Patterns of Resistance and Incomplete Response to Docetaxel by Gene Expression Profiling in Breast Cancer Patients
http://www.100md.com
《临床肿瘤学》
the Breast Center and the Departments of Medicine, Pathology, and Molecular and Cellular Biology, Baylor College of Medicine
the Methodist Hospital, Houston, TX
ABSTRACT
PATIENTS AND METHODS: Core biopsies from 24 patients were obtained before treatment with neoadjuvant docetaxel (four cycles, 100 mg/m2 every 3 weeks), and response was assessed after chemotherapy. After 3 months of neoadjuvant chemotherapy, surgical specimens (n = 13) were obtained, and laser capture microdissection (LCM; n = 8) was performed to enrich for tumor cells. From each core, surgical, and LCM specimen, sufficient total RNA (3 to 6 μg) was extracted for cDNA array analysis using the Affymetrix HgU95-Av2 GeneChip (Affymetrix, Santa Clara, CA).
RESULTS: From the initial core biopsies, differential patterns of expression of 92 genes correlated with docetaxel response (P = .001). However, the molecular patterns of the residual cancers after 3 months of docetaxel treatment were strikingly similar, independent of initial sensitivity or resistance. This relative genetic homogeneity after treatment was observed in both LCM and non-LCM surgical specimens. The residual tumor after treatment in tumors that were initially sensitive indicates selection of a residual and resistant subpopulation of cells. The gene expression pattern was populated by genes involved in cell cycle arrest at G2M (eg, mitotic cyclins and cdc2) and survival pathways involving the mammalian target of rapamycin.
CONCLUSION: A specific and consistent gene expression pattern was found in residual tumors after docetaxel treatment. These profiles provide therapeutic targets that could lead to improved treatment.
INTRODUCTION
Therefore, we set out to identify gene expression patterns in primary breast cancer specimens that might predict suboptimal response and resistance to taxanes. Neoadjuvant chemotherapy (treatment before primary surgery) allows for sampling of the primary tumor both before and after treatment for gene expression analysis and for direct assessment of response to chemotherapy by following changes in tumor size during the first few months of treatment.9,10 Response to neoadjuvant chemotherapy has been shown to be a valid surrogate marker of survival, with significantly better survival in those patients whose tumors completely regress compared with all other responses.9,10 Using high-throughput quantitation of gene expression, we have identified a 92-gene expression pattern in pretreatment breast cancers that seems to correlate with and, thereby, predict excellent clinical response to neoadjuvant docetaxel.11 However, although these profiles have great potential to penetrate the genetic heterogeneity of this disease and help to prioritize different treatment strategies based on their likelihood of success in individual patients, clinical responses may be incomplete. Therefore, we set out to examine the gene expression patterns associated with this incomplete response by determining the gene expression patterns after 3 months of docetaxel therapy in laser capture microdissection (LCM) and non-LCM surgical specimens of residual breast cancers and by comparing the molecular patterns before and after docetaxel exposure to identify gene pathways that could be important in the mechanism of incomplete response to docetaxel.
PATIENTS AND METHODS
After 3 months of docetaxel, all 24 patients underwent primary surgery. The surgical specimens were examined by a pathologist (M.C.G.) who then immediately snap froze a portion of tumor for gene expression analysis. Of these 24 patients, the surgical specimens of eight patients were lost during a disastrous flood, and frozen specimens were not obtained in three patients. Of the 13 remaining surgical specimens, eight had sufficient tumor for LCM (Pixcell IIe; Arcturus Bioengineering, Mountain View, CA). The selection of specimens for gene expression analysis is included in Figure 1. From each of these surgical specimens, 5-μm sections were cut, and invasive breast carcinoma cells were laser microdissected using standard published techniques. RNA extraction was then carried out on the LCM epithelial cancer cells.
RNA Extraction, Amplification, and GeneChip Hybridization
Total RNA was isolated from the frozen core biopsy and the non-LCM and LCM specimens according to protocols recommended by Affymetrix GeneChip (Affymetrix, Santa Clara, CA). Briefly, total RNA was isolated using TRIzol reagent (Invitrogen Corporation, Carlsbad, CA). Samples were subsequently passed over a Qiagen RNeasy column (Qiagen, Valencia, CA) for control of small fragments that have been shown to affect reverse transcriptase reaction and hybridization quality (unpublished data). Each core and surgical specimen yielded 3 to 6 μg of total RNA.
After RNA recovery, double-stranded cDNA was synthesized by a chimeric oligonucleotide with an oligo-dT and a T7 RNA polymerase promoter at a concentration of 100 pmol/μL. Reverse transcription was carried out according to protocols recommended by Affymetrix GeneChip using commercially available buffers and proteins (Invitrogen). Biotin labeling and approximately 250-fold linear amplification followed phenol-chloroform cleanup of the reverse transcription reaction product and was carried out by in vitro transcription (Enzo Biochem, New York, NY) over a reaction time of 8 hours.
each biopsy, 15 μg of labeled cRNA was then hybridized onto an Affymetrix U95Av2 GeneChip following the recommended procedures for prehybridization, hybridization, washing, and staining with streptavidin-phycoerythrin. Antibody amplification was accomplished using a biotin-linked antistreptavidin antibody (Vector Laboratories, Burlingame, CA) with a goat immunoglobulin G–blocking antibody (Sigma, St Louis, MO). A second application of streptavidin-phycoerythrin was used subsequent to additional wash steps. Following automated staining and wash protocols (Affymetrix protocol EukGE-2v4), the arrays were scanned on an Affymetrix GeneChip Scanner (Agilent, Palo Alto, CA) and quantitated (Affymetrix). The Affymetrix U95Av2 GeneChip represents approximately 12,625 genes. The raw, unnormalized data was then analyzed by ArrayAnalyzer (Insightful Corporation, Seattle, WA; http://www.insightful.com) for normalization and expression estimation.
Statistical Analysis
We first analyzed the gene expression patterns correlating with response and de novo resistance to docetaxel from the initial pretreatment core biopsies; this analysis has been published.11 Second, we determined the patterns of gene expression in the non-LCM and LCM surgical specimens after 3 months of docetaxel treatment to identify the molecular pattern after exposure to treatment. Third, we examined changes in expression in tumors that did not regress completely with therapy to determine patterns that might predict for incomplete response to docetaxel.
After scanning and low-level quantitation using MicroArray Suite (Affymetrix), we used ArrayAnalyzer and DNA-Chip Analyzer (http://www.dchip.org/). Low-level analysis (expression estimation) was performed with ArrayAnalyzer,12 because it produces more accurate results for low expressed genes. Quantile normalization method was used. For expression estimation, we used the robust multiarray analysis method that implements a robust linear fitting procedure called median polish.
After estimation of expression values, we exported the data from ArrayAnlyzer to dChip (version 1.3)13 for filtering and class comparisons. We filtered genes to eliminate those with very low expression values in most samples or low variability across all samples. We retained 7,276 genes for further analysis after filtering; for group comparison, we used paired or two-group t tests. In all comparisons, the multiple comparison problem was addressed by estimation of distribution of false discovery rate by permutation of class labels. Only gene lists with a low median false discovery rate of 0% to 2% were used, which ensures a low chance of false discoveries.
With this statistical analysis, we set out to determine the molecular patterns to distinguish whether differences existed in the surgical specimens that were initially sensitive and resistant. We then correlated changes in gene expression patterns before versus after docetaxel exposure with clinical response for clues to the mechanisms of incomplete response to docetaxel.
RESULTS
Of the 24 patients, 11 (46%) were sensitive to docetaxel, and 13 (54%) were resistant. Of the 11 patients with sensitive tumors, five patients (45%) had minimal residual disease (< 10% residual tumor); whereas of 13 patients with resistant tumors, seven patients had residual tumors ≥ 60% of baseline (58%), and three of the 13 women (23%) had residual tumors that were 100% or greater of baseline.
Core and Surgical Biopsies and RNA Yield
Before treatment, six core biopsies were obtained from each primary breast cancer. Two to three core biopsy specimens were immediately snap frozen at –80°C for later gene expression array analysis, and the remaining cores were processed for pathologic evaluation. Each core biopsy measured approximately 1 cm by 1 mm. Because these biopsies were too small for microdissection, we ascertained the tumor cellularity of the pretreatment core biopsies. The core biopsies showed good tumor cellularity, with a median tumor cellularity of 75% (range, 40% to 100%).
After docetaxel treatment, 13 surgical specimens were evaluated for tumor cellularity. The median tumor cellularity of the surgical specimens was lower at 40% (range, 10% to 100%). Because of the lower tumor cellularity in the surgical specimens, we performed LCM to enrich for tumor cells, and this was successfully performed in eight specimens with a median of 500 microdissected invasive cancer cells (range, 300 to 1,500 cells). Each frozen core, surgical, and surgical LCM specimen yielded 3 to 6 μg of total RNA, which was more than sufficient to generate the approximately 20 μg of labeled cRNA needed for hybridization with the Affymetrix HgU95Av2 GeneChip, using the manufacturer's standard protocol.
Discriminatory Genes After Docetaxel Exposure in Non-LCM and LCM Surgical Specimens
We previously published that, from the predocetaxel initial core biopsies, 92 genes were classed as most significantly differentially expressed between responsive and nonresponsive tumors at a nominal P value of less than .001.11 Among these genes are numerous members of cell signaling, immunologic response, DNA damage detection and repair, cell cycle regulation, and tumor-suppressor/oncogene families.
We analyzed gene expression patterns of the posttreatment surgical specimens, selecting a subset of candidate genes by filtering on signal intensity to eliminate genes with uniformly low expression or genes that had expression that did not vary significantly across the samples; subsequent to this filtering, 7,672 genes remained. After log transformation, a t test was used to select discriminatory genes. To evaluate the possibility of spurious results caused by multiple comparisons, we performed a permutation test using only gene lists with low a (0% to 2%) median false discovery rate (genes that may be differentially expressed by chance alone).
Unlike the untreated initial pretreatment biopsies, in which genes discriminating between sensitive and resistant classes were present, t tests on posttreatment specimens with nominal P values of .001, .01, or .05 did not identify any differentially expressed genes that would not have been selected by chance alone, as reflected by the false discovery rate in both the LCM and non-LCM surgical specimens, even at the least stringent P value of .05 (Table 2). In other words, unlike the pretreatment biopsies, we found no statistically significant differentially expressed genes in either LCM or non-LCM surgical specimens, so that the genetic profile of tumors that were initially sensitive or resistant was similar after 3 months of docetaxel.
Changes in Expression Patterns: Non-LCM Surgical Specimens
Non-LCM surgical versus initial pretreatment specimens. We compared patterns of gene expression of tumors before and after docetaxel treatment in 13 of the non-LCM specimens (seven paired sensitive samples and six paired resistant samples; Table 3). In these non-LCM specimens, for tumors that were initially sensitive tumors, pair-wise comparison between expression patterns yielded 118 and 15 genes that were at least two- and three-fold differentially expressed, respectively, at a P value of .01. In the initially resistant tumors, pair-wise comparison between expression patterns yielded 604 and 194 genes that were at least two- and three-fold differentially expressed, respectively, at a P value of .01.
LCM surgical versus initial pretreatment specimens. We then compared patterns of gene expression of tumors before and after docetaxel treatment in LCM specimens (four paired sensitive samples and four paired resistant samples). The results of this supervised clustering are demonstrated in Figure 2 and Table 3.
In the tumors that were initially sensitive, pair-wise comparison between expression patterns yielded 119 and 39 genes that were at least two- and three-fold differentially expressed, respectively, at a P value of .01 as shown in Figure 3. In the initially resistant tumors, pair-wise comparison between expression patterns yielded 251 and 105 genes that were at least two- and three-fold differentially expressed, respectively, at a P value of .01.
Comparison between non-LCM and LCM surgical specimens. The direction of gene expression changes in initial versus LCM and initial versus non-LCM specimens was the same for all significant genes, although not surprisingly, the fold changes were different. In other words, significant differential genes were found to be overexpressed or underexpressed regardless of whether microdissection of the surgical specimens was performed.
Functional Classification of Genes
Pretreatment specimens. From the pretreatment initial samples, the 92 genes that had been classed as most significantly "differentially expressed" showed 4.2- to 2.6-fold decreases or 2.5- to 15.7-fold increases in expression in resistant versus sensitive tumors. Only 14 of the 92 genes were overexpressed in the resistant cluster, with major categories including unknown function, protein translation, cell cycle, and RNA transcription. Beta-tubulin isoforms were also associated with docetaxel resistance. Of the 78 genes overexpressed in docetaxel-sensitive tumors, major categories were stress/apoptosis, adhesion/cytoskeleton (none was overexpressed in resistant tumors), protein transport, signal transduction, and RNA splicing/transport.
Changes in gene expression. Incomplete response to chemotherapy is of major clinical importance. In the tumors that were initially sensitive, changes in gene expression before and after docetaxel treatment may indicate pathways involved in this incomplete response and, hence, reflect a mechanism of inherent resistance to treatment. Looking specifically at changes in gene expression in cancers that were initially sensitive, genes involved in fatty acid/phospholipids metabolism or involved in cell survival involving mammalian target of rapamycin (mTOR) were overexpressed. These pathways have been reported to involve vesicular trafficking, oxidative bursts, and protein and organelle metabolism. Comparing microdissected surgical specimens with the pretreatment biopsies, critical players of the mTOR pathway, such as peroxisome receptor, lipoprotein lipases, and alcohol dehydrogenases, were differentially expressed (Table 4). These observations give preliminary clues to possible mechanisms of incomplete response and resistance to docetaxel. The gene expression pattern before and after docetaxel in tumors that were initially sensitive is shown in Figure 3 (118 genes).
DISCUSSION
the initial biopsies, the resistant tumors overexpressed transcription and signal transduction genes, whereas sensitive tumors had higher expression of genes involved in cytoskeletal/adhesion and protein transport, signal transduction, transcription, cell cycle, and apoptosis. In addition, in sensitive tumors, higher expression of genes involved in stress-related pathways was also found, in particular heat shock proteins.
Breast cancers are highly heterogeneous.14-18 However, we found that after docetaxel treatment, the residual tumors showed similar gene expression patterns, with no significant differential gene expression in the surgical specimens (whether LCM or non-LCM) in tumors that were originally sensitive versus resistant. There are numerous possible explanations for this observation, including clonal selection of a resistance phenotype by treatment, enhanced detection of stromal or even residual normal gene expression patterns, or many other potential causes for the development of a seemingly homogeneous end point from a heterogeneous initial population.14-18
Changes in gene expression patterns associated with docetaxel resistance and residual tumors are highly complex and may give clues to pathways that could be important therapeutic targets. This preliminary data could support a hypothesis that expression of genes involved in the mTOR survival pathway leads to incomplete response to docetaxel. This survival pathway has been reported to involve vesicular trafficking, oxidative bursts, and protein/organelle metabolism. In tumors, mTOR is a survival factor that increases tumor mass through the recruitment of mitogens and nutrients19 and is unusually sensitive to cellular adenosine triphosphate levels, perhaps directly sensing and regulating adenosine triphosphate concentration.19 In addition, phosphatidic acid has been identified as a critical element of mTOR signaling; current models suggest that parallel signaling pathways in translational regulation converge on common effector proteins, which could reveal an important function of this lipid in the regulation of divergent signal transduction and protein synthesis cascades, as well as providing a direct link between mTOR and mitogens, including xenobiotics.20,21 It should be noted that although phosphoinositide 3-kinase and mTOR share roles in the same overall pathway, a linear regulatory relationship between the two has not yet been shown.21 Although mTOR's main role seem to be involved in the assessment of amino acid availability, it may also have a role in the detection of DNA damage.22 Many other members of the growing phosphoinositide kinase kinase-related protein family (which includes ATM, ATR, and DNA-PK) are known to perform roles in DNA damage detection as well as checkpoint control. Inhibitors of the mTOR pathway, including rapamycin, are being tested in phase II clinical studies, including studies in breast cancer patients.
This study shows that expression array technology can effectively begin to penetrate the genetic complexity and heterogeneity of human breast cancers and give pertinent clues behind tumor biology including sensitivity and resistance to therapies. This kind of molecular profiling could have profound clinical applications in designing synergistic combinations of agents and defining the optimal treatment selection for each individual patient. Incomplete responses to docetaxel are frequent. These pilot results suggest that adding mTOR inhibitors might increase the overall efficacy of docetaxel, and if so, increase the overall likelihood of survival and cure in women with breast cancer.
Authors' Disclosures of Potential Conflicts of Interest
NOTES
Supported in part by the US Army Medical Research and Materiel Command grant No. DAMD17-01-0132, a Grant-in-Aid (US 11115) from Aventis Pharmaceutical Inc, the Emma Jacobs Clinical Breast Cancer Fund, and the Breast Cancer Specialized Program of Organized Research Excellence grant No. P50 CA50183 from the National Cancer Institute.
Presented in part at the 39th Annual American Society of Clinical Oncology Meeting, Chicago IL, May 31-June 3, 2003.
J.C.C. and E.C.W. contributed equally to this work.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Early Breast Cancer Trialists' Collaborative Group: Systemic treatment of early breast cancer by hormonal, cytotoxic or immune therapy: 133 randomised trials involving 31,000 recurrences and 24,000 deaths among 75,000 women. Lancet 339:1-15, 1992
2. Early Breast Cancer Trialists' Collaborative Group: Tamoxifen for early breast cancer: An overview of the randomised trials. Lancet 351:1451-1467, 1998
3. Early Breast Cancer Trialists' Collaborative Group: Polychemotherapy for early breast cancer: An overview of the randomised trials. Lancet 352:930-942, 1998
4. Hortobagyi GN: Docetaxel in breast cancer and a rationale for combination therapy. Oncology 11:11-15, 1997
5. Chan S, Friedrichs K, Noel D, et al: Prospective randomized trial of docetaxel versus doxorubicin in patients with metastatic breast cancer: The 303 Study Group. J Clin Oncol 17:2341-2354, 1999
6. Nabholtz JM, et al: A phase III trial comparing docetaxel (T), doxorubicin (A) and cyclophosphamide (C) (TAC) to (FAC) as first line chemotherapy for patients with metastatic breast cancer. Proc Am Soc Clin Oncol 20:22a, 2001 (abstr 83)
7. Henderson IC, Berry DA, Demetri G, et al: Improved outcomes from adding sequential paclitaxel but not from escalating doxorubicin in an adjuvant chemotherapy regimen for patients with node-positive primary breast cancer. J Clin Oncol 21:976-983, 2003
8. Aapro MS: Adjuvant therapy of primary breast cancer: A review of key findings from the 7th international conference, St. Gallen, February 2001. Oncologist 6:376-385, 2001
9. Fisher B, Bryant J, Wolmark N, et al: Effect of preoperative chemotherapy on the outcome of women with operable breast cancer. J Clin Oncol 16:2672-2685, 1998
10. Mamounas EP: Preoperative doxorubicin plus cyclophosphamide followed by preoperative or postoperative docetaxel. Oncology 11:37-40, 1997
11. Chang JC, Wooten EC, Tsimelzon A, et al: Gene expression profiling predicts therapeutic response to docetaxel (Taxotere) in breast cancer patients. Lancet 362:362-369, 2003
12. Irizarry RA, Bolstad BM, Collin F, et al: Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 31:e15, 2003
13. Li C, Wong WH: Model-based analysis of oligonucleotide arrays: Expression index computation and outlier detection. Proc Natl Acad Sci U S A 98:31-36, 2001
14. Schena M, Shalon D, Davis RW, et al: Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270:467-470, 1995
15. Khan J, Simon R, Bittner M, et al: Gene expression profiling of alveolar rhabdomyosarcoma with cDNA microarrays. Cancer Res 58:5009-5013, 1998
16. Lockhart DJ, Dong H, Byrne MC, et al: Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 14:1675-1680, 1996
17. Sgroi DC, Teng S, Robinson G, et al: In vivo gene expression profile analysis of human breast cancer progression. Cancer Res 59:5656-5661, 1999
18. Perou CM, Jeffrey SS, van de Rijn M, et al: Distinctive gene expression patterns in human mammary epithelial cells and breast cancers. Proc Natl Acad Sci U S A 96:9212-9217, 1999
19. Dennis PB, Jaeschke A, Saitoh M, et al: Mammalian TOR: A homeostatic ATP sensor. Science 294:1102-1105, 2001
20. Fang Y, Vilella-Bach M, Bachmann R, et al: Phosphatidic acid-mediated mitogenic activation of mTOR signaling. Science 294:1942-1945, 2001
21. Martin KA, Blenis J: Coordinate regulation of translation by the PI 3-kinase and mTOR pathways. Adv Cancer Res 86:1-39, 2002
22. Tee AR, Proud CG: DNA-damaging agents cause inactivation of translational regulators linked to mTOR signalling. Oncogene 19:3021-3031, 2000(Jenny C. Chang, Eric C. W)
the Methodist Hospital, Houston, TX
ABSTRACT
PATIENTS AND METHODS: Core biopsies from 24 patients were obtained before treatment with neoadjuvant docetaxel (four cycles, 100 mg/m2 every 3 weeks), and response was assessed after chemotherapy. After 3 months of neoadjuvant chemotherapy, surgical specimens (n = 13) were obtained, and laser capture microdissection (LCM; n = 8) was performed to enrich for tumor cells. From each core, surgical, and LCM specimen, sufficient total RNA (3 to 6 μg) was extracted for cDNA array analysis using the Affymetrix HgU95-Av2 GeneChip (Affymetrix, Santa Clara, CA).
RESULTS: From the initial core biopsies, differential patterns of expression of 92 genes correlated with docetaxel response (P = .001). However, the molecular patterns of the residual cancers after 3 months of docetaxel treatment were strikingly similar, independent of initial sensitivity or resistance. This relative genetic homogeneity after treatment was observed in both LCM and non-LCM surgical specimens. The residual tumor after treatment in tumors that were initially sensitive indicates selection of a residual and resistant subpopulation of cells. The gene expression pattern was populated by genes involved in cell cycle arrest at G2M (eg, mitotic cyclins and cdc2) and survival pathways involving the mammalian target of rapamycin.
CONCLUSION: A specific and consistent gene expression pattern was found in residual tumors after docetaxel treatment. These profiles provide therapeutic targets that could lead to improved treatment.
INTRODUCTION
Therefore, we set out to identify gene expression patterns in primary breast cancer specimens that might predict suboptimal response and resistance to taxanes. Neoadjuvant chemotherapy (treatment before primary surgery) allows for sampling of the primary tumor both before and after treatment for gene expression analysis and for direct assessment of response to chemotherapy by following changes in tumor size during the first few months of treatment.9,10 Response to neoadjuvant chemotherapy has been shown to be a valid surrogate marker of survival, with significantly better survival in those patients whose tumors completely regress compared with all other responses.9,10 Using high-throughput quantitation of gene expression, we have identified a 92-gene expression pattern in pretreatment breast cancers that seems to correlate with and, thereby, predict excellent clinical response to neoadjuvant docetaxel.11 However, although these profiles have great potential to penetrate the genetic heterogeneity of this disease and help to prioritize different treatment strategies based on their likelihood of success in individual patients, clinical responses may be incomplete. Therefore, we set out to examine the gene expression patterns associated with this incomplete response by determining the gene expression patterns after 3 months of docetaxel therapy in laser capture microdissection (LCM) and non-LCM surgical specimens of residual breast cancers and by comparing the molecular patterns before and after docetaxel exposure to identify gene pathways that could be important in the mechanism of incomplete response to docetaxel.
PATIENTS AND METHODS
After 3 months of docetaxel, all 24 patients underwent primary surgery. The surgical specimens were examined by a pathologist (M.C.G.) who then immediately snap froze a portion of tumor for gene expression analysis. Of these 24 patients, the surgical specimens of eight patients were lost during a disastrous flood, and frozen specimens were not obtained in three patients. Of the 13 remaining surgical specimens, eight had sufficient tumor for LCM (Pixcell IIe; Arcturus Bioengineering, Mountain View, CA). The selection of specimens for gene expression analysis is included in Figure 1. From each of these surgical specimens, 5-μm sections were cut, and invasive breast carcinoma cells were laser microdissected using standard published techniques. RNA extraction was then carried out on the LCM epithelial cancer cells.
RNA Extraction, Amplification, and GeneChip Hybridization
Total RNA was isolated from the frozen core biopsy and the non-LCM and LCM specimens according to protocols recommended by Affymetrix GeneChip (Affymetrix, Santa Clara, CA). Briefly, total RNA was isolated using TRIzol reagent (Invitrogen Corporation, Carlsbad, CA). Samples were subsequently passed over a Qiagen RNeasy column (Qiagen, Valencia, CA) for control of small fragments that have been shown to affect reverse transcriptase reaction and hybridization quality (unpublished data). Each core and surgical specimen yielded 3 to 6 μg of total RNA.
After RNA recovery, double-stranded cDNA was synthesized by a chimeric oligonucleotide with an oligo-dT and a T7 RNA polymerase promoter at a concentration of 100 pmol/μL. Reverse transcription was carried out according to protocols recommended by Affymetrix GeneChip using commercially available buffers and proteins (Invitrogen). Biotin labeling and approximately 250-fold linear amplification followed phenol-chloroform cleanup of the reverse transcription reaction product and was carried out by in vitro transcription (Enzo Biochem, New York, NY) over a reaction time of 8 hours.
each biopsy, 15 μg of labeled cRNA was then hybridized onto an Affymetrix U95Av2 GeneChip following the recommended procedures for prehybridization, hybridization, washing, and staining with streptavidin-phycoerythrin. Antibody amplification was accomplished using a biotin-linked antistreptavidin antibody (Vector Laboratories, Burlingame, CA) with a goat immunoglobulin G–blocking antibody (Sigma, St Louis, MO). A second application of streptavidin-phycoerythrin was used subsequent to additional wash steps. Following automated staining and wash protocols (Affymetrix protocol EukGE-2v4), the arrays were scanned on an Affymetrix GeneChip Scanner (Agilent, Palo Alto, CA) and quantitated (Affymetrix). The Affymetrix U95Av2 GeneChip represents approximately 12,625 genes. The raw, unnormalized data was then analyzed by ArrayAnalyzer (Insightful Corporation, Seattle, WA; http://www.insightful.com) for normalization and expression estimation.
Statistical Analysis
We first analyzed the gene expression patterns correlating with response and de novo resistance to docetaxel from the initial pretreatment core biopsies; this analysis has been published.11 Second, we determined the patterns of gene expression in the non-LCM and LCM surgical specimens after 3 months of docetaxel treatment to identify the molecular pattern after exposure to treatment. Third, we examined changes in expression in tumors that did not regress completely with therapy to determine patterns that might predict for incomplete response to docetaxel.
After scanning and low-level quantitation using MicroArray Suite (Affymetrix), we used ArrayAnalyzer and DNA-Chip Analyzer (http://www.dchip.org/). Low-level analysis (expression estimation) was performed with ArrayAnalyzer,12 because it produces more accurate results for low expressed genes. Quantile normalization method was used. For expression estimation, we used the robust multiarray analysis method that implements a robust linear fitting procedure called median polish.
After estimation of expression values, we exported the data from ArrayAnlyzer to dChip (version 1.3)13 for filtering and class comparisons. We filtered genes to eliminate those with very low expression values in most samples or low variability across all samples. We retained 7,276 genes for further analysis after filtering; for group comparison, we used paired or two-group t tests. In all comparisons, the multiple comparison problem was addressed by estimation of distribution of false discovery rate by permutation of class labels. Only gene lists with a low median false discovery rate of 0% to 2% were used, which ensures a low chance of false discoveries.
With this statistical analysis, we set out to determine the molecular patterns to distinguish whether differences existed in the surgical specimens that were initially sensitive and resistant. We then correlated changes in gene expression patterns before versus after docetaxel exposure with clinical response for clues to the mechanisms of incomplete response to docetaxel.
RESULTS
Of the 24 patients, 11 (46%) were sensitive to docetaxel, and 13 (54%) were resistant. Of the 11 patients with sensitive tumors, five patients (45%) had minimal residual disease (< 10% residual tumor); whereas of 13 patients with resistant tumors, seven patients had residual tumors ≥ 60% of baseline (58%), and three of the 13 women (23%) had residual tumors that were 100% or greater of baseline.
Core and Surgical Biopsies and RNA Yield
Before treatment, six core biopsies were obtained from each primary breast cancer. Two to three core biopsy specimens were immediately snap frozen at –80°C for later gene expression array analysis, and the remaining cores were processed for pathologic evaluation. Each core biopsy measured approximately 1 cm by 1 mm. Because these biopsies were too small for microdissection, we ascertained the tumor cellularity of the pretreatment core biopsies. The core biopsies showed good tumor cellularity, with a median tumor cellularity of 75% (range, 40% to 100%).
After docetaxel treatment, 13 surgical specimens were evaluated for tumor cellularity. The median tumor cellularity of the surgical specimens was lower at 40% (range, 10% to 100%). Because of the lower tumor cellularity in the surgical specimens, we performed LCM to enrich for tumor cells, and this was successfully performed in eight specimens with a median of 500 microdissected invasive cancer cells (range, 300 to 1,500 cells). Each frozen core, surgical, and surgical LCM specimen yielded 3 to 6 μg of total RNA, which was more than sufficient to generate the approximately 20 μg of labeled cRNA needed for hybridization with the Affymetrix HgU95Av2 GeneChip, using the manufacturer's standard protocol.
Discriminatory Genes After Docetaxel Exposure in Non-LCM and LCM Surgical Specimens
We previously published that, from the predocetaxel initial core biopsies, 92 genes were classed as most significantly differentially expressed between responsive and nonresponsive tumors at a nominal P value of less than .001.11 Among these genes are numerous members of cell signaling, immunologic response, DNA damage detection and repair, cell cycle regulation, and tumor-suppressor/oncogene families.
We analyzed gene expression patterns of the posttreatment surgical specimens, selecting a subset of candidate genes by filtering on signal intensity to eliminate genes with uniformly low expression or genes that had expression that did not vary significantly across the samples; subsequent to this filtering, 7,672 genes remained. After log transformation, a t test was used to select discriminatory genes. To evaluate the possibility of spurious results caused by multiple comparisons, we performed a permutation test using only gene lists with low a (0% to 2%) median false discovery rate (genes that may be differentially expressed by chance alone).
Unlike the untreated initial pretreatment biopsies, in which genes discriminating between sensitive and resistant classes were present, t tests on posttreatment specimens with nominal P values of .001, .01, or .05 did not identify any differentially expressed genes that would not have been selected by chance alone, as reflected by the false discovery rate in both the LCM and non-LCM surgical specimens, even at the least stringent P value of .05 (Table 2). In other words, unlike the pretreatment biopsies, we found no statistically significant differentially expressed genes in either LCM or non-LCM surgical specimens, so that the genetic profile of tumors that were initially sensitive or resistant was similar after 3 months of docetaxel.
Changes in Expression Patterns: Non-LCM Surgical Specimens
Non-LCM surgical versus initial pretreatment specimens. We compared patterns of gene expression of tumors before and after docetaxel treatment in 13 of the non-LCM specimens (seven paired sensitive samples and six paired resistant samples; Table 3). In these non-LCM specimens, for tumors that were initially sensitive tumors, pair-wise comparison between expression patterns yielded 118 and 15 genes that were at least two- and three-fold differentially expressed, respectively, at a P value of .01. In the initially resistant tumors, pair-wise comparison between expression patterns yielded 604 and 194 genes that were at least two- and three-fold differentially expressed, respectively, at a P value of .01.
LCM surgical versus initial pretreatment specimens. We then compared patterns of gene expression of tumors before and after docetaxel treatment in LCM specimens (four paired sensitive samples and four paired resistant samples). The results of this supervised clustering are demonstrated in Figure 2 and Table 3.
In the tumors that were initially sensitive, pair-wise comparison between expression patterns yielded 119 and 39 genes that were at least two- and three-fold differentially expressed, respectively, at a P value of .01 as shown in Figure 3. In the initially resistant tumors, pair-wise comparison between expression patterns yielded 251 and 105 genes that were at least two- and three-fold differentially expressed, respectively, at a P value of .01.
Comparison between non-LCM and LCM surgical specimens. The direction of gene expression changes in initial versus LCM and initial versus non-LCM specimens was the same for all significant genes, although not surprisingly, the fold changes were different. In other words, significant differential genes were found to be overexpressed or underexpressed regardless of whether microdissection of the surgical specimens was performed.
Functional Classification of Genes
Pretreatment specimens. From the pretreatment initial samples, the 92 genes that had been classed as most significantly "differentially expressed" showed 4.2- to 2.6-fold decreases or 2.5- to 15.7-fold increases in expression in resistant versus sensitive tumors. Only 14 of the 92 genes were overexpressed in the resistant cluster, with major categories including unknown function, protein translation, cell cycle, and RNA transcription. Beta-tubulin isoforms were also associated with docetaxel resistance. Of the 78 genes overexpressed in docetaxel-sensitive tumors, major categories were stress/apoptosis, adhesion/cytoskeleton (none was overexpressed in resistant tumors), protein transport, signal transduction, and RNA splicing/transport.
Changes in gene expression. Incomplete response to chemotherapy is of major clinical importance. In the tumors that were initially sensitive, changes in gene expression before and after docetaxel treatment may indicate pathways involved in this incomplete response and, hence, reflect a mechanism of inherent resistance to treatment. Looking specifically at changes in gene expression in cancers that were initially sensitive, genes involved in fatty acid/phospholipids metabolism or involved in cell survival involving mammalian target of rapamycin (mTOR) were overexpressed. These pathways have been reported to involve vesicular trafficking, oxidative bursts, and protein and organelle metabolism. Comparing microdissected surgical specimens with the pretreatment biopsies, critical players of the mTOR pathway, such as peroxisome receptor, lipoprotein lipases, and alcohol dehydrogenases, were differentially expressed (Table 4). These observations give preliminary clues to possible mechanisms of incomplete response and resistance to docetaxel. The gene expression pattern before and after docetaxel in tumors that were initially sensitive is shown in Figure 3 (118 genes).
DISCUSSION
the initial biopsies, the resistant tumors overexpressed transcription and signal transduction genes, whereas sensitive tumors had higher expression of genes involved in cytoskeletal/adhesion and protein transport, signal transduction, transcription, cell cycle, and apoptosis. In addition, in sensitive tumors, higher expression of genes involved in stress-related pathways was also found, in particular heat shock proteins.
Breast cancers are highly heterogeneous.14-18 However, we found that after docetaxel treatment, the residual tumors showed similar gene expression patterns, with no significant differential gene expression in the surgical specimens (whether LCM or non-LCM) in tumors that were originally sensitive versus resistant. There are numerous possible explanations for this observation, including clonal selection of a resistance phenotype by treatment, enhanced detection of stromal or even residual normal gene expression patterns, or many other potential causes for the development of a seemingly homogeneous end point from a heterogeneous initial population.14-18
Changes in gene expression patterns associated with docetaxel resistance and residual tumors are highly complex and may give clues to pathways that could be important therapeutic targets. This preliminary data could support a hypothesis that expression of genes involved in the mTOR survival pathway leads to incomplete response to docetaxel. This survival pathway has been reported to involve vesicular trafficking, oxidative bursts, and protein/organelle metabolism. In tumors, mTOR is a survival factor that increases tumor mass through the recruitment of mitogens and nutrients19 and is unusually sensitive to cellular adenosine triphosphate levels, perhaps directly sensing and regulating adenosine triphosphate concentration.19 In addition, phosphatidic acid has been identified as a critical element of mTOR signaling; current models suggest that parallel signaling pathways in translational regulation converge on common effector proteins, which could reveal an important function of this lipid in the regulation of divergent signal transduction and protein synthesis cascades, as well as providing a direct link between mTOR and mitogens, including xenobiotics.20,21 It should be noted that although phosphoinositide 3-kinase and mTOR share roles in the same overall pathway, a linear regulatory relationship between the two has not yet been shown.21 Although mTOR's main role seem to be involved in the assessment of amino acid availability, it may also have a role in the detection of DNA damage.22 Many other members of the growing phosphoinositide kinase kinase-related protein family (which includes ATM, ATR, and DNA-PK) are known to perform roles in DNA damage detection as well as checkpoint control. Inhibitors of the mTOR pathway, including rapamycin, are being tested in phase II clinical studies, including studies in breast cancer patients.
This study shows that expression array technology can effectively begin to penetrate the genetic complexity and heterogeneity of human breast cancers and give pertinent clues behind tumor biology including sensitivity and resistance to therapies. This kind of molecular profiling could have profound clinical applications in designing synergistic combinations of agents and defining the optimal treatment selection for each individual patient. Incomplete responses to docetaxel are frequent. These pilot results suggest that adding mTOR inhibitors might increase the overall efficacy of docetaxel, and if so, increase the overall likelihood of survival and cure in women with breast cancer.
Authors' Disclosures of Potential Conflicts of Interest
NOTES
Supported in part by the US Army Medical Research and Materiel Command grant No. DAMD17-01-0132, a Grant-in-Aid (US 11115) from Aventis Pharmaceutical Inc, the Emma Jacobs Clinical Breast Cancer Fund, and the Breast Cancer Specialized Program of Organized Research Excellence grant No. P50 CA50183 from the National Cancer Institute.
Presented in part at the 39th Annual American Society of Clinical Oncology Meeting, Chicago IL, May 31-June 3, 2003.
J.C.C. and E.C.W. contributed equally to this work.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Early Breast Cancer Trialists' Collaborative Group: Systemic treatment of early breast cancer by hormonal, cytotoxic or immune therapy: 133 randomised trials involving 31,000 recurrences and 24,000 deaths among 75,000 women. Lancet 339:1-15, 1992
2. Early Breast Cancer Trialists' Collaborative Group: Tamoxifen for early breast cancer: An overview of the randomised trials. Lancet 351:1451-1467, 1998
3. Early Breast Cancer Trialists' Collaborative Group: Polychemotherapy for early breast cancer: An overview of the randomised trials. Lancet 352:930-942, 1998
4. Hortobagyi GN: Docetaxel in breast cancer and a rationale for combination therapy. Oncology 11:11-15, 1997
5. Chan S, Friedrichs K, Noel D, et al: Prospective randomized trial of docetaxel versus doxorubicin in patients with metastatic breast cancer: The 303 Study Group. J Clin Oncol 17:2341-2354, 1999
6. Nabholtz JM, et al: A phase III trial comparing docetaxel (T), doxorubicin (A) and cyclophosphamide (C) (TAC) to (FAC) as first line chemotherapy for patients with metastatic breast cancer. Proc Am Soc Clin Oncol 20:22a, 2001 (abstr 83)
7. Henderson IC, Berry DA, Demetri G, et al: Improved outcomes from adding sequential paclitaxel but not from escalating doxorubicin in an adjuvant chemotherapy regimen for patients with node-positive primary breast cancer. J Clin Oncol 21:976-983, 2003
8. Aapro MS: Adjuvant therapy of primary breast cancer: A review of key findings from the 7th international conference, St. Gallen, February 2001. Oncologist 6:376-385, 2001
9. Fisher B, Bryant J, Wolmark N, et al: Effect of preoperative chemotherapy on the outcome of women with operable breast cancer. J Clin Oncol 16:2672-2685, 1998
10. Mamounas EP: Preoperative doxorubicin plus cyclophosphamide followed by preoperative or postoperative docetaxel. Oncology 11:37-40, 1997
11. Chang JC, Wooten EC, Tsimelzon A, et al: Gene expression profiling predicts therapeutic response to docetaxel (Taxotere) in breast cancer patients. Lancet 362:362-369, 2003
12. Irizarry RA, Bolstad BM, Collin F, et al: Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 31:e15, 2003
13. Li C, Wong WH: Model-based analysis of oligonucleotide arrays: Expression index computation and outlier detection. Proc Natl Acad Sci U S A 98:31-36, 2001
14. Schena M, Shalon D, Davis RW, et al: Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270:467-470, 1995
15. Khan J, Simon R, Bittner M, et al: Gene expression profiling of alveolar rhabdomyosarcoma with cDNA microarrays. Cancer Res 58:5009-5013, 1998
16. Lockhart DJ, Dong H, Byrne MC, et al: Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 14:1675-1680, 1996
17. Sgroi DC, Teng S, Robinson G, et al: In vivo gene expression profile analysis of human breast cancer progression. Cancer Res 59:5656-5661, 1999
18. Perou CM, Jeffrey SS, van de Rijn M, et al: Distinctive gene expression patterns in human mammary epithelial cells and breast cancers. Proc Natl Acad Sci U S A 96:9212-9217, 1999
19. Dennis PB, Jaeschke A, Saitoh M, et al: Mammalian TOR: A homeostatic ATP sensor. Science 294:1102-1105, 2001
20. Fang Y, Vilella-Bach M, Bachmann R, et al: Phosphatidic acid-mediated mitogenic activation of mTOR signaling. Science 294:1942-1945, 2001
21. Martin KA, Blenis J: Coordinate regulation of translation by the PI 3-kinase and mTOR pathways. Adv Cancer Res 86:1-39, 2002
22. Tee AR, Proud CG: DNA-damaging agents cause inactivation of translational regulators linked to mTOR signalling. Oncogene 19:3021-3031, 2000(Jenny C. Chang, Eric C. W)