Clinical Cancer Research

Concerns have been raised from some investigators about the positive predictive value (PPV) described in our recent publication, "Diagnostic markers for the early detection of ovarian cancer," and an erratum table published later (1, 2). An additional concern is the recent announcement by LabCorp(1) regarding the use of OvaSure as a commercial test. In this response, we focus on the questions associated with the data presented in our article, as we were not involved in the commercial marketing of this test. In our article, there are no recommendations for commercialization; that is solely the decision of LabCorp, which licensed the test from Yale University, and any concerns about commercialization should be taken up with them.A corrected table was published to correct a typographical error reported for the value of CA125 included in the details of the original Table 2, which described the three different models used in the training set for the study (2). This was not a retraction of any information originally described in the original article (1). We added to this corrected Table 2 information about the PPV for the general population based on the incidence of ovarian cancer in the general population, which is 1:2,500, which was not included in the original manuscript for reasons discussed below.The critique of the PPV is referring to the reported values of 6.5% and 99.3%. In the original article, we clearly indicate that the PPV for the test sample was 99.3%. We never indicated the value for the general population. Because the population studied at Yale did not represent the general population, it is not our intention for the test to be used for the general population. Therefore, calculating PPV based on an incidence in the general population is not relevant for this assay.The participants included in the study were enrolled from the Yale Early Detection Program for Ovarian Cancer in the Department of Obstetrics and Gynecology, a program begun in 1990 (3). The characteristics of these women have been described in previous publications and include a high percentage of participants with BRCA1 or BRCA2 gene mutations (4). High-risk populations for ovarian cancer are often seen in academic medical settings or highly specialized clinics. Thus, under those circumstances, the PPV could be much higher than that in a general population. Furthermore, because the incidence in a high-risk population may vary substantially from institution to institution or from clinic to clinic, a fixed or constant PPV can never be achieved in reality. Thus, when making a diagnostic decision, one should bear this in mind.In response to the letter from McIntosh et al. (5): The statement in their commentary that "[the authors'] conclusion, however, was determined from the combined data and for the classifier that performed best in the test set. This violates fundamental principles of statistical analysis, which requires..." is not correct. On page 1066 of the original article (1), in Materials and Methods, under the Multiplex Study Design subheading, we indicated that the study and samples were divided into two stages: stage 1 was the characterization stage, whereas stage 2 was the validation or test stage. It is clearly stated in the article that the test set was not used for constructing the classification model. Furthermore, it is clearly stated in the Results section of the Abstract and on page 1069 that the sensitivity of the four models at cutoffs corresponding to 95% specificity on the test set ranges from 84% to 98%, and not "84% or 88%" as erroneously stated in the commentary. The differences between the training and the test sets are precisely the point that we emphasize for the importance of always using a different data set for the validation stage. That is clearly described in Discussion.We also indicated that with the purpose of additional future validations, we wanted to provide a fixed classification model. Because the training samples in our article serve as a training set not only for this specific study but also for future validation studies, it was deemed more efficient to combine all samples from the training and test sets to refine the model. On page 1070 of the original article, it is acknowledged that "the final model provides more optimistic results compared with the test group, because it includes the training set... Nevertheless, this final model will be validated in a multicenter validation study." In addition, we used 10-fold cross-validation for the total cohort, which is a well-accepted approach. There are numerous studies using this analysis model. Therefore, in our article, we explain the differences between and implications of the two approaches and we report both results.It is a well-known epidemiologic fact that a case-control study design is unable to eliminate all forms of potential bias in the collection of biospecimens as a cohort study could do. On the other hand, a cohort study is prohibitively expensive and requires many years of follow-up to obtain sufficient numbers of appropriate specimens without potential biases. We are not aware of any case-control studies on ovarian cancer biomarker discovery, which include almost all ovarian biomarker studies for ovarian cancer clinical diagnosis and detection, that can control for all systematic differences in collection conditions between ovarian cancer cases and controls.It should be pointed out that in view of anticipated potential bias incorporated into the biospecimens by the collection and processing conditions, the inclusion criteria of our study population are quite rigorous relative to those of other reported studies on ovarian cancer biomarker discovery. However, as we have indicated above, it is just not possible to eliminate all potential sources of bias in a case-control study design.In response to the "stress" argument as a source of bias, we point out that both the ovarian cancer patients and the patients who had benign tumors were likely in the same stress category when the samples were collected, and the results of these two groups were different; therefore, the assertion that the results obtained in these studies may be due to a differential stress response is unlikely.In the letter from Greene et al., they indicated that with a PPV of 6.5%, only 1 of 15 women tested positive would in fact have ovarian cancer (6). We agree that this is true for all forms of cancers that have a low incidence such as ovarian and pancreatic cancers. That is the reason that there is general consensus in the biomarkers community that screening the general population has no value and it is necessary to focus on specific groups who have higher risk of developing cancer and who might benefit from the test (7).Ovarian cancer is a "rare" but lethal disease. Any test, even with a specificity of 99.9%, will have a PPV not larger than 10% if it is calculated based on the general population. If that approach is adopted, there is no value in developing a test. As correctly pointed out by Coates et al., the value of a test exists when it is applied to what is considered a high-risk population, which may further differ in incidences due to different genetic backgrounds and family histories (8). In that case, we need to calculate the PPV for each subgroup. The data reported in our article are based on the clearly defined samples that we used for testing. Therefore, it would be misleading to apply our results to a general population. Greene et al. indicate that the use of this test will lead to unnecessary surgery. We agree that this would be the case if the test had been applied to a general population (6). However, to a woman with a strong family history of ovarian cancer or known BRCA1/2 mutations, the test may offer additional choices besides surgery. Without the test, women who are tested positive for BRCA often face the only clinical option recommended by the U.S. Preventive Services Task Force-prophylactic oophorectomy. The use of this test may help to postpone or completely avoid the surgical procedure if the results of the test are negative. Furthermore, no blood test, in any clinical context, even with a high PPV, is meant to be used as a final diagnosis. A blood test is usually inexpensive and it can be tested repeatedly. This provides an excellent alternative to avoid the use of additional, more expensive but not more sensitive, tests unless clearly indicated.Coates et al. indicate in their letter that "research suggests that women at high risk of hereditary cancers may be more likely to have ovarian cancers with a poorer prognosis than are women in the general population (8). In addition, research on previous screening technologies suggests that cancers detected by screening may be more likely to have clinicopathologic features with a better prognosis than cancers diagnosed clinically. Therefore, to obtain valid estimates of sensitivity, specificity, and predictive values for screening women at high risk, research may need to be conducted within the specified high-risk populations, not among women in the general population." We agree with their suggestion, and that has been the position of our work from the beginning. Additional studies confirming our findings are necessary and will help to better understand the clinical relevance of the markers.In terms of the definition of the clinical setting used for the test, we defined the high-risk population used in our study in the Supplementary Material(2) included with the original article (1). In relation to the use of the commercial test, although it is not part of the original article (1), we would like to point out that the target population that has been offered the test is the high-risk population. The complete description of high risk is available online through the LabCorp Web page.(3)Coates et al. indicate that more information is needed on the analytic validity of the test (8). All of that information is available and accessible online.(4) The validation of the multiplex panel was done independently, and the Human Cancer Biomarker Panel kit is commercially available through Millipore(5) to anyone who would like to make additional evaluations. A complete protocol on how to run the test is available within the kit. The coefficient of variation, sensitivity, accuracy, standard curve range, and precision are clearly described.With respect to the last comment on Coates et al.'s letter: "Recommendations are likely to be more credible if they are made by third-party panels minimizing conflicts of interest" (8): The information reported in Visintin et al. (1) was done in a way that avoid as much bias as possible. All the raw data used in this study are available online for evaluation (9), and an independent statistician assigned by the Early Detection Research Network/National Cancer Institute, Dr. Ziding Feng, did the analysis of those data. The values reported in the article are those provided by Dr. Feng's group.Independent validation of the value of the six biomarkers for early detection of ovarian detection is essential. Since the article's publication in Clinical Cancer Research, 287 samples obtained from the Gynecology Oncology Group, provided by Dr. Michael Birrer from National Cancer Institute, were evaluated in independent laboratory and analyzed by an independent statistician. The samples included 193 ovarian cancer and 94 healthy individuals collected under the Gynecology Oncology Group protocol 136. Ninety-three samples were from patients diagnosed with stages I and II ovarian cancer and 100 were from patients with stages III and IV. The final results of that study showed a 96.8% specificity and 95.8% sensitivity. The healthy group samples, according to the information provided by Dr Birrer, was obtained from healthy donors randomly selected from samples collected by different institutions in the United States.(6).

