of dFdC or 5-azadC.

RNA isolation and cDNA synthesis

RNA was isolated using TRIzol (Invitrogen) according to the manufacturer’s protocol. 2 μg total RNA was converted to cDNA using Superscript II (Invitrogen) according to the manufacturer’s protocol. cDNAs were diluted 1:6 in dH2O. Primer sets were designed to include exon–exon boundaries and tested for specificity by performing a meltcurve. QRT-PCR analysis was performed in a total volume of 20 μl, using 8 μl of the diluted cDNA, 1 μl of each primer [10 μM stock] and 10 μl FastStart Universal SYBR Green Master (ROX) (Roche). QRT-PCR was performed on the 7500 Fast Real-Time PCR System (Applied Biosystems) under the following conditions: 95°C for 10 min, 40 cycles of 95°C for 3 s, and 60°C for 30 s. Relative mRNA levels of each gene shown were normalized to the expression of at least two house keeping genes: GAPDH, PGK, RPL4, or ACTIN. The sequences of the primers for assays using SYBR Green master mix (Roche) are as follows: GAPDH_QPCR_Forward, 5′-AAGGTGAAGGTCGGAGTCAA-3′; GAPDH_QPCR_Reverse, 5′-AATGAAGGGGTCATTGATGG-3′; RPL4_QPCR_Forward, 5′-GCTCTGGCCAGGGTGCTTTTG-3′; RPL4_QPCR_Reverse, 5′-ATGGCGTATCGTTTTTGGGTTGT-3′; PGK_QPCR_Forward, 5′-ATTAGCCGAGCCAGCCAAAATAG-3′; PGK_QPCR_Reverse, 5′-TCATCAAAAACCCACCAGCCTTCT-3′; ACTIN_QPCR_Forward, 5′-GCTGGCACCCAGCACAA-3′; ACTIN_QPCR_Reverse, 5′-GCCGATCCACACGGAGTACT-3′; DCK_QPCR_Forward, 5′-GACCATCGTTCAGGTTTCTCA-3′; DCK_QPCR_Reverse, 5′-TCCTGAACCTGTTGCCAGAT-3′.

Protein extraction and western blotting

Cell pellets were resuspended in RIPA lysis buffer (150 nM NaCl, 50 mM Tris pH 8.0, 1% NP40, 0.5% DOC, and 0.1% SDS) and rocked for 30′ at 4°C. Cell lysates were cleared by centrifugation at 14,000 rpm for 10′ at 4°C. Cell lysates were corrected for protein concentration by performing a BCA protein concentration determination assay (PIERCE). Cell lysates were processed and loaded (40 μg/sample) on NuPAGE BIS–TRIS precast gels (Invitrogen) according to the manufacturer’s protocol. Gels were transferred to a PVDF membrane (Milipore) for 2 h at 4°C using 450 mA. Western blots were incubated with antibodies targeting DCK (Santa Cruz: sc-81245, 1:1000 in 5% BSA/TBST), α-tubulin (Santa Cruz: sc-5286, 1:5000 in 5% BSA/TBST), or GFP (Santa Cruz: sc-8334, 1:5000 in 5% BSA/TBST) o/n at 4°C.

Microarray analysis: NKI-AVL cohort

For the MammaPrint profiles, the mRNA level of DCK was retrieved from gene expression data sets of 295 patients with primary breast carcinomas. Values are presented as log2 DCK ratios versus MammaPrint Reference Pool [19]. Central line indicates median, the upper side, and the lower side of the box indicate upper quartile and lower quartile, respectively. Two black whiskers extend from the box out to the value of 1.5* interquartile range. Outliers are indicated as crosses. The corresponding MammaPrint prognosis profiles of these 295 patients were previously reported [12]. The analysis of DCK as a tumor marker was in full concordance with the REMARK guidelines [20].

Microarray analysis: oncomine

The mRNA levels of DCK were retrieved from gene expression data of seven normal breast samples and 40 ductal breast carcinoma samples using Oncomine [21]. The thick bars in the boxes are average expression levels. The error bars are above or below the boxes, and the range of expression levels is enclosed by two dots.

Results

Deoxycytidine kinase is overexpressed in breast cancer and is associated with poor outcome

DCK is one of the 70 genes used in the MammaPrint prognostic classifier to identify primary breast cancer patients that are at high risk for recurrence [11]. We examined DCK expression in the series of 295 consecutive patients that have been classified into low or high risk for recurrence (MammaPrint low risk vs. MammaPrint high risk) [12]. We found that DCK is expressed at significantly higher levels in tumors of patients that have a MammaPrint high risk profile compared to those having a low risk profile, regardless of whether patients received adjuvant treatment (Fig. 1a; P = 8.5 × 10−7 and P = 4.3 × 10−4). We subsequently analyzed DCK expression in the Oncomine database (www.oncomine.org), a cancer-profiling database containing data from over >10,000 microarrays [21]. Analysis of breast-derived datasets showed that DCK is overexpressed in ductal breast carcinoma in comparison with normal breast tissue (Fig. 1b; P = 3.09 × 10−6) [22]. To confirm these findings, we analyzed DCK mRNA levels in a primary breast epithelial cell line (MCF10A) and various breast cancer carcinoma cell lines (HCC1954, BT-474, MDA-MB-231, SK-BR-3, MCF7, and T47D) by QRT-PCR. Consistent with the Oncomine data, cell lines derived from malignant breast cancer lesions displayed a clear induction of DCK expression (Fig. 1c).
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-011-1477-3/MediaObjects/10549_2011_1477_Fig1_HTML.gif
Fig. 1

DCK is overexpressed in poor outcome breast cancer. a Analysis of DCK mRNA expression in MammaPrint high risk patients (n = 180) and MammaPrint low risk patients (n = 115) showing that DCK is overexpressed in MammaPrint high risk patients. b Analysis of DCK mRNA expression in ductal breast carcinoma (n = 40) compared to normal breast (n = 7) documents upregulation of DCK in ductal breast carcinoma. c Analyis of relative DCK levels by QRT-PCR in a panel of breast epithelial and breast carcinoma cell lines showing that DCK is upregulated in breast cancer cell lines

Deoxycytidine kinase expression predicts sensitivity to nucleoside analogs in untransformed breast epithelial cells

As DCK is essential for the pharmacological activation of 5-azadC and dFdC, we hypothesized that DCK expression is a causal factor in determining responsiveness of breast cancer cell lines to 5-azadC and dFdC. To test this, we retrovirally transduced MCF10A cells that have relatively low levels of DCK expression (Fig. 1c) with two different shRNA knockdown vectors targeting DCK (pRS-DCK.2 and pRS-DCK.4) or a DCK expression vector (pQXCIP-DCK). We used a shRNA vector targeting green fluorescent protein (pRS-GFP) or a GFP expression vector (pQXCIP-GFP) as negative controls throughout this study. First, we confirmed that the two independent DCK shRNA vectors downregulated DCK mRNA levels and that DCK were expressed at higher levels in cells transduced with pQXCIP-DCK (Fig. 2a). Next, we determined sensitivity of DCK knockdown cells to dFdC. The two individual shRNA vectors targeting DCK conferred potent resistance to dFdC in long-term colony formation assays (Fig. 2b). We could recapitulate these findings by performing 3T3 proliferation assays of MCF10A expressing pRS-GFP, pRS-DCK.2, or pRS-DCK.4 in absence or presence of dFdC (Fig. 2b). Moreover, DCK knockdown cells were much less sensitive to 5-azadC (Fig. 2b). Importantly, the knockdown efficiency of the two vectors was directly correlated to their ability to induce resistance to dFdC and 5-azadC (Fig. 2a, b). Conversely, we found that over-expression of DCK sensitized MCF10A cells to both dFdC and 5-azadC in long-term colony formation and proliferation assays (Fig. 2c). Collectively, these data show that DCK expression crisply predicts responsiveness to multiple nucleoside analogs in MCF10A cells, consistent with a direct and causal role for this enzyme in activating these produgs.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-011-1477-3/MediaObjects/10549_2011_1477_Fig2_HTML.gif
Fig. 2

DCK expression predicts sensitivity to nucleoside analogs in MCF10A cells. a MCF10A cells were transduced with shRNAs targeting GFP or DCK (#2 and #4). The knockdown abilties of the shRNAs were determined by examing relative DCK mRNA by QRT-PCR (left panel). Error bars denote standard deviation (SD). MCF10A cells were transduced with plasmids overexpressing GFP or DCK. Overexpression was verified by analyzing DCK and GFP protein expression by western blotting (right panel), α-tubulin serves as a loading control. b The shRNA transduced MCF10A cells were seeded for a colony formation assay in the absence of drug or indicated concentrations of dFdC. Cells were fixed, stained, and scanned after 14 days (left panel). Alternatively, MCF10A cells expressing shRNAs targeting GFP or DCK were seeded for proliferation assays according to the 3T3 protocol in absence of drug, presence of dFdC (10 nM) or 5-azadC (500 nM) (right panel). Error bars denote SD of triplicate independent experiments. c MCF10A cells, transduced with plasmids overexpressing GFP or DCK, were seeded for a colony formation assay (left panel) or examined by proliferation curves (right panel) using the same conditions as in b. Error bars denote SD of triplicate independent experiments

Deoxycytidine kinase expression predicts sensitivity to nucleoside analogs in malignant breast cancer cells

Next, we assessed whether modulation of DCK levels in HCC1954 breast cancer cells that express higher levels of DCK can also alter sensitivity to 5-azadC and dFdC. We thus generated both DCK knockdown and DCK over-expressing HCC1954 cells (Fig. 3a). We subsequently examined sensitivity of DCK knockdown and DCK over-expressing HCC1954 cells to both dFdC and 5-azadC. Similar to MCF10A cells, the two individual shRNA vectors targeting DCK induced potent resistance to dFdC and 5-azadC in both long-term colony formation assays and proliferation assays (Fig. 3b). Conversely, over-expression of DCK sensitized HCC1954 cells to both dFdC and 5-azadC (Fig. 3c). However, the hypersensitivity to nucleoside analogs by overexpression of DCK in HCC1954 cells was not as pronounced as in MCF10A cells, suggesting that DCK enzyme levels are less rate-limiting in HCC1954 cells for prodrug conversion as compared to MCF10A cells (see Fig. 1c). In summary, these findings demonstrate that modulation of DCK levels can critically affect responsiveness to nucleoside analogs in both primary breast epithelial cells and malignant breast carcinoma cells.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-011-1477-3/MediaObjects/10549_2011_1477_Fig3_HTML.gif
Fig. 3

DCK expression predicts sensitivity to nucleoside analogs in HCC1954 cells. a HCC1954 cells were transduced with shRNAs targeting GFP or DCK (#2 and #4). The knockdown abilties of the shRNAs were assessed by examing relative DCK mRNA by QRT-PCR (left panel). Error bars denote standard deviation (SD). HCC1954 cells were transduced with plasmids overexpressing GFP or DCK. Overexpression was verified by analyzing DCK and GFP protein expression by western blotting (right panel), α-tubulin serves as a loading control. b The shRNA transduced HCC1954 cells were seeded for a colony formation assay in the absence of drug or indicated concentrations of dFdC or 5-azadC. Cells were fixed, stained, and scanned after 14 days (left panel). Alternatively, HCC1954 cells expressing shRNAs targeting GFP or DCK were seeded for proliferation assays according to the 3T3 protocol in absence of drug or presence of dFdC (5 nM) or 5-azadC (250 nM) (right panel). Error bars denote SD of triplicate independent experiments. c HCC1954 cells, transduced with plasmids overexpressing GFP or DCK, were seeded for a colony formation assay (left panel) or examined by proliferation curves (right panel) using the same conditions as in b. Error bars denote SD of triplicate independent experiments

Deoxycytidine kinase expression predicts sensitivity to nucleoside analogs in both breast cancer cells and lung carcinoma cells

We then investigated whether the relation between DCK expression and responsiveness to nucleoside analogs is also applicable to other breast cancer cell lines. We thus knocked down DCK in MCF7 breast carcinoma cells (Fig. 4a). Similar to MCF10A and HCC1954 cells, the two individual shRNA vectors targeting DCK induced potent resistance to dFdC and 5-azadC but also AraC in long-term colony formation assays and proliferation assays (Fig. 4b).
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-011-1477-3/MediaObjects/10549_2011_1477_Fig4_HTML.gif
Fig. 4

DCK expression predicts sensitivity to nucleoside analogs in MCF7 and A549 cells. a MCF7 and A549 cells were transduced with shRNAs targeting GFP or DCK (#2 and #4). The knockdown abilties of the shRNAs were assessed by examing relative DCK mRNA by QRT-PCR. Error bars denote standard deviation (SD). b The shRNA transduced MCF7 cells were seeded for a colony formation assay in the absence of drug or indicated concentrations of dFdC, 5-azadC, or AraC. Cells were fixed, stained, and scanned after 14 days (left panel). Alternatively, MCF7 cells expressing shRNAs targeting GFP or DCK were seeded for proliferation assays according to the 3T3 protocol in absence of drug or presence of dFdC (10 nM) (right panel). Error bars denote SD of triplicate independent experiments. c The shRNA transduced A549 cells were seeded for a colony formation assay in the absence of drug or indicated concentrations of dFdC, 5-azadC, or AraC. Cells were fixed, stained, and scanned after 14 days (left panel). Alternatively, A549 cells expressing shRNAs targeting GFP or DCK were seeded for proliferation assays according to the 3T3 protocol in absence of drug or presence of dFdC (10 nM) (right panel). Error bars denote SD of triplicate independent experiments

At present, the use of dFdC is only considered when patients with metastatic breast cancer relapse from treatment with anthracyclins. However, in lung cancer, dFdC is commonly used in combination with carboplatin as a first line treatment. We, therefore, also examined whether DCK knockdown provides resistance to dFdC in lung carcinoma cells. We transduced A549 lung carcinoma cells with the shRNA vectors targeting GFP and DCK and assessed sensitivity to dFdC, to 5-azadC, and to AraC (Fig. 4a, c). Similar to breast epithelial and breast carcinoma cell lines, knockdown of DCK in A549 cells resulted in marked resistance to multiple nucleoside analogs (Fig. 4c). In conclusion, these data clearly demonstrate that DCK levels critically determine sensitivity of multiple cancer cell lines to nucleoside analogs.

Discussion

This study was inspired by the observation that DCK is one of the 70 genes in the MammaPrint breast cancer prognosis signature, which can foretell disease outcome in early stage breast cancer [12, 23]. In this study, we found that DCK expression is significantly higher in breast tumors having a high risk MammaPrint prognosis profile, as compared to those having the low risk profile. As DCK is essential for pharmacological activation of nucleoside analogs that are frequently used in cancer therapy, we hypothesized that poor outcome patients may benefit from treatment with nucleoside analogs. In support of this, we demonstrate that DCK levels determine responsiveness to nucleoside analogs in multiple breast and lung cancer cell lines. Collectively, these findings suggest that the heterogeneity in DCK expression in breast cancer can be exploited to guide effective treatment with nucleoside analogs.

DCK has previously been found to be important for mediating sensitivity to nucleoside analogs in leukemic, pancreatic, lung, colorectal, and glioma cell lines. Loss of DCK expression by de novo mutations or gene knockdown has been reported to confer resistance to dFdC, AraC, or 5-azadC [13, 15, 17, 2426]. Conversely, DCK over-expression can induce hypersensitivity to these nucleoside analogs [14, 16, 17, 2729]. However, the relation of DCK with responsiveness to nucleoside analogs has not been explored in breast cancer. We provide mechanistic evidence for treatment of breast cancer patients with nucleoside analogs by demonstrating that the strong correlation between DCK levels and sensitivity to nucleoside analogs is also applicable to breast cancer cell lines. The data thus suggests a novel adjuvant treatment option for poor outcome breast cancer patients for which treatment guidelines currently do not encourage the use of nucleoside analogs.

Patients with pancreatic cancer usually receive adjuvant dFdC therapy after surgical resection. Remarkably, and in contrast with the findings in breast cancer, low levels of DCK protein expression were found to correlate significantly with poor clinical outcome in pancreatic cancer [30]. In breast cancer, mitotic index is a significant prognostic factor in predicting disease outcome [31]. The finding that DCK, an enzyme in the nucleoside biosynthetic pathway, is expressed at higher levels in poor outcome breast cancer patients, and is consistent with this notion. In contrast to the reported inverse correlation between DCK expression and overall survival in pancreatic cancer, but in support of our in vitro data in breast cancer, Sebastiani et al. and another independent study showed that low levels of DCK protein expression correlate significantly with poor clinical outcome in patients with resected pancreatic cancer who received dFdC as adjuvant therapy [30, 32]. Moreover, a recent study reported that the quantitative analysis of DCK mRNA using FFPE tissue samples can be used to predict the sensitivity to gemcitabine-based adjuvant therapy in patients with pancreatic cancer. Collectively, these studies not only support the utility of DCK expression as a biomarker to predict responsiveness to nucleoside analogs but also suggest that both DCK protein levels and DCK mRNA expression may serve as biomarkers to predict nucleoside analog sensitivity [33]. We are currently setting up a retrospective study in which we will correlate DCK expression of dFdC-treated breast cancer patients to clinical outcome to further study the utility of DCK expression for individualized cancer therapy.

Acknowledgments

We thank Lucas Bruurs for technical assistance. The work of the authors was supported by the SPINOZA grant from the Netherlands Organization for Scientific Research (NWO).

Conflict of Interest

RB and ST are employees of Agendia BV, the company that markets the MammaPrint prognosis test for breast cancer.

Copyright information

© Springer Science+Business Media, LLC. 2011