AT-101 or placebo (P) with docetaxel (D) in second-line NSCLC with gene signature biomarker development

3577

Background: AT-101 inhibits the Bcl-2 protein family (Bcl-2, Bcl-xL, Mcl-1, Bcl-W) with broad preclinical activity including synergy with D. Methods: We conducted a randomized, double-blind, P-controlled phase 2 study. Patients had received 1 prior chemotherapy regimen for advanced NSCLC. Patients received AT-101 (40 mg b.i.d. days 1–3) or P with D (75 mg/m2 day 1) Q 21 days. Primary endpoint was progression-free survival (PFS) by independent review (IR). Secondary endpoints were overall survival (OS) and investigator determined PFS. 102 patients were planned for 70 events (80% power, HR 0.6, 1-sided alpha 0.1). A biomarker of AT-101 activity was developed by treating 55 NSCLC cell lines with AT-101 and using the corresponding gene expression data to identify a genomic predictor of sensitivity to AT-101. Results: 106 patients were randomized. Baseline factors were balanced. Median OS for patients on AT-101+D was 7.3 months versus 5.6 months for P + D arm (HR 0.60, p = 0.05). The median investigator determined PFS was 12.6 weeks and 10.7 weeks for AT-101+D and P+D arms respectively (HR 1.0, p = 0.49); IR is ongoing. Common adverse events were fatigue (18%), anemia (18%), and dyspnea (18%). Grade 1/2 headaches occurred more frequently in the AT-101 arm. Analysis of gene expression data by Bayesian regression revealed a robust 500 gene predictor of sensitivity to AT-101 that was cross validated in a leave one out analysis and in an independent cohort of 32 NSCLC cell lines. Validation of the predictor in patient samples from the trial is ongoing. Conclusions: In this phase II trial AT-101, an oral pan Bcl-2 family inhibitor, had favorable OS compared to placebo when combined with docetaxel in previously treated NSCLC and was well tolerated. A genomic predictor of AT-101 sensitivity is likely to enrich for responders and identify novel therapeutic combinations for future

[Table: see text]

Implementation of genomic predictors of chemotherapy response for guiding preoperative therapy in a prospective breast cancer trial

11057

Background: Personalized approaches to breast cancer therapy depend on genomic assays. While assays based on fixed tissues offer greater convenience, the spectrum of biology interrogated is limited. Full-transcriptome assays using microarrays are more challenging, but have the advantage of providing multiple prognostic and predictive signatures in one assay. We have created a clinical infrastructure with the objective of obtaining full genome expression data on breast cancer samples as a clinical assay for use in a prospective trial. Methods: “Performance of Genomic Expression Profiles to Direct the Use of Preoperative Chemotherapy for Early Stage Breast Cancer” is a prospective trial validating genomic signatures for predicting response to doxorubicin (A) or docetaxel (T) treatment in HER2 negative cancers. Fresh-frozen cores are reviewed by the study pathologist for tumor content. RNA is then extracted and probe generated to hybrize to an Affymetrix U133Plus2.0 microarray. Microarray data quality is determined using summary metrics for U133Plus2.0 arrays and principal component analysis (PCA) plots. The data is then used for predicting A or T sensitivity. Results: Thirteen cancers have been analyzed in the context of the above trial. Histologic type was lobular for 1, and predominantly ductal for 12 (10 ER positive, 3 ER negative). Median tumor size was 3.4 cm (range, 1.8–5.7). Microarray analysis was successful on 11 tumors (84%), providing data of sufficient quality to make predictions of A and T sensitivity. One sample hybridization failed QC as detected by PCA analysis, and one sample had insufficient RNA. The median “tissue to array data” time and “study consent to initiation of treatment” time were 5 days (range, 3–8) and 14 days (range, 11–36), respectively. Conclusions: Our initial experience shows that full-genome expression analysis on frozen tumor using an Affymetrix platform is feasible as a clinical assay for breast cancer. The resulting data is being used in a prospective marker validation protocol for predicting chemosensitivity. The data can also be analyzed for a variety of other potential prognostic and predictive signatures for guiding therapy.

No significant financial relationships to disclose.

Age-specific differences in oncogenic pathway deregulation and chemosensitivity in patients with acute myeloid leukemia: Strategies to maximize response to induction chemotherapy

7013

Background: Despite all the advances made in understanding the poor prognosis of acute myeloid leukemia (AML) in the elderly, the underlying biology at a molecular signaling pathway level has yet to be defined. Methods: Clinically annotated, microarray data from 425 patients with newly diagnosed AML from two publicly available datasets GSE1159 and GSE12417 were analyzed. Age-specific cohorts (young ≤45 years; n = 175 and elderly ≥55 years; n = 144) were prospectively identified. Gene expression analysis was conducted by applying previously defined and tested signature profiles reflecting deregulation of oncogenic signaling pathways, altered tumor environment, and signatures of chemotherapy sensitivity. Standard Kaplan-Meier survival curves were generated using the two-sided log-rank test and individual differences in the probability of oncogenic pathway deregulation between young vs. elderly were analyzed via the non-parametric Mann-Whitney U test and a one-sided p-value ≤ 0.05 was considered statistically significant. Results: Elderly AML patients had worse OS (median 8.8 months vs. 24.1 months in younger patients; p = 0.001) and EFS (median 7.1 months vs. 15.3 months in younger patients; p < 0.0001). Analysis of oncogenic pathways revealed that older patients had higher probability of Ras, Src, and TNF pathway activation, p < 0.0001. Older patients were also less sensitive to anthracycline compared to younger AML patients, p < 0.0001. Unsupervised hierarchical clustering of younger AML patients revealed two clusters and clinically better survival for cluster 1 compared to cluster 2 (high Ras, Src, TNF pathway activation) and the latter were in turn less sensitive to adriamycin. However, in elderly patients those in cluster 2 also had high Ras, Src, TNF but this did not translate into differences in survival or chemotherapy sensitivity. Conclusions: AML arising in the elderly represents a distinct biologic entity characterized by unique patterns of deregulated signaling pathways that contributes to poor survival and resistance to adriamycin. We hope these findings will enable clinically meaningful adjustments of treatment strategies in the older patient population.

No significant financial relationships to disclose.

Characterizing the clinical relevance of an embryonic stem cell phenotype in lung adenocarcinoma

11001

Background: Cancer cells possess traits reminiscent of those ascribed to normal stem cells. It is unclear whether these phenotypic similarities between normal/embryonic stem cells and mature tumor cells, specific to lung cancer, are a result of underlying biologic processes, such as specific molecular pathways and regulatory networks. Methods: Using a large cohort of lung cancer cell lines with associated gene expression data, genes associated with an embryonic stem cell identity were used to develop a ‘signature’ representative of embryonic stemness (ES) activity specific to lung adenocarcinoma. Differential biology was evaluated using Gene Set Enrichment Analysis (GSEA) and signatures of oncogenic pathway deregulation. The ES signature was applied to three independent early (stage I - IIIa) lung adenocarcinoma data sets (N = 634) with clinically annotated gene expression data. The relationship between the ES phenotype and cisplatin sensitivity was also evaluated. Results: Using Bayesian regression analysis, a 100 gene signature representative of ES activity in lung adenocarcinoma was developed and validated in a leave-one-out-analysis. GSEA identified gene sets significantly represented in the ES signature: signature of neoplastic transformation, signature of undifferentiated cancer, BRCA pathway, and fibroblast serum response pathway, all associated with cancer invasiveness. Adenocarcinomas with ES demonstrated increased activation of RAS (p = 0.0002), MYC (p = 0.0057), wound healing (angiogenesis) (p < 0.0001), chromosomal instability (p < 0.0001), and invasiveness (p < 0.0001) gene signatures. Adenocarcinomas (N= 634) with ES had a decreased survival (p<0.04). The ES signature was not prognostic in prostate, ovarian, or breast adenocarcinomas. Lung tumors (N=634) and adenocarcinoma cell lines (N=31) with ES were more resistant to cisplatin (p<0.0001 and p=0.0063, respectively). Conclusions: Lung adenocarcinomas that share a common gene expression pattern with normal stem cells were associated with decreased survival and increased likelihood of resistance to cisplatin, indicating the aggressiveness of lung tumors with a stem cell phenotype.

No significant financial relationships to disclose.

Integration of mRNA and microRNA profiles as prognostic and predictive markers in lung adenocarcinoma

7522

Background: Lung adenocarcinoma (ADC) is a distinct biologic entity with unique gene amplifications (Weir B, Nature 2008). Yet, comprehensive transcriptomic analysis, including microRNAs, specific to lung ADC are lacking. Methods: Using mRNA expression data from a discovery cohort of 154 patients with histologically proven early stage (I and II) lung ADC, signatures of oncogenic pathway and tumor microenvironment status were applied and further organized by hierarchical clustering to develop a metagene model. Further, using in vitro assays in a large cohort of lung ADC cell lines (n = 42) with corresponding mRNA and microRNA data, novel microRNAs associated with a poor prognosis and their relationship to cisplatin resistance was elucidated. Results: In the discovery cohort of 154 patients with early stage disease, activation of oncogenic pathways associated with wound healing (angiogenesis), chromosomal instability, and STAT signaling were associated with an increased risk of recurrence (p<0.001). Utilizing the extremes of survival to identify cohorts of patients as high and low risk phenotypes, using bayesian regression, a 100 gene signature (‘metagene') that captured the diversity of signaling pathways unique to patients at increased risk of recurrence was identified and validated in an independent cohort (n = 364) of lung ADC samples with 78.3% accuracy. Kaplan Meier survival analysis and multivariate analysis further confirmed the independent prognostic value of the 100 gene signature (p= 0.007). Using in vitro cell proliferation assays, predicted high risk lung ADC cell lines were identified as being more resistant to cisplatin therapy than those predicted to be low risk (p=0.001). In a novel manner, we also identified several microRNAs (miR-215, miR-98, miR- 643, let-7b, miR-665, miR-629) associated with a high risk of recurrence and more importantly cisplatin resistance. Conclusions: mRNA and microRNA profiles reflect unique aspects of individual tumors and may characterize histology-specific tumor heterogeneity in lung ADC, providing an opportunity to better characterize the oncogenic process and refine therapeutic options.

No significant financial relationships to disclose.

The multi-institutional myeloma group clinico-genomic risk stratification system: A blinded validation

8521

Background: Several clinical and molecular prognostic factors (e.g, International Staging System [ISS] stage, plasma cell labeling index, genomic models) exist for multiple myeloma (MM). We hypothesized that exploiting gene signatures representative of oncogenic pathway deregulation (i.e., Ras, Myc, etc.), would improve MM prognostication and also aid with the identification of novel therapeutic targets. Methods: Using a discovery cohort (n=47) of patients with MM and corresponding gene expression data, we built upon current molecular risk-stratification and devised a Bayesian genomic (“metagene”) model for prognosis. We validated that model in an independent patient cohort (n=207). Finally, we incorporated ISS staging and clinical variables to construct a combined Clinico-Genomic Risk Stratification System. We further validated the combined model in a separate cohort (n=72), in a blinded manner. Results: Using gene signatures predictive of oncogenic pathway activation in the discovery cohort, we identified specific patterns (metagenes) of signaling pathway activation with prognostic relevance. In an independent validation cohort, this metagene-based model accurately predicted event free survival (EFS) independently of ISS (multivariate hazard ratio [HR] 3.4 for ISS stage, and 5.4 for the metagene model, p=0.002). Using multivariate risk modeling, we incorporated ISS staging and the metagene model into a Clinico-Genomic System and successfully stratified the validation cohort into three groups (low, intermediate, and high risk) with markedly different EFS (HR 4.2 for intermediate risk and 14.0 for high risk vs. the low risk cohort, p<0.0001). In an additional blinded validation, the Clinico-Genomic System again accurately predicted median overall survival (68.7 [low risk] vs 24.7 [intermediate risk] vs 18.7 months [high risk], p<0.0001); more accurately than either ISS or other reported genomic models. Conclusions: A combined Clinico-Genomic Risk Stratification System, building on patterns of oncogenic pathway activation and ISS staging system, improves upon current prognostic models in MM and identifies novel pathway targets for future therapeutic consideration.

No significant financial relationships to disclose.

Gene expression patterns of oncogenic signaling pathways provide a novel approach to targeted therapeutics in advanced non-small cell lung cancer (NSCLC)

7503

Gene expression profiles reflect unique aspects of individual tumors and may characterize tumor heterogeneity in NSCLC. We used gene expression signatures that reflect the deregulation of major oncogenic signaling pathways (Ras, PI3 kinase, Akt, Src, β-catenin, E2F, and Myc) to identify patterns of pathway activation in 118 samples from patients with recurrent (n = 69) and metastatic (n = 49) NSCLC. These individuals, in particular, the patients with metastatic disease are likely to be resistant to standard chemotherapy pointing to the need for alternative therapeutic options. As one approach to this challenge, we used Affymetrix expression data to identify patterns of pathway deregulation in patients with advanced (recurrent/metastatic) disease and compared them to our previously described data on early stage NSCLC (n = 91). Unique to the recurrent and metastatic disease cohorts, a majority of the tumors (greater than 80%) had deregulation of multiple signaling pathways. In particluar, NSCLC tumors with deregulated Src and Myc pathways had a much worse prognosis than those with deregulated Ras and Akt pathways [median survival: 15.5 months vs. 55.3 months, hazard ratio: 2.6, p < 0.01]. Further, using cell proliferation assays, we have shown that the predicted deregulation of Ras, Src and PI3K pathways was directly proportional (p < 0.01, log rank test) to the sensitivity of NSCLC cell lines (n = 17) to agents that specifically target these pathways, suggesting that an ability to profile the status of oncogenic pathways in a tumor sample provides an opportunity to better characterize the oncogenic process in patients with recurrent and advanced NSCLC. More importantly, such a strategy provides a novel approach to targeted therapeutics in NSCLC by guiding the appropriate use of pathway specific inhibitors. Finally, data on gene expression patterns involving other signatures representing unique cancer biology relevant to NSCLC (e.g. hypoxia, chromosomal instability, and angiogenesis) will be also be presented.

No significant financial relationships to disclose.

Molecular signatures characterize early stage breast cancer arising in young women and have prognostic and therapeutic implications independent of ER status

522

Background: Breast cancer in young women [YW] has a poorer prognosis, is less likely hormone sensitive, and has yet to be defined as a unique biologic phenotype. The identification of molecular pathways unique to YW should enable refinement of treatment strategies. Methods: To optimize characterization of breast cancer arising in YW vs. older women [OW], we prospectively defined 2 age groups constituting pre- and post-menopausal women [YW = 45 yrs (median 40 yrs); OW = 65 yrs (median 71 yrs)]. Age groups were applied to 784 clinically-annotated microarray (Affymetrix) data. 411 patients (n = 200 = 45yrs; n = 211 = 65 yrs) were eligible. In addition to clinical parameters and differential gene expression analysis, genomic signatures of oncogenic pathway deregulation (Ras, E2F, β-catenin, PI3K, Myc, Src) and chemotherapeutic sensitivity were evaluated. Results: In the 2 age-defined tumor sets, there was no difference between ER or PR status by IHC (p = ns). Consistent with previous data sets, YW demonstrated higher IHC expression of ErbB2 (p = 0.037), higher grade tumors (p < 0.0001), larger tumor size (p = 0.0002) and greater lymph node positivity (p = 0.045). Quantitative ER expression (mRNA) was lower (p < 0.0001) among YW. In comparing deregulation of oncogenic pathways between age groups, a statistically higher probability of PI3K (p = 0.005) and Myc (p = 0.02) pathway deregulation was observed among YW. Importantly, hierarchical clustering revealed unique patterns of pathway deregulation. In YW, PI3K and β-catenin deregulation conferred a worse prognosis compared to those with Myc and Src deregulation (HR = 4.15; p = 0.008). In contrast, among OW, E2F deregulation with concurrent low probability of PI3K and Myc deregulation was associated with a poorer outcome (HR = 2.7; p = 0.006). Conclusion: Results of this large study demonstrate that breast cancer arising in YW represents a distinct biologic entity characterized by unique patterns of deregulated signaling pathways that are prognostic, independent of currently available clinical variables such as ER. Further data, including unique patterns of chemotherapeutic sensitivity as they relate to clinical parameters and oncogenic pathway status, will also be presented.

[Table: see text]

A genomic approach to identify mechanisms associated with chemotherapy resistance

2534

Background: Gene expression profiling has shown an ability to predict chemotherapeutic response (Potti et al. Nature Medicine 2006). Building on this work, we used a genomic strategy to explore the biology associated with the development of chemotherapy resistance and sought to determine if disease context impacted results. Methods: Gene set enrichment analysis (GSEA) was performed on expression data for NCI60 cell lines sensitive and resistant to specific chemotherapeutic agents (adriamycin, cyclophosphamide, docetaxel, etoposide, 5-fluoruracil, paclitaxel, topotecan). GSEA was additionally performed on a series of lung cancer cell lines with a defined sensitivity to cisplatin and docetaxel. Sensitive and resistant cell lines with individual mean and confidence intervals greater than 1SD from the mean across all samples were included. Adjusting for multiple hypothesis testing, gene sets with a false discovery rate (FDR) <0.25 were deemed statistically significant. In the discovery mode, gene sets with a nominal p-value <0.05 were also of interest. Finally, overlapping gene sets between agents were assessed. Results: Statistically-significant gene sets, representing biologic pathways associated with resistance, were identified for the various chemotherapeutic agents (i.e., cell death, erbb3, and bad pathways associated with docetaxel resistance). No gene sets with FDR <0.25 or nominal p-value <0.05 were common to all drugs. In assessing disease specific resistance, 22 lung cancer cell lines for cisplatin (15 sensitive, 7 resistant) and 14 lung cancer cell lines for docetaxel (10 sensitive, 4 resistant) were analyzed. GSEA identified the bcl-2 pathway (p<0.002) to be associated with cisplatin resistance. In contrast, the proteosome (p=0.01) and akt (p=0.02) pathways were associated with docetaxel resistance. Pathways involved in the production of V-H+-ATPase were enriched in cisplatin and docetaxel resistant lung cancer cell lines suggesting a global mechanism of resistance. Conclusions: These results support the use of a genomic approach to identify unique drug-specific and global therapeutic targets associated with the development of chemotherapy resistance. Interestingly, disease context appears important in identifying novel targets.

No significant financial relationships to disclose.

Pharmacogenomic strategies provide a rational approach to the treatment of cisplatin-resistant patients with advanced non- small cell lung cancer (NSCLC)

7501

Background: Standard treatment for advanced non-small lung cancer (NSCLC) includes platinum-based therapy. Response is however suboptimal and unpredictable. Newer agents, such as pemetrexed, have shown significant activity in the second-line setting, but their role in the first-line setting is unknown. Methods: Using in vitro drug sensitivity data, coupled with Affymetrix microarray data, gene expression signatures predicting sensitivity to cisplatin and pemetrexed were developed, using methods previously described (Nature Medicine, 2006). Signatures were validated with response data from 32 independent ovarian and lung cancer cell lines as well as 59 patient samples. Results: Genomic-derived signatures of cisplatin and pemetrexed sensitivity were shown to accurately predict response in vitro and, in the case of cisplatin, in 59 samples obtained from advanced disease patients previously treated with cisplatin. The accuracy of the cisplatin predictor, using clinical response as the endpoint, was 83.1% (Sensitivity: 100%, Specificity: 57%). In comparison, determination of ERCC1 status (IHC) had a predictive accuracy of 62% (Sensitivity: 82.8%, Specificity: 33.3%). Interestingly, a significant inverse correlation (p = 0.004) was found between the likelihood of cisplatin and pemetrexed sensitivity in the NSCLC cell lines. This relationhip was further validated in another indepedent cohort of 91 patients with NSCLC (p < 0.01). Finally, using novel cell proliferation experiments, we show that tumors from cisplatin refractory patients that are initially sensitive to pemetrexed, if treated with a taxane (docetaxel or paclitaxel) prior to pemetrexed therapy, develop resistance to subsequent pemetrexed therapy. Conclusions: The use of gene expression signatures that predict cisplatin and pemetrexed sensitivity can be incorporated into strategies to refine the selection of chemotherapeutic drugs chosen as first-line therapy in advanced NSCLC. It is critical for future prospective studies to validate the use of pharmacogenomic predictors of chemosensitivity in NSCLC, to optimize response rates and survival benefit.

No significant financial relationships to disclose.

A genomic signature of chemotherapeutic multidrug resistance provides a rational approach to identify patients for targeted therapy in advanced solid tumors

10613

Background: Multidrug resistance (MDR), the process by which cells resist many structurally and functionally unrelated drugs, remains a major challenge for cancer treatment. Strategies to reliably predict MDR will revolutionize cancer therapy. Methods: We made use of in vitro drug sensitivity data on 47 FDA approved chemotherapeutic agents from the NCI-60 database coupled with Affymetrix microarray data, to develop a gene expression signature of chemotherapeutic MDR, using a 'metagene’ approach to gene expression analysis (Potti et al, Nature Medicine, 2006). The capacity of the MDR signature to predict in vitro and in vivo resistance was evaluated in multiple independent cell line and patient cohorts. Results: A gene expression signature of chemotherapeutic MDR developed from cancer cell lines using a novel ‘metagene’ approach can accurately predict patients resistant to standard chemotherapeutic agents, including multi-drug regimens. This includes cohorts of patients with early stage breast cancer (n = 171; accuracy: 75.5%, sensitivity: 49.3%, specificity: 92%), patients who received neoadjuvant chemotherapy (n = 51; accuracy: 78.5%, sensitivity: 73.7%, specificity: 92.3%), and ovarian cancer (n = 105; accuracy: 83.8%, sensitivity: 71.8%, specificity: 93.7%). We further show that MDR signature expression is an independent prognostic factor in early stage breast cancer (p = 0.02, odds ratio: 3.49) and a large cohort of non-small cell lung cancer (p = 0.03, odds ratio: 5.70), in a multivariate analysis. Finally, to identify a rational therapeutic strategy in patients with the MDR phenotype, we have integrated the prediction of MDR with profiles of oncogenic pathway deregulation (Bild A, Nature, 2006) and show that targeting the Src pathway using a Src specific inhibitor (SU6656) circumvents MDR in the in vitro setting. Conclusions: The development of a gene expression signature of MDR, that can predict resistance to commonly used cytotoxic chemotherapies, provides a rational approach to developmental therapeutics, by identifying novel 'customized’ candidate drugs (e.g. targeting Src) for further clinical development.

No significant financial relationships to disclose.

Clinical validation of in vitro drug sensitivity microarray data: Regimen-specific signatures predict pathological complete response to neo-adjuvant chemotherapy for breast cancer in a randomized trial (EORTC 10994/BIG 00–01)

544

Background: We previously described gene expression signatures that predict sensitivity to common chemotherapeutic agents and published promising results of their applicability in patients (Nature Med 2006). The goal of this study was to confirm their validity in a larger series of breast cancer patients with hormone-receptor negative (HR negative) since these tumours are more sensitive to chemotherapy. We used pathological complete response as a surrogate for chemosensitivity. We analyzed samples from a subset of patients included in a recently completed large neoadjuvant phase III trial. The trial compares a non-taxane regimen (fluorouracil + epirubicin + cyclophosphamide × 6; FEC arm) with a taxane regimen (docetaxel × 3 then epirubicin + docetaxel × 3; T->ET arm). Methods: RNA prepared from frozen samples obtained at diagnosis were hybridized to Affymetrix arrays. In vitro single agent signatures generated using a metagene approach were combined to obtain a FEC and a T->ET regimen-specific signatures. Predictions were blinded to patient outcome. With both signatures we calculated the receiver operating curve, its AUC, and the cut-point with maximal Youden index- accuracy, positive predictive value (PPV), sensitivity (Sens), negative predictive value (NPV) and specificity (Spec). Results: Samples from 124 patients (55 pCR) with HR negative tumours underwent a successful gene-expression array: 65 patients were treated in FEC arm and 59 patients in T->ET arm. The results are summarized below. Conclusions: We have validated the approach of using regimen-specific genomic signatures developed in vitro, in the context of a multicenter randomized trial. These results support the activation of a prospective trial comparing the conventional random choice of chemotherapy versus a specific array based approach.

[Table: see text]

[Table: see text]

A genomic strategy to combinatorial therapeutics in solid tumors

2031

Background: For most advanced solid tumors, the response rate to cytotoxic drugs is generally low, highlighting the importance of identifying those patients most likely to respond, either to single agents or combinations of cytotoxic or targeted therapies. Methods: We have made use of in vitro drug response data generated on the NCI-60 panel of cancer cell lines, coupled with Affymetrix U133 2.0 plus gene expression data, to develop genomic predictors of chemotherapy sensitivity. These models were then validated in independent cancer cell lines as well as response data from patient treatment studies. Results: Predictive models making use of gene expression data were developed for docetaxel, adriamycin, 5-flourouracil, cyclophosphamide, paclitaxel, and topotecan. These models were shown to accurately predict sensitivity to the drugs in an independent set (n = 30) of cancer cell lines. Importantly, three of the predictors (docetaxel, topotecan, paclitaxel) also accurately (> 80%) predicted response in patient studies. When evaluated in a large collection of human cancers (n = 381), these gene expression signatures of drug response identified patterns of predicted sensitivity suggesting potential opportunities for novel combinations. We also combined the predictions of chemotherapy sensitivity with predictions of pathway deregulation (Bild A, Nature 2005), to develop further opportunities for combination therapy. For instance, this analysis revealed a significant relationship between PI3 kinase pathway deregulation and docetaxel resistance (p = 0.001), and a correlation between docetaxel sensitivity and the activation of the Rb/E2F pathway (p = 0.009). Furthermore, cell lines showing an increased probability of PI3 kinase and Rb/E2F activation were also more likely to respond to a PI3 kinase (LY-294002) inhibitor (p = 0.01) or R-Roscovitine (p = 0.03), a cell cycle inhibitor, respectively. Conclusions: The development and validation of chemotherapeutic response predictors, together with oncogenic pathway signatures that can guide the use of targeted agents, provides an opportunity to develop effective combinatorial therapeutic strategies geared to the individual patient.

No significant financial relationships to disclose.

Gene expression patterns identify patients with non-small cell lung cancer (NSCLC) who are at increased risk for venous thromboembolism (VTE)

8535

Background: VTE is the leading cause of death in patients with cancer. The 1 year survival rate in patients diagnosed with cancer at time of VTE is 12% compared to 36% in cancer patients who are free of VTE .Also, a thrombotic event in a hospitalized cancer patient leads to higher in-hospital mortality.Thus,VTE in cancer patients may indicate a more aggressive phenotype. Methods: With the overarching goal of testing the ability of genomic profiling to detect biologically and clinically significant differences in patients with cancer and VTE, using NSCLC as a proof of principle, we expanded on previous work wherein we have shown the ability of multiple gene expression patterns (‘metagenes’) to predict a thrombotic event .96 patients with NSCLC were enrolled in this study. RNA was extracted and gene array data obtained using an Affymetrix U133 2.0 plus GeneChip. The clinical history of all 96 patients was reviewed to identify patients with a definitive VTE episode after the initial diagnosis of cancer and not within six weeks of surgery.14/96 (14.5%) patients met these criteria. We then identified 14 additional patients with NSCLC matched by age, gender and clinical stage patients who did not have VTE for at least 2 years following the diagnosis of cancer. Gene expression data was analyzed using a binary regression analysis. Results: Using the metagene approach, a discriminator gene set (n = 45) that differentiated patients with NSCLC and VTE from patients with NSCLC without VTE was identified. A leave-one-out cross validation performed to further assess the reliability of the discriminator metagene set was more than 85% accurate in identifying patients with NSCLC and VTE. Also, significant biologically relevant differences were seen between the comparison groups, to include genes such as P53, VEGFC, E2F4, TFPI and EPHB2 Conclusions: Our data suggests that a genomic approach can be used to identify with NSCLC that develop VTE, while also providing information important to an understanding of the underlying biology of the association between cancer and thrombosis. We are in the process of validating these results in a prospective trial.

No significant financial relationships to disclose.

A genomic strategy to refine prognosis in early stage non-small cell lung carcinoma (NSCLC)

7026

Background. Although stage-specific classification identifies appropriate populations for adjuvant chemotherapy, this is likely an imprecise predictor for the individual patient with early stage NSCLC. Methods. Using previously-described methodologies that employ DNA microarray data, multiple gene expression profiles (‘metagenes’) that predict risk of recurrence in patients with stage I disease were identified. This analysis used an initial ‘test’ cohort of patients with NSCLC (n = 89) that represented an equal mix of squamous cell and adenocarcinoma. Also, each histologic subset had equal number of patients who survived more than 5 years and those who died within 2.5 years of initial diagnosis. The performance of the metagene-based model generated on the training cohort was then evaluated in independent ‘validation’ sets, including two multi-center cooperative group studies (ACOSOG Z0030 and CALGB 9761). Importantly, the CALGB validation was performed in a blinded fashion. Results. Classification tree analyses that sample multiple gene expression profiles were used to develop a model of recurrence, termed the Lung Metagene Model, that accurately assesses prognosis (risk of recurrence and survival), performing significantly (p<0.001, odds ratio: 16.1, multivariate analysis) better than pathologic stage, T-size, nodal status, age, gender, histologic subtype and smoking history. The accuracy of prognosis using the Lung Metagene Model exceeded 90% (leave-one-out cross validation) in the initial training set (n = 89), 72% in the ACOSOG (n = 25), and 81% in the CALGB (n = 84) datasets. The prognostic accuracy was consistent across histologic subtypes and stages of NSCLC. Importantly, this provides an opportunity to re-classify stage IA patients to identify a subset of ‘high risk’ patients that may benefit from adjuvant chemotherapy. Further, stage IB and II patients identified as ‘low risk’ for recurrence, and who present co-morbidities, could potentially be candidates for observation, and those patients predicted to be at ‘high risk’ may benefit from novel therapeutic trials. Conclusions. The Lung Metagene Model provides a mechanism to refine the estimation of an individual patient’s risk for disease recurrence and thus guide the use of adjuvant chemotherapy in NSCLC.

No significant financial relationships to disclose.

CD117 (c-KIT) overexpression in patients with extensive-stage small-cell lung carcinoma

BACKGROUND

The aim of this study was to determine the incidence and role of CD117 (c-KIT) overexpression as a predictive/prognostic marker in extensive-stage small-cell lung carcinoma (ESSCLC). We performed a retrospective study on subjects with a biopsy-proven diagnosis of ESSCLC.

PATIENTS AND METHODS

A chart review for demographic and clinical data was performed on patients with ESSCLC diagnosed between 1991 and 2001. CD117 overexpression was evaluated using immunohistochemistry (A4052 polyclonal antibody) performed on archival paraffin-embedded specimens.

RESULTS

Two hundred and twenty-three patients with ESSCLC were identified, of whom 193 (84 females, 109 males) with a mean age of 68.5 years (range 42-90) had adequate tissue specimens available for CD117 testing. The most commonly presenting symptom was weight loss, seen in 61 patients (31.6%). Of the 193 specimens, 54 (27.9%) showed CD117 overexpression. The median length of survival for CD117-positive patients was 9 months as compared with the CD117-negative population, in whom the survival was 6 months (P = 0.025, Cox proportional hazard method).

CONCLUSIONS

CD117 overexpression detected using immunohistochemistry is observed in about a third of patients with ESSCLC and does not have statistically significant prognostic value. However, CD117 may be a potential target for site-specific immunotherapy in ESSCLC. Our findings suggest a role for clinical trials assessing the role of selective tyrosine kinase inhibitor STI-571 (alone or in combination with conventional therapy) in patients with ESSCLC.

Thrombocytopenia associated with c7E3 Fab (abciximab)

Abciximab (c7E3 Fab) inhibits platelet aggregation and is used to prevent complications of percutaneous coronary intervention. Thrombocytopenia is an often-cited complication of abciximab. Pseudothrombocytopenia is due to ethylenediaminetetraacetate (EDTA)-activated platelet agglutination, resulting in a spuriously low platelet count. We have looked at both "true" and pseudothrombocytopenia after infusion of abciximab. Sixty-six patients receiving their first exposure to abciximab after an unstable coronary event/revascularization were eligible. All the patients received a bolus of c7E3 Fab followed by a continuous infusion. Platelets were monitored in all patients at 2, 4, 12, 24, and 48 h, and more frequently if required. The incidence of thrombocytopenia and acute severe thrombocytopenia (platelet count < or =20,000/microl) was evaluated. A peripheral blood smear was performed on all patients showing thrombocytopenia to evaluate for pseudothrombocytopenia. Seventeen (25.6%) developed thrombocytopenia and nine (13.6%) developed acute severe thrombocytopenia. However, 18 of these patients had pseudothrombocytopenia. The onset of true thrombocytopenia was at 4 h after the infusion, while pseudothrombocytopenia occurred at anytime during the first 24 h. Only two (3.03%) patients required platelet transfusions. No life-threatening hemorrhagic complications were recognized. Five of six subjects with true thrombocytopenia had positive laboratory findings of disseminated intravascular coagulation; however, none had an adverse outcome. Acute severe thrombocytopenia was noted to be a relatively benign adverse effect of abciximab. There is an increasing incidence of pseudothrombocytopenia in this subgroup of patients. It would be worthwhile examining a peripheral blood smear or collecting blood for platelet counts in a heparin-coated tube in order to exclude this phenomenon and thereby prevent inappropriate discontinuation of this drug.

Identification of HER-2/neu overexpression and the clinical course of lung carcinoma in non-smokers with chronic lymphocytic leukemia

Patients with CLL have an excess risk of developing second primary malignancies. The etiology of this excess risk is unclear, and has been thought to be related to smoking. HER-2/neu overexpression has evolved as a prognostic/predictive factor in some solid tumors. We reviewed our experience with non-smokers who had CLL and subsequently developed lung carcinoma, in an effort to better understand the clinical course of these patients, and to evaluate the role of HER-2/neu overexpression. We reviewed the records of all patients who had a diagnosis of both CLL and lung carcinoma between 1986 and 2000. HER-2/neu overexpression was estimated by immunohistochemistry (IHC) using the Hercep test (DAKO). An IHC score of 2+ or greater was considered positive. Overall survival was calculated from the date of diagnosis of lung carcinoma by the Kaplan-Meier product limit method. Fourteen non-smokers in whom a diagnosis of CLL was made at least 6 months prior to the diagnosis of lung carcinoma were identified. The median age for diagnosis of CLL in this group was 67 years while that for lung carcinoma was 70 years. The lung carcinomas included 10 non-small cell (NSCLC) and four small cell (SCLC) carcinomas. Nine specimens (six NSCLC and three SCLC) showed HER-2/neu overexpression. Interestingly, 90% of patients with advanced stage cancer (stage IIIB/IV NSCLC or extensive SCLC) overexpressed HER-2/neu. The presence of CLL did not alter outcome in patients with early stage lung cancer. However, after adjustment for age and performance status, patients with advanced stage NSCLC and CLL had a worse than expected outcome. HER-2/neu overexpression (independent of smoking) may be involved in the development/progression of lung cancer in patients with CLL, and has an associated worse outcome. It is appropriate to consider heightened surveillance of CLL patients for lung carcinoma.

Severe Evans's syndrome secondary to interleukin-2 therapy: treatment with chimeric monoclonal anti-CD20 antibody

Interleukin-2 (IL-2) acts by increasing the efficiency of the immune system to exert a tumoricidal effect. Although it is well known that immune stimulation with IL-2 plays a role in unmasking autoimmune phenomena such as autoimmune thyroiditis, hematological effects such as anemia and thrombocytopenia are more frequently due to toxic non-immune mechanisms. We describe a patient who developed severe Evans's syndrome [autoimmune hemolytic anemia (AHA) and immune thrombocytopenic purpura (ITP)] secondary to IL-2 therapy. ITP was refractory to multiple treatment modalities including steroids and splenectomy. ITP and AHA were initially managed with intravenous gamma globulin therapy and frequent blood transfusions, respectively. Ultimately, immunosuppressive therapy with cyclophosphamide and chimeric monoclonal anti-CD20 antibody (rituximab) were successful in inducing complete remission of Evans's syndrome.

Fluconazole-induced torsade de pointes

OBJECTIVE

To present a case of fluconazole-associated torsade de pointes (TDP) and discuss fluconazole's role in causing TDP.

CASE SUMMARY

A 68-year-old white woman with Candida glabrata isolated from a presacral abscess developed TDP eight days after commencing oral fluconazole The patient had no other risk factors for TDP, including coronary artery disease, cardiomyopathy, congestive heart failure, and electrolyte abnormalities There was a temporal association between the initiation of fluconazole and TDP. The TDP resolved when fluconazole was discontinued; however, the patient continued to have premature ventricular contractions and nonsustained ventricular tachycardia (NSVT) until six days after drug cessation

DISCUSSION

Use of the Naranjo probability scale indicates a probable relationship between the use of fluconazole and the development of TDP. The possible mechanism is depression of rapidly activating delayed rectifier potassium currents. In our patient, there was no other etiology identified that could explain QT prolongation or TDP The complete disappearance of NSVT and premature ventricular contractions followed by normalization of QT interval after the drug was stopped strongly suggests fluconazole as the etiology.

CONCLUSIONS

Clinicians should be aware that fluconazole, even at low doses, may cause prolongation of the QT interval, leading to TDP. Serial electrocardiographic monitoring may be considered when fluconazole is administered in patients who are at risk for ventricular arrhythmias.