We investigated the molecular effect of the epidermal growth factor receptor (EGFR) inhibitors erlotinib and gefitinib in vivo on all available tumors from patients treated on North American Brain Tumor Consortium trials 01-03 and 00-01 for recurrent or progressive malignant glioma. EGFR expression and signaling during treatment with erlotinib or gefitinib were analyzed by Western blot and compared with pre-erlotinib/gefitinib-exposed tissue or unexposed controls. Tumors were also analyzed for EGFR mutations and for other genomic abnormalities by array-based comparative genomic hybridization. Clinical data were used to associate molecular features with tumor sensitivity to erlotinib or gefitinib. Erlotinib and gefitinib did not markedly affect EGFR activity in vivo. No lung signature mutations of EGFR exons 18 to 21 were observed. There was no clear association between erlotinib/gefitinib sensitivity and deletion or amplification events on array-based comparative genomic hybridization analysis, although novel genomic changes were identified. As erlotinib and gefitinib were generally ineffective at markedly inhibiting EGFR phosphorylation in these tumors, other assays may be needed to detect molecular effects. Additionally, the mechanism of erlotinib/gefitinib sensitivity likely differs between brain and lung tumors. Finally, novel genomic changes, including deletions of chromosomes 6, 21, and 22, represent new targets for further research.

To evaluate prognostic and predictive molecular biomarkers in early-stage non-small cell lung carcinoma (NSCLC) receiving neoadjuvant chemotherapy. The IFCT-0002 trial compared two neoadjuvant regimens in 528 stages I to II NSCLC patients. DNA extraction of snap-frozen surgical samples taken from 208 patients receiving gemcitabine-cisplatin or paclitaxel-carboplatin regimens allowed for the identification of 3p allelic imbalance, Ras association domain family 1A (RASSF1A) and death-associated protein kinase 1 (DAPK1) promoter methylation, and epidermal growth factor receptor, K-ras, and TP53 mutations. Multivariate analysis identified prognostic and predictive effects of molecular alterations. A Bootstrapping approach was used to assess stability of the prognostic models generating optimism corrected indexes. RASSF1A methylation correlated significantly with shorter disease-free survival (DFS; adjusted HR = 1.88, 95% CI: 1.25-2.82, P = 0.0048) and shorter median overall survival (OS; adjusted HR = 2.01, 95% CI: 1.26-3.20, P = 0.020). A computed bootstrap resampling strategy led to a prognostic model, including RASSF1A, DAPK1, and tumor stage, dividing patients into three prognostic groups, with median OS ranging from 34 months for high-risk patients (HR for death = 3.85, 95% CI: 1.79-6.40) to more than 84 months for moderate (HR = 1.85, 95% CI: 0.97-3.52) and low-risk patients (reference group; P = 0.00044). In addition, RASSF1A methylation predicted longer DFS in patients treated with paclitaxel-carboplatin compared with gemcitabine-cisplatin (adjusted HR = 0.47, 95% CI: 0.23-0.97, P(interaction) = 0.042). Following neoadjuvant chemotherapy, RASSF1A methylation negatively impacted prognosis of early-stage NSCLC. Along with DAPK1 methylation and tumor stage, RASSF1A methylation allowed definition of three subgroups with strikingly different prognosis. Conversely, significantly longer DFS following paclitaxel-based neoadjuvant chemotherapy for patients whose tumors showed RASSF1A methylation suggested its predictive interest in stages I and II NSCLC.

Seneca Valley Virus (SVV-001) is a novel naturally occurring replication-competent picornavirus with potent and selective tropism for neuroendocrine cancer cell types, including small cell lung cancer. We conducted a first-in-human, first-in-class phase I clinical trial of this agent in patients with cancers with neuroendocrine features, including small cell lung cancer. Clinical evaluation of single intravenous doses in patients with cancers with neuroendocrine features was performed across five log-increments from 10(7) to 10(11) vp/kg. Toxicity, viral titers and clearance, neutralizing antibody development, and tumor response were assessed. A total of 30 patients were treated with SVV-001, including six with small cell carcinoma at the lowest dose of 10(7) vp/kg. SVV-001 was well tolerated, with no dose-limiting toxicities observed in any dose cohort. Viral clearance was documented in all subjects and correlated temporally with development of antiviral antibodies. Evidence of in vivo intratumoral viral replication was observed among patients with small cell carcinoma, with peak viral titers estimated to be >10(3)-fold higher than the administered dose. One patient with previously progressive chemorefractory small cell lung cancer remained progression-free for 10 months after SVV-001 administration, and is alive over 3 years after treatment. Intravenous SVV-001 administration in patients is well tolerated at doses up to 10(11) vp/kg, with predictable viral clearance kinetics, intratumoral viral replication, and evidence of antitumor activity in patients with small cell lung cancer. Phase II clinical evaluation in small cell lung cancer is warranted, and has been initiated.

There are currently no validated factors predictive of response to taxanes in patients with breast cancer. We analyzed specimens from patients included in the Breast Cancer International Research Group (BCIRG) 001 trial, a randomized study which showed the superiority of docetaxel/doxorubicin/cyclophosphamide over fluorouracil/doxorubicin/cyclophosphamide as adjuvant therapy for node-positive operable breast cancer in terms of disease-free survival (DFS) and overall survival (OS). Immunohistochemical assessment of biological markers included histologic grade, tumor size, estrogen and progesterone receptors, lymph node status, HER2, MUC1, Ki-67/MIB-1, p53, Bcl-2, Bax, Bcl-XL, BAG-1, beta-tubulin isotypes II, III and IV, tau protein, and detyrosinated alpha tubulin. Associations between selected parameters and survival were tested through univariate analyses, then completed with multivariate analyses and a bootstrap resampling technique. In univariate analysis histologic grade, tumor size, number of involved nodes, estrogen and progesterone receptor status, p53, Ki-67, tubulin III, and tau protein were associated both with DFS and with OS. In multivariate analysis estrogen and progesterone receptors, tumor size, number of involved nodes, and Ki-67 protein were associated both with DFS and with OS, whereas tau protein levels were correlated with DFS and tubulin III and P53 were correlated with OS. No interaction was observed between Ki-67 and treatment allocation. We conclude that the expression in primary tumors of Ki-67 and p53 protein, as well as of the microtubule-related parameters tau protein and tubulin III, are independent prognostic factors in patients receiving adjuvant chemotherapy for node-positive breast cancer but are not predictive of benefit from docetaxel-containing adjuvant chemotherapy.

A phase I trial of intrathecal Spartaject Busulfan (SuperGen, Inc., San Ramon, CA) was conducted in children with neoplastic meningitis following recurrent primary brain tumors to describe toxicities, estimate the maximum tolerated dose (MTD), and document evidence of responses to this agent. The continuous reassessment method was used to assign cohorts of patients to doses of intrathecal Spartaject Busulfan via an Ommaya reservoir and/or lumbar puncture twice weekly for 2 weeks followed by an assessment of toxicity and response. Patients with stable disease or an objective response continued to receive intrathecal Spartaject Busulfan plus systemic chemotherapy at regular intervals. Cerebrospinal fluid and blood were obtained for pharmacokinetic studies in patients with Ommaya reservoirs after the first dose of intrathecal Spartaject Busulfan. Seven evaluable patients were assigned to the starting dose of 5 mg, two patients to 7.5 mg, three patients to 10 mg, seven patients to 13 mg, and four patients to 17 mg. Between September 2000 and May 2003, 28 patients were enrolled in this study. Twenty-three patients (median age, 8.8 years; range, 2.5-19.5 years) were evaluable for estimating the MTD, and dose-limiting toxicities were observed in three and included grade 3 vomiting (n = 1 at 5 mg), grade 3 headache (n = 1 at 17 mg), and grade 3 arachnoiditis (n = 1 at 17 mg). Pharmacokinetic data showed that post-infusion concentrations of busulfan ranged from 50 to 150 microg/mL and declined to <1 microg/mL within 5 hours. Intrathecal Spartaject Busulfan was well tolerated in children with neoplastic meningitis from brain tumors, and the recommended dose for future phase II studies is 13 mg.

Purpose: Endosialin (TEM-1, CD248) is a protein expressed on the surface of activated mesenchymal cells, including certain subsets of tumors. Preclinical models suppressing endosialin function have shown antitumor activity. A humanized monoclonal antibody, MORAb-004, was engineered to target endosialin and is the first agent in clinical development for this mesenchymal cell target. Experimental design: This first-in-human, open-label, phase I study recruited patients with treatment-refractory solid tumors. MORAb-004 was administered intravenously once weekly in 4-week cycles. Objectives included determination of the safety of multiple infusions of MORAb-004, identification of the maximum tolerated dose (MTD), pharmacokinetic modeling, detection of any anti-human antibody response, and assessment of objective radiographic response to therapy. Results: Thirty-six patients were treated at 10 dose levels of MORAb-004, ranging from 0.0625 to 16 mg/kg. Drug-related adverse events were primarily grade 1-2 infusion toxicities. Dose-limiting toxicity of grade 3 vomiting was observed at 16 mg/kg. Eighteen of 32 evaluable patients across all doses achieved disease stability, with minor radiographic responses observed in 4 patients (pancreatic neuroendocrine, hepatocellular, and sarcoma tumor types). Pharmacokinetics showed MORAb-004 accumulation beginning at 4 mg/kg and saturable elimination beginning at 0.25 mg/kg. Exposure increased in a greater-than-dose-proportional manner with terminal half-life increasing proportionally with dose. The MTD was identified as 12 mg/kg. Conclusions: Preliminary antitumor activity was observed. Safety profile, pharmacokinetics, and early antitumor activity suggest that MORAb-004 is safe at doses up to 12 mg/kg and should be studied further for efficacy.

Checkpoint kinase 1 (Chk1) plays a critical role in the activation of mitotic spindle checkpoint and DNA damage checkpoint. We examined the preclinical use of the Chk1 inhibitor PF-00477736 as a docetaxel-sensitizing agent. Specifically, we investigated the correlation between PF-00477736-mediated modulation of biomarkers and the sensitization of docetaxel efficacy. In vitro and in vivo studies using COLO205 and other cell lines were done to assess PF-00477736-induced enhancement of docetaxel efficacy and effects on associated biomarkers. PF-00477736 significantly enhanced the docetaxel-induced efficacy in tumor cells and xenografts. Docetaxel induced dose- and time-dependent increase in the levels of phosphorylated Chk1 (Ser(345)), phosphorylated histone H3 (Ser(10)), and gammaH2AX foci and promoted the cytoplasmic localization of phosphorylated Cdc25C (Ser(216)). PF-00477736 cotreatment suppressed docetaxel-induced changes in phosphorylated histone H3 and cytoplasmic phosphorylated Cdc25C (Ser(216)) levels and concurrently sensitized the docetaxel-induced apoptosis. Docetaxel alone or in combination with PF-00477736 induced significant antiproliferative activity in xenografts, shown via [18F]FLT-PET imaging. However, changes in [18F]FLT uptake did not reflect the potentiation of docetaxel efficacy. In contrast, bioluminescence imaging showed that PF-00477736 sensitized docetaxel-induced suppression of tumor survival. Docetaxel triggers mitotic spindle checkpoint activation at low concentrations and activates both the DNA damage checkpoint and the spindle checkpoint at high concentrations. In combination with docetaxel, PF-00477736 abrogates the mitotic checkpoint, as well as the DNA damage checkpoint, and results in sensitization to docetaxel. Chk1 inhibitor PF-00477736 offers a therapeutic potential for the enhancement of taxane therapy.

In the current study, we examine the effects of a novel proteasome inhibitor, NPI-0052 (salinosporamide A), on proteasome function and nuclear factor-kappaB activation and evaluate its ability to enhance treatment response in colon cancer xenografts when administered orally. The effects of treatment on nuclear factor-kappaB activation, cell cycle regulation, and apoptosis were determined. The pharmacodynamic effect of NPI-0052 on 20S proteasome function was assayed in vivo following oral and i.v. drug administration and compared with treatment with bortezomib. The effect of combined treatment with chemotherapy was determined in a colon cancer xenograft model. We found that NPI-0052 is a potent, well-tolerated proteasome inhibitor that has pharmacodynamic properties distinct from bortezomib in that it achieves significantly higher and more sustained levels of proteasome inhibition. When combined with chemotherapy, NPI-0052 increases apoptosis and shifts cells toward G2 cell cycle arrest. When added to chemotherapy in vivo [using combinations of 5-fluorouracil (5-FU), CPT-11, Avastin (bevacizumab), leucovorin, and oxaliplatin], NPI-0052 significantly improved the tumoricidal response and resulted in a 1.8-fold increased response to CPT-11, 5-FU, and leucovorin triple-drug combination (P=0.0002, t test), a 1.5-fold increased response to the oxaliplatin, 5-FU, and leucovorin triple-drug combination (P=0.013, t test), and a 2.3-fold greater response to the CPT-11, 5-FU, leucovorin, and Avastin regimen (P=0.00057). The high level of proteasome inhibition achieved by NPI-0052 is well tolerated and significantly improves the tumoricidal response to multidrug treatment in a colon cancer xenograft model. Further evaluation of this novel proteasome inhibitor in clinical trials is indicated.

Purpose: Carfilzomib, a selective proteasome inhibitor, has shown safety and efficacy in relapsed and/or refractory multiple myeloma. This phase I study in patients with relapsed or progressive multiple myeloma assessed the safety and tolerability of escalating doses of carfilzomib in combination with lenalidomide and low-dose dexamethasone (CRd) to identify the dose for a phase II expansion study. Experimental design: Patients with multiple myeloma who relapsed after 1 to 3 prior regimens enrolled into dose-escalation cohorts. CRd was administered on 28-day dosing cycles: carfilzomib 15 to 27 mg/m(2) on days 1, 2, 8, 9, 15, and 16; lenalidomide 10 to 25 mg on days 1 to 21; and dexamethasone 40 mg weekly. Results: Forty patients enrolled in six cohorts. Prior treatment included bortezomib (75%) and lenalidomide (70%); 20% and 36% were refractory overall. The maximum tolerated dose was not identified, and the highest dose combination tested was recommended for the phase II study. The most common toxicities of any grade were fatigue (62.5%), neutropenia (55.5%), and diarrhea (52.5%). Grade 3/4 toxicities included neutropenia (42.5%), thrombocytopenia (32.5%), and lymphopenia (27.5%), with no grade 3/4 neuropathy reported. Proteasome inhibition 1-hour after dose was more than 80% in cycles 1 and 2. Among all patients, the overall response rate was 62.5%, the clinical benefit response rate was 75.0%, and the median duration of response and progression-free survival were 11.8 and 10.2 months, respectively. Conclusion: The maximum planned CRd dose, carfilzomib 27 mg/m(2), lenalidomide 25 mg, and dexamethasone 40 mg, was recommended for further study, with promising safety and efficacy.

The oncogenic PI3K/Akt/mTOR pathway is an attractive therapeutic target in cancer. However, it is unknown whether the pathway blockade required for tumor growth inhibition is clinically achievable. Therefore, we conducted pharmacodynamic studies with GDC-0068, an ATP competitive, selective Akt1/2/3 inhibitor, in preclinical models and in patients treated with this compound. We used a reverse phase protein array (RPPA) platform to identify a biomarker set indicative of Akt inhibition in cell lines and human-tumor xenografts, and correlated the degree of pathway inhibition with antitumor activity. Akt pathway activity was measured using this biomarker set in pre and post-dose tumor biopsies from patients treated with GDC-0068 in the dose escalation clinical trial. The set of biomarkers of Akt inhibition is composed of 10 phosphoproteins, including Akt and PRAS40, and is modulated in a dose-dependent fashion, both in vitro and in vivo. In human-tumor xenografts, this dose-dependency significantly correlated with tumor growth inhibition. Tumor biopsies from patients treated with GDC-0068 at clinically achievable doses attained a degree of biomarker inhibition that correlated with tumor growth inhibition in preclinical models. In these clinical samples, compensatory feedback activation of ERK and HER3 was observed, consistent with preclinical observations. This study identified a set of biomarkers of Akt inhibition that can be used in the clinical setting to assess target engagement. Here, it was used to show that robust Akt inhibition in tumors from patients treated with GDC-0068 is achievable, supporting the clinical development of this compound in defined patient populations.

Purpose: We describe the preclinical pharmacology and antitumor activity of GDC-0068, a novel highly selective ATP-competitive pan-Akt inhibitor currently in clinical trials for the treatment of human cancers. Experimental design: The effect of GDC-0068 on Akt signaling was characterized using specific biomarkers of the Akt pathway, and response to GDC-0068 was evaluated in human cancer cell lines and xenograft models with various genetic backgrounds, either as a single agent or in combination with chemotherapeutic agents. Results: GDC-0068 blocked Akt signaling both in cultured human cancer cell lines and in tumor xenograft models as evidenced by dose-dependent decrease in phosphorylation of downstream targets. Inhibition of Akt activity by GDC-0068 resulted in blockade of cell-cycle progression and reduced viability of cancer cell lines. Markers of Akt activation, including high-basal phospho-Akt levels, PTEN loss, and PIK3CA kinase domain mutations, correlate with sensitivity to GDC-0068. Isogenic PTEN knockout also sensitized MCF10A cells to GDC-0068. In multiple tumor xenograft models, oral administration of GDC-0068 resulted in antitumor activity ranging from tumor growth delay to regression. Consistent with the role of Akt in a survival pathway, GDC-0068 also enhanced antitumor activity of classic chemotherapeutic agents. Conclusions: GDC-0068 is a highly selective, orally bioavailable Akt kinase inhibitor that shows pharmacodynamic inhibition of Akt signaling and robust antitumor activity in human cancer cells in vitro and in vivo. Our preclinical data provide a strong mechanistic rationale to evaluate GDC-0068 in cancers with activated Akt signaling.

To compare the preclinical and clinical pharmacokinetic properties of paclitaxel formulated as a Cremophor-free, albumin-bound nanoparticle (ABI-007) and formulated in Cremophor-ethanol (Taxol). ABI-007 and Taxol were given i.v. to Harlan Sprague-Dawley male rats to determine pharmacokinetic and drug disposition. Paclitaxel pharmacokinetic properties also were assessed in 27 patients with advanced solid tumors who were randomly assigned to treatment with ABI-007 (260 mg/m(2), 30 minutes; n = 14) or Taxol (175 mg/m(2), 3 hours; n = 13), with cycles repeated every 3 weeks. The volume of distribution at steady state and clearance for paclitaxel formulated as Cremophor-free nanoparticle ABI-007 were significantly greater than those for paclitaxel formulated with Cremophor (Taxol) in rats. Fecal excretion was the main elimination pathway with both formulations. Consistent with the preclinical data, paclitaxel clearance and volume of distribution were significantly higher for ABI-007 than for Taxol in humans [21.13 versus 14.76 L/h/m(2) (P = 0.048) and 663.8 versus 433.4 L/m(2) (P = 0.040), respectively]. Paclitaxel formulated as ABI-007 differs from paclitaxel formulated as Taxol, with a higher plasma clearance and a larger volume of distribution. This finding is consistent with the absence of paclitaxel-sequestering Cremophor micelles after administration of ABI-007. This unique property of ABI-007 could be important for its therapeutic effectiveness.

ABI-007 is a novel Cremophor-free, protein-stabilized, nanoparticle formulation of paclitaxel. The absence of Cremophor EL may permit ABI-007 to be administered without the premedications used routinely for the prevention of hypersensitivity reactions. Furthermore, this novel formulation permits a higher paclitaxel concentration in solution and, thus, a decreased infusion volume and time. This Phase I study examines the toxicity profile, maximum tolerated dose (MTD), and pharmacokinetics of ABI-007. ABI-007 was administered in the outpatient setting, as a 30-min infusion without premedications. Doses of ABI-007 ranged from 135 (level 0) to 375 mg/m2 (level 3). Sixteen patients participated in pharmacokinetic studies. Nineteen patients were treated. No acute hypersensitivity reactions were observed during the infusion period. Hematological toxicity was mild and not cumulative. Dose-limiting toxicity, which occurred in 3 of 6 patients treated at level 3 (375 mg/m2), consisted of sensory neuropathy (3 patients), stomatitis (2 patients), and superficial keratopathy (2 patients). The MTD was thus determined to be 300 mg/m2 (level 2). Pharmacokinetic analyses revealed paclitaxel C(max) and area under the curve(inf) values to increase linearly over the ABI-007 dose range of 135-300 mg/m2. C(max) and area under the curve(inf) values for individual patients correlated well with toxicity. ABI-007 offers several features of clinical interest, including rapid infusion rate, absence of requirement for premedication, and a high paclitaxel MTD. Our results provide support for Phase II trials to determine the antitumor activity of this drug.

ABI-007, an albumin-bound, 130-nm particle form of paclitaxel, was developed to avoid Cremophor/ethanol-associated toxicities in Cremophor-based paclitaxel (Taxol) and to exploit albumin receptor-mediated endothelial transport. We studied the antitumor activity, intratumoral paclitaxel accumulation, and endothelial transport for ABI-007 and Cremophor-based paclitaxel. Antitumor activity and mortality were assessed in nude mice bearing human tumor xenografts [lung (H522), breast (MX-1), ovarian (SK-OV-3), prostate (PC-3), and colon (HT29)] treated with ABI-007 or Cremophor-based paclitaxel. Intratumoral paclitaxel concentrations (MX-1-tumored mice) were compared for radiolabeled ABI-007 and Cremophor-based paclitaxel. In vitro endothelial transcytosis and Cremophor inhibition of paclitaxel binding to cells and albumin was compared for ABI-007 and Cremophor-based paclitaxel. Both ABI-007 and Cremophor-based paclitaxel caused tumor regression and prolonged survival; the order of sensitivity was lung > breast congruent with ovary > prostate > colon. The LD(50) and maximum tolerated dose for ABI-007 and Cremophor-based paclitaxel were 47 and 30 mg/kg/d and 30 and 13.4 mg/kg/d, respectively. At equitoxic dose, the ABI-007-treated groups showed more complete regressions, longer time to recurrence, longer doubling time, and prolonged survival. At equal dose, tumor paclitaxel area under the curve was 33% higher for ABI-007 versus Cremophor-based paclitaxel, indicating more effective intratumoral accumulation of ABI-007. Endothelial binding and transcytosis of paclitaxel were markedly higher for ABI-007 versus Cremophor-based paclitaxel, and this difference was abrogated by a known inhibitor of endothelial gp60 receptor/caveolar transport. In addition, Cremophor was found to inhibit binding of paclitaxel to endothelial cells and albumin. Enhanced endothelial cell binding and transcytosis for ABI-007 and inhibition by Cremophor in Cremophor-based paclitaxel may account in part for the greater efficacy and intratumor delivery of ABI-007.

MORAb-009 is a chimeric monoclonal antibody that targets mesothelin, a tumor differentiation antigen overexpressed in pancreatic cancer, ovarian cancer, mesothelioma, and other malignancies. We conducted a phase I clinical trial of MORAb-009 in patients with advanced mesothelin-expressing cancers to determine its safety, dose-limiting toxicity (DLT), and maximum tolerated dose (MTD). Methods: Cohorts consisting of 3 to 6 subjects each received MORAb-009 intravenously on days 1, 8, 15, and 22 at progressively increasing doses ranging from 12.5 to 400 mg/m(2). Disease evaluation with computed tomography occurred on day 35. Subjects with responding or stable disease could receive additional cycles of MORAb-009. A total of 24 subjects were treated including 13 mesothelioma, 7 pancreatic cancer, and 4 ovarian cancer patients. The median number of MORAb-009 infusions was 4 (range 1-24 infusions). At the 400 mg/m(2) dose level, 2 subjects experienced DLT (grade 4 transaminitis and a grade 3 serum sickness). Thus, although there were other contributing causes of these adverse events, 200 mg/m(2) was considered the MTD. Other adverse events at least possibly related to MORAb-009 included 7 drug hypersensitivity events (all grade 1 or 2) and a thromboembolic event (grade 4). Eleven subjects had stable disease. There was a dose-dependent increase in serum MORAb-009 concentration. MORAb-009 is well tolerated and the MTD when administered weekly is conservatively set at 200 mg/m(2). In this group of previously treated patients, 11 subjects had stable disease. Phase II studies of MORAb-009 in different mesothelin-expressing cancers are ongoing.

Antineoplastic agents often achieve antitumor activity at the expense of close to unacceptable toxicity. One potential avenue to improve therapeutic index might combine agents targeting distinct components of the same growth regulatory pathway. This might lead to more complete modulation of the target pathway at concentrations lower than those associated with limiting adventitious toxicities from either agent alone. The protein kinase antagonist UCN-01 is currently used in Phase I/II trials and has recently been demonstrated to inhibit potently PDK1. We have recently documented that the alkylphospholipid perifosine potently also inhibits Akt kinase (PKB) activation by interfering with membrane localization of Akt. This leads to the hypothesis that these two agents might act synergistically through distinct mechanisms in the PI3K/Akt proliferation and survival-related signaling pathway. The synergistic effects of UCN-01 and perifosine, on two cell lines (A-549 and PC-3), were examined using various long-term in vitro assays for cell growth, cell cycle distribution, clonogenicity, survival morphology, and apoptosis. Along with Western blotting experiments were performed to determine whether this synergistic combination of two drugs has significant effect on their downstream targets and on biochemical markers of apoptosis. After 72 h, perifosine at concentrations of 1.5 and 10 microM UCN-01 at 40 and 250 nM did not significantly affect the growth of PC-3 and A459 cells, respectively. However, in combination at the same respective individual concentrations (1.5 microM and 40 nM of perifosine and UCN-01, respectively, in PC-3 cells and 10 microM perifosine and 0.25 microM UCN-01 in the somewhat more resistant A549 cells), virtually complete growth inhibition of both the cell lines resulted. Supra-additive inhibition of growth was also demonstrated in independent clonogenic assays. Mechanistic studies in cell culture models suggest enhanced depletion of the S-phase population in cells treated by the combination. This correlated with enhanced inactivation of Akt along with activation of caspases 3 and 9 and poly(ADP-ribose) polymerase cleavage. Evidence of synergy was formally demonstrated and occurred across a wide range of drug concentrations and was largely independent of the order or sequence of drug addition. As the concentrations of UCN-01 and perifosine causing synergistic inhibition of cell growth are clinically achievable without prominent toxicity, these data support the development of clinical studies with this combination.

UCN-01, a protein kinase C/cyclin-dependent kinase inhibitor, suppressed thymidylate synthase (TS) protein expression in a dose-dependent manner with near complete suppression at 1 microM after a 24-h exposure in human gastric cancer cell line SK-GT5. Other protein kinase C/cyclin-dependent kinase inhibitors, including flavopiridol and safingol, had a similar effect on TS protein expression, but to a lesser degree. Moreover, UCN-01 repressed the induction of TS after 5-fluorouracil (FU) exposure by 90-95% and significantly enhanced the induction of apoptosis by FU from 4-8% with either FU or UCN-01 alone to 46+/-1% (P < 0.005 versus either single drug, reverse sequence, or the combination) when UCN-01 was given after FU. The effect of UCN-01 on TS was associated with a dose-dependent suppression of the E2F-1 protein, a transcriptional activator of TS. Northern blot analysis revealed that TS mRNA levels decreased gradually as the concentration of UCN-01 increased, but that E2F-1 mRNA levels remained relatively unchanged. UCN-01 may provide a novel way to enhance cellular sensitivity toward FU by means of suppressing TS expression mediated mainly by down-regulation of E2F-1.

The purpose of this study was to characterize interactions between the farnesyltransferase inhibitor L744832 and the checkpoint abrogator UCN-01 in drug-sensitive and drug-resistant human myeloma cell lines and primary CD138+ multiple myeloma cells. Wild-type and drug-resistant myeloma cell lines were exposed to UCN-01 +/- L744832 for 24 hours, after which mitochondrial injury, caspase activation, apoptosis, and various perturbations in signaling and survival pathways were monitored. Simultaneous exposure of myeloma cells to marginally toxic concentrations of L744832 and UCN-01 resulted in a synergistic induction of mitochondrial damage, caspase activation, and apoptosis, associated with activation of p34cdc2 and c-Jun-NH2-kinase and inactivation of extracellular signal-regulated kinase, Akt, GSK-3, p70(S6K), and signal transducers and activators of transcription 3 (STAT3). Enhanced lethality for the combination was also observed in primary CD138+ myeloma cells, but not in their CD138- counterparts. L744832/UCN-01-mediated lethality was not attenuated by conventional resistance mechanisms to cytotoxic drugs (e.g., melphalan or dexamethasone), addition of exogenous interleukin-6 or insulin-like growth factor-I, or the presence of stromal cells. In contrast, enforced activation of STAT3 significantly protected myeloma cells from L744832/UCN-01-induced apoptosis. Coadministration of the farnesyltransferase inhibitor L744832 promotes UCN-01-induced apoptosis in human multiple myeloma cells through a process that may involve perturbations in various survival signaling pathways, including extracellular signal-regulated kinase, Akt, and STAT3, and through a process capable of circumventing conventional modes of myeloma cell resistance, including growth factor- and stromal cell-related mechanisms. They also raise the possibility that combined treatment with farnesyltransferase inhibitors and UCN-01 could represent a novel therapeutic strategy in multiple myeloma.

The transcription factor Wilms tumor protein 1 (WT1) belongs to a new generation of tumor antigens, as it is essential for tumor cell proliferation and is highly expressed in various hematologic and solid malignancies. The aim of this study was to apply a modified reverse immunology strategy to identify immunogenic epitopes of WT1 which could be useful for immunotherapy. Potential HLA-A*01 epitopes predicted by a MHC binding algorithm were screened for recognition by peripheral blood mononuclear cells (PBMC) from patients with spontaneous T cell responses using intracellular cytokine cytometry. Epitope processing was shown by proteasomal cleavage. Epitope-specific T cells were generated from CD4+CD25+ regulatory T cell-depleted PBMC. One of five predicted HLA-A*01-binding candidate epitopes showed high immunogenicity as 5 of 14 patients with hematologic malignancies had WT1.317-327-reactive T cells ranging from 0.4% to 1.5% of CD3+CD8+ T cells. Proteasomal degradation assays indicated the cleavage of WT1.317-327. The depletion of regulatory T cells from PBMCs enabled the rapid expansion of WT1.317-327-specific CTL, whereas no CTL could be generated from unfractionated PBMC. WT1.317-327-specific CTL efficiently lysed an autologous WT1-expressing tumor cell line but not HLA-A*01-negative WT1-expressing tumor cells. Immunogenicity of the epitope across histologies was verified by the demonstration of spontaneous ex vivo WT1.317-327-specific T cell responses in two of six patients with HLA-A*01-positive melanoma or lung cancer. In this study, a modified reverse immunology strategy was employed to identify a first immunogenic HLA-A*01-restricted T cell epitope of the tumor antigen WT1, which is of considerable interest for use in vaccination trials.

7-Hydroxystaurosporine (UCN-01), a protein kinase inhibitor in clinical development, demonstrates potent antineoplastic activity. To determine whether specific genetic abnormalities would modulate the response to UCN-01, a model of human non-small cell lung carcinoma (NSCLC) cell lines with differential abnormalities of p16CDKN2, RB, and p53 was used for these studies. Cell growth was measured by the 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide assay, and cell cycling was studied using flow cytometric analysis of DNA content. Changes in protein levels and phosphorylation were assessed by Western blotting. In cell lines expressing wild-type RB (A549 and Calul), UCN-01 treatment resulted in dose-dependent growth inhibition, arrest of cells in G1, and a reduction of cells in S phase. p16CDKN2-null cells showed similar growth inhibition to normal fetal lung fibroblasts. UCN-01-induced growth arrest was accompanied by induction of p21CDKN1 and a shift of Rb to the hypophosphorylated state in both p53 wild-type and mutant cell lines. In contrast, UCN-01 treatment of the RB-null cell line H596 resulted in less growth inhibition. To test the role of RB in response to UCN-01, effects of treatment were examined in two human isogenic models of RB expression: the bladder cancer cell line 5637 (RB-null) and the prostate cancer cell line DU-145 (RB-mutant). In the Rb-expressing 5637 subline (RB5), UCN-01 treatment resulted in Rb hypophosphorylation and an accