Article Text
PDF
Abstract
Background Chemotherapy benefits relatively few patients with cutaneous melanoma. The assessment of tumour chemosensitivity by the ATP-based tumour chemosensitivity assay (ATP-TCA) has shown strong correlation with outcome in cutaneous melanoma, but requires fresh tissue and dedicated laboratory facilities.
Aim To examine whether the results of the ATP-TCA correlate with the expression of genes known to be involved in resistance to chemotherapy, based on the hypothesis that the molecular basis of chemosensitivity lies within known drug resistance mechanisms.
Method The chemosensitivity of 47 cutaneous melanomas was assessed using the ATP-TCA and correlated with quantitative expression of 93 resistance genes measured by quantitative reverse transcriptase PCR (qRT-PCR) in a Taqman Array after extraction of total RNA from formalin-fixed paraffin-embedded tissue.
Results Drugs susceptible to particular resistance mechanisms showed good correlation with genes linked to these mechanisms using signatures of up to 17 genes. Comparison of these signatures for DTIC, treosulfan and cisplatin showed several genes in common. HSP70, at least one human epidermal growth factor receptor, genes involved in apoptosis (IAP2, PTEN) and DNA repair (ERCC1, XPA, XRCC1, XRCC6) were present for these agents, as well as genes involved in the regulation of proliferation (Ki67, p21, p27). The combinations tested included genes represented in the single agent signatures.
Conclusions These data suggest that melanoma chemosensitivity is influenced by known resistance mechanisms, including susceptibility to apoptosis. Use of a candidate gene approach may increase understanding of the mechanisms underlying chemosensitivity to drugs active against melanoma and provide signatures with predictive value.
- Melanoma
- ATP
- chemosensitivity
- gene expression
- RT-PCR
- chemotherapy
- PCR
Statistics from Altmetric.com
Introduction
Chemotherapy provides limited benefit to patients with cutaneous melanoma, and, in the UK, most patients are offered a single agent, darcarbazine (DTIC), which remains the standard against which other drugs are judged.1 Other active agents include platinum, particularly in combination with paclitaxel,2 vinca alkaloids,3 4 and treosulfan, usually combined with gemcitabine.5–7 The finding that some patients respond to second-line agents suggested a degree of heterogeneity of chemosensitivity between melanoma patients, confirmed in primary cell cultures using an ATP-based tumour chemosensitivity assay (ATP-TCA).8 Subsequent studies9 confirmed this, and were the basis for a phase II clinical trial conducted by DeCOG in Germany,10 which showed an objective response in 36% of chemosensitive patients compared with 16% in chemoresistant patients, and a clinical benefit rate (including stable disease) of 59% in chemosensitive patients versus 23% in chemoresistant patients. Overall survival was significantly improved in chemosensitive patients (14.6 months) compared with chemoresistant patients (7.4 months), justifying a phase III trial, which is in progress.
The clinical use of such cellular assays is problematic, as accessible tumour tissue is not available in the quantity required. Molecular assays offer an alternative capable of wider use. While studies of single genes rarely show predictive efficacy unless they happen to be the targets of the drugs concerned, multigene signatures have shown greater promise. There are two methods available to generate such signatures. One uses hybridisation arrays to screen the entire genome, and statistical methods to generate predictive gene signatures.11 12 The other has the advantage of being hypothesis-driven and requires accurate measurement of fewer genes, based on the known mechanisms of resistance or sensitivity to the drugs of interest. We have recently published evidence that this approach correlates well with the sensitivity of lung cancer to chemotherapeutic agents in the ATP-TCA,13 and now extend this proof of principle to melanoma, a very different tumour type.
The gene set used in this study was based on published papers and knowledge of the pathways involved in resistance and sensitivity to chemotherapy. The mechanisms of resistance to the single agents used in melanoma have been studied, although there are many gaps. In addition, while it is generally assumed that the same mechanisms used by cells to resist the activity of single agents will also apply to combinations of those agents, there are few studies to support this assumption. Furthermore, most studies of chemoresistance in melanoma have used cell lines that may not reflect the situation in tumours well, as we have previously reported in ovarian cancer14 and found in melanoma (Fernando A., Cree I.A unpublished data). The mechanisms of cellular resistance to chemotherapy include downregulation of target expression, drug metabolism, membrane-located xenobiotic pumps, altered susceptibility to apoptosis, and altered growth/cell cycle or differentiation.15 16 For this study, we designed a Taqman Array microfluidic quantitative reverse transcriptase (qRT)-PCR card to include 93 genes known to be involved in drug resistance/sensitivity to chemotherapeutic agents, including those used in melanoma: DTIC, alkylating agents, platinum, vinca alkaloids and taxanes. The card includes five housekeeping genes to allow standardisation of the results for comparison of individual tumour data. The gene set chosen is not comprehensive, but was designed to establish proof of principle for the use of chemosensitivity data with gene expression data to determine the contribution of individual genes to drug sensitivity and resistance. This study tested the hypothesis that the molecular basis of the observed heterogeneity of chemosensitivity of melanoma is determined by the known resistance mechanisms expressed by these patients' tumours.
Methods
qRT-PCR was used to examine the expression of genes previously shown to be involved in the resistance of melanoma to chemotherapy, as previously published.13 The expression profiles have been compared with quantitative chemosensitivity data obtained for the same tumours using the ATP-TCA.
Patients and samples
A series of 47 metastatic melanoma samples were obtained from surgical specimens, with written consent according to the Declaration of Helsinki with ethics committee approval. The fresh samples were used to obtain in vitro sensitivity data from the ATP-TCA, and all patients had formalin-fixed paraffin-embedded (FFPE) material taken for histology, providing a source of material for qRT-PCR. Patient ages ranged from 28 to 89 years (median 68).
ATP-TCA
The ATP-TCA was performed as previously published.17 18 Samples were sent to the laboratory in cooled transport medium (Dulbecco's modified Eagle's medium; Sigma, Poole, Dorset, UK) with added antibiotics. Melanoma cells were obtained by enzymatic dissociation, washed in serum-free complete assay medium (available from DCS, Hamburg, Germany), and purified by density centrifugation with Hypaque (Sigma). The cells were plated in 96-well polypropylene plates at 20 000 cells per well with six dilutions of four drugs or combinations (table 1), tested in triplicate. Controls in each plate were one row with medium only and one with a maximum inhibitor. Each plate was incubated for 6 days at 37°C with 5% CO2, after which ATP was extracted with tumour cell extraction reagent (DCS). ATP levels were measured by luciferin–luciferase assay, and luminescence was read in a microplate luminometer (MPLX; Berthold Diagnostic Systems, Hamburg, Germany). The results were expressed as the percentage inhibition at each concentration tested, and a summary index was calculated as the sum of the surviving cell fraction at each concentration calculated as 600 – Sum(Inhibition600:Inhibition6.25) for comparison with qRT-PCR results, where IndexSUM = 0 indicates complete inhibition and IndexSUM = 600 indicates no effect.19
Concentrations and suppliers for each drug used in the study, together with the number of tumours tested with each drug
Extraction of mRNA from FFPE melanoma tissue
H&E-stained sections of selected FFPE blocks of metastatic melanoma were marked by a pathologist to identify areas of interest, from which punches were taken using a manual tissue arrayer (MTA1; Beecher Instruments Inc, Sun Prairie, WI, USA) fitted with a 0.6 mm diameter punch stylet. The stylet was decontaminated (Ambion DNA & RNA Zap, Applied Biosystems, Foster City, CA, USA) and cleaned (70% alcohol) between each FFPE block. Two punches were obtained from each block and placed in a sterile labelled 1.5 ml microcentrifuge tube for extraction. We preferred this to sections, as it avoids excessive wax being present in the early part of the extraction process.
RNA extraction was performed as previously published in detail.13 In brief, tubes containing the FFPE punches were heated at 70°C in a Stuart SBH200D heating block for 20 min. Xylene was added to the tube for 3 min at 50°C. The Microfuge tubes were then centrifuged at 13 000 ×g for 2 min in a Sanyo MSE Microcentaur centrifuge, and waste xylene removed with a fine-tipped pipette. Subsequently 1.0 ml 100% ethanol was added to the tubes at room temperature before centrifugation at 12 000 ×g for 2 min. The ethanol was removed by pipette, and the process repeated. The Microfuge tube lids were opened to allow evaporation of any residual ethanol at 50°C for 15 min, before protease digestion.
Protease digestion and RNA extraction were performed by the use of an Ambion Recoverall kit (Applied Biosystems, AM1975) according to the manufacturer's instructions, as previously described.13 The lysates resulting from protease digestion were stored at −20°C before RNA extraction.
Two-step qRT-PCR
Reverse transcription was performed in 0.2 ml PCR tubes using an ABI High-Capacity cDNA Archive Kit (4322171) according to the manufacturer's instructions and as previously published.13 In the RT mix, the final RNA concentration was 1–20 ng/μl. Cycling conditions were: step 1, 25°C for 10 min; step 2, 37°C for 120 min. The tubes were pulse spun in a Microfuge at 12 000 ×g for 30 s after removal from the thermal cycler and stored overnight at +4°C. cDNA content was measured using a NanoDrop spectrophotometer before use in the Taqman Array.
Taqman Arrays (Applied Biosystems Inc) were run according to the manufacturer's instructions.13 Taqman x2 Universal Master Mix with UNG Amperase (ABI, 4364338) was mixed with an equal volume of cDNA to give a final concentration of 300 ng/μl DNA. Each sample was pipetted into two ports (100 μl per port) of the 384-well Taqman Array card, for the 96 genes arrayed in a Chemosensitivity Gene Expression Array (CGEA-1; CanTech Ltd) and spun twice for 60 s at 380 ×g. The card was then placed in a Taqman Array slide sealer for sealing, and the loading ports cut from the card before it was read in an AB 7900HT thermal cycler. PCR was performed for 90 min with the following conditions: AmpErase UNG Activation for 2 min at 50°C; AmpliTaq Gold DNA Polymerase Activation for 10 min at 94.5°C; followed by 40 cycles each of Melt Anneal/Extend for 30 s at 97°C and 1 min at 59.7°C. The ‘Auto Threshold Cycle’ function was performed at the end of the run, and Ct data transferred to a Microsoft Excel spreadsheet, controls checked, and the data transferred to a Microsoft Access database for further analysis. The list of genes present on each card is given in table 2.
List of genes present on each card, ordered by main mechanism of action
Ct values were standardised against HMBS (PBGD), the least variable housekeeping gene, to avoid errors due to differences in efficiency between the housekeeping and test genes. A logarithmic gene expression ratio was calculated as Ln(2-Ct[test]/2-Ct[PBGD]) and used for comparison with ATP-TCA data by multiple linear regression using SPSS V.14.0 and Analyse-It software.
Data analysis
The qRT-PCR data were compared with ATP-TCA results for individual drugs or combinations by multiple linear regression with forward selection of variables using SPSS V.15.0. For each variable, inclusion was dependent on a probability of F > 0.1—that is, the threshold for inclusion of a gene in the model within SPSS, based on initial assessment of the most appropriate model size using the Akaike Information Criterion, a function of model error and size that penalises large models (data not shown). The PRESS (prediction residual sum of squares) method was used, as this is an adjusted regression method helpful in preventing overfitting of the data, as it uses a ‘leave one out’ method of analysis. Genes were added by forward regression according to their univariate correlations following entry of each gene into the model, and no intercept term was included.
Results
ATP-based tumour chemosensitivity assay
In keeping with previous papers,8 9 there was considerable heterogeneity of ATP-TCA results for each of the drugs and combinations tested. The box plots for each drug tested (figure 1) show the greatest activity (ie, lowest IndexSUM) for treosulfan + gemcitabine, while other options showed weaker but more variable activity. Not all drugs were tested in each tumour, because of constraints imposed by the number of cells available (table 1).
Box plots showing the heterogeneity of chemosensitivity in the ATP-based tumour chemosensitivity assay for each drug used in the study. A low index indicates greater activity in the assay. The graph suggests considerable resistance to DTIC and cisplatin, while T+G shows greatest activity.
Quantitative reverse transcriptase PCR
All samples were regarded as evaluable on the basis of the housekeeping gene Ct levels (ie, PBGD Ct >37 cycles), despite extraction of RNA from FFPE tissue up to 7 years old. Variation in housekeeping gene levels was limited, with a normal distribution of Ct levels (PBGD, mean=32, range 28–37). The oldest blocks used were from 2001, and the youngest from 2008. Comparison of the PBGD housekeeping gene for 2001–2003 and 2005–2008 showed a reduction of 1.6 cycles in Ct values, suggesting good survival of RNA within older blocks. For most genes studied, Ct levels were within the detectable range, although some were more rarely expressed than others. NTC remained undetectable throughout the study, showing that the reagents and cards in use had not been contaminated. Control data showed an intra-assay coefficient of variation for one sample of 4.8% for PBGD.
Correlation of mechanisms with ATP-TCA data
Drugs susceptible to particular resistance mechanisms showed good correlation with genes linked to these mechanisms (figure 2), provided that the drugs were sufficiently active and showed heterogeneity of chemosensitivity. As all of the genes on the Taqman Arrays had been previously related to drug resistance or sensitivity, this cannot therefore be regarded as a naïve dataset, but the SPSS analysis included all genes present on the card. The genes involved in each of the forward multiple regression models were identified, and the coefficients for each are shown in table 3 in the order of greatest contribution to the model. Strong correlation was observed with all drugs tested, apart from vinorelbine in which the adjusted R2 correlation coefficient was 0.316, giving a SE of estimated IndexSUM of 154 points.
Correlation of ATP-based tumour chemosensitivity assay IndexSUM data with predicted IndexSUM data (PRESS) from gene expression in the Taqman Arrays. (A) Dacarbazine (DTIC), (B) treosulfan, (C) cisplatin, (D) gemcitabine, (E) paclitaxel, (F) vinorelbine, (G) cisplatin + paclitaxel, (H) treosulfan + gemcitabine.
(a) Summary of models obtained by SPSS for each drug or combination tested. (b) Coefficients for each model, including the genes present in each. SE is reported as predicted chemosensitivity index points
Two alkylating agents were tested (DTIC and tresosulfan) as well as cisplatin, all of which target DNA directly. Comparison of the gene expression signatures for each drug shows several genes in common. HSP70 is present in all three signatures, and all three contain at least one human epidermal growth factor receptor—particularly EGFR (HER1). There were two sets of primers for the APC gene on the array, and APC showed the highest level of entry to both alkylating agent signatures by forward selection. Genes involved in apoptosis (IAP2, PTEN) and DNA repair (ERCC1, XPA, XRCC1, XRCC6) are present for all three DNA-active agents, but in addition there are genes involved in the regulation of proliferation (Ki67, p21, p27). Detoxification and xenobiotic pump molecules are represented by MRP molecules in particular.
The gemcitabine signature also involves a number of pump molecules (MRP again), and proliferation-associated genes, including Ki67 and HER4 as well as p16 and p21. The two microtubule-active agents, vinorelbine and paclitaxel, show different signatures, and that for vinorelbine may not be informative, as the correlation coefficient is relatively low. The paclitaxel signature includes apoptosis-related genes (Bax, IGFBP2, Akt, IGF1 and Fas-L), as well as several genes that may be associated with DNA repair (BRCA1, Rad51, ERCC2). Detoxification and xenobiotic pump molecules are represented by SOD1, MRP2 and TAP4
Two combinations were tested. The cisplatin + paclitaxel signature includes four genes represented in the single-agent signatures, one for both agents (MRP2). Cisplatin activity is also associated with MTII and p21 expression, while paclitaxel is associated with Akt expression. The combination is associated with apoptosis and DNA repair-associated genes, but detoxification/pump and proliferation-associated molecules are also present. The treosulfan + gemcitabine signature includes two genes in the treosulfan signature, APC and HSP70, and three in the gemcitabine signature, GSTπ, DPD and Bad. The majority of the remainder again fall into the apoptosis and DNA repair categories, with some detoxification and pump genes.
Discussion
The availability of chemosensitivity data from the ATP-TCA has enabled us to examine several single-agent and combination effects in the same tumours, something that is impossible in patients. We have compared the chemosensitivity data with the expression of a large number of potential resistance mechanisms using qRT-PCR to study FFPE biopsy material. In contrast with the ATP-TCA, which requires large amounts of fresh tissue, Taqman Arrays can be performed with a few nanograms of RNA extracted from FFPE blocks, and in each case we have used just two 0.6 mm punches from tumour within the blocks. Our success in doing this is certainly in part due to the use of punches rather than sections, which avoids having to deal with too much wax in the early part of the extraction process. Although the amount of RNA obtained was small, the use of Taqman Arrays optimises the use of this material. Fragmentation of RNA remains an issue, but can be obviated by the use of short (<100 bp) product sizes. The comparison of quantitative gene expression data (coefficient of variation 5%) with chemosensitivity data from the ATP-TCA (coefficient of variation 15%) means that relatively small numbers of tumours are needed to obtain data. This could be of benefit for the investigation of mechanisms of sensitivity and resistance, and to develop predictive molecular assays for new drugs entering the clinic. It is of course likely that our array does not include all the genes of relevance to the drugs tested. In some instances, the lack of good correlation may reflect the absence of key genes from the array (possibly unknown as resistance-associated), or general resistance, which results in a lack of heterogeneity in the cellular chemosensitivity assay and prevents good correlation. However, it should be noted that most previous functional research on chemosensitivity has been performed using cell lines that show differences in their chemosensitivity to tumour-derived cells tested in xenografts20 or primary cultures.14 Since manufacture of the Taqman Arrays, several new correlations of gene expression with drugs active in melanoma have been described—for example, the putative role of structural proteins such as gp100.21 Further work is therefore required to define gene sets that might be clinically useful.
To avoid the danger of overfitting the data, we used PRESS statistics, essentially a ‘leave one out’ adjusted regression model. The genes included in each model are involved in the expected mechanisms described in the introduction, but it should be noted that we were looking for predictive markers, and those genes that are upregulated as part of resistance, and are therefore important but not predictive, may not be represented. An example may be MGMT, which is important in resistance to alkylating agents such as DTIC, temozolomide and fotemustine, but the expression of which may not be predictive, as it is upregulated rapidly after drug exposure, as are many genes.22 23 Covariance is also an issue: genes involved in similar mechanisms (eg, drug pumps) may be related in their expression, and which appears in the model may be of little importance.23 Despite these provisos, there is generally good correlation between expression of genes associated with chemosensitivity to chemotherapy and the activity of anti-cancer drugs in the ATP-TCA.
Alkylating agents and platinum target DNA directly, and it is therefore no surprise that DNA repair genes figure prominently in the signatures for these agents, alone and in combination. Perhaps more of a surprise is the presence of these genes in the signatures for drugs such as paclitaxel, although in this instance they may be markers of proliferation rather than DNA repair capability. Cytotoxic agents target growing cells preferentially, and proliferation markers are present in most of the signatures obtained. Equally, melanoma is known to be one of the more chemoresistant tumour types because of its resistance to apoptosis. This has been targeted therapeutically through the use of antisense bcl-2 treatment,24 although with limited success, possibly because of the expression of many anti-apoptotic mechanisms in melanoma cells (eg, Mcl-1, survivin).25 26 Apoptosis-related genes figure prominently in nearly all of the signatures obtained, and may measure the apoptotic potential of the melanoma cells, which is likely to be a major determinant of response. In addition, most signatures include pump and detoxification molecules involved in classical drug resistance.
There are some notable absences of previously suggested resistance mechanisms from the signatures we have obtained. For instance, the relationship of paclitaxel sensitivity to tubulin gene expression is controversial, although recent data in primary melanoma27 suggest that it may be important. This may of course represent a covariance effect, but the lack of correlation of in vitro chemosensitivity with tubulin expression is not surprising given the limited correlation with individual genes observed. The number of melanomas included in this study is relatively small, and it is important not to draw too much from these data—the likelihood of the signatures described here being optimal for prediction of chemosensitivity is small, and much larger studies using clinical samples will be required to develop and validate signatures for clinical use. However, the use of a similar candidate gene approach to improve prediction for new drugs is attractive, and it would be interesting to compare this approach with others to guide the use of tyrosine kinase inhibitors and other drugs targeting BRAF, c-KIT and MEK in melanoma.28 Many of the tyrosine kinase inhibitors kill cells by apoptosis, and some are susceptible to drug pumps or detoxification.28 Combination of mutation or amplification data from DNA assays with gene expression signatures may improve the predictive capacity of both.
Conclusion
This study supports the hypothesis that the molecular basis of the observed difference in sensitivity between melanoma cases lies within the known resistance mechanisms inherent to these patients' tumours, as we have previously described in lung cancer.13 It seems likely that this approach will be able to produce predictive signatures related to clinical outcome, which could be used to optimise therapy for individual patients. Correlative studies of these signatures with clinical outcome, ideally in the context of clinical trials, would be helpful.
Take-home messages
The generation of melanoma primary cell cultures allows assessment of their sensitivity to drugs, which can then be compared with gene expression in paraffin-processed tissue.
The molecular basis of the observed heterogeneity in the sensitivity of melanoma cases to drugs lies within the known resistance mechanisms for those drugs.
Knowledge of resistance mechanisms for anti-melanoma drugs should allow the development of predictive tests applicable to paraffin-processed tissue to guide treatment.
Acknowledgments
We are grateful to Mr Chirag Patel (Applied Biosystems) for assistance with the estimates of model size.
References
Footnotes
Funding The tissue collection was funded in part by the Wessex Cancer Trust, Bellis House, 11 Westwood Road, Southampton SO17 1DL. The ATP-TCA analyses were funded by CanTech Ltd, c/o Translational Oncology Research Centre, Level F - Pathology Centre, Queen Alexandra Hospital, Portsmouth PO6 3LY; the molecular analysis was funded by Applied Biosystems, Foster City, California, USA, and the analysis performed by staff within TORC funded by CanTech Ltd. KP is funded by the Skin Cancer Research Fund, Department of Plastic Surgery, Frenchay Hospital, Bristol BS16 1LE.
Competing interests IAC is a Director of CanTech Ltd, which part funded this project.
Ethics approval This study was conducted with the approval of the North West Research Ethics Committee, NHS, North West Room, 155 Gateway House, Piccadilly, South Manchester M60 7LP, UK.
Provenance and peer review Not commissioned; externally peer reviewed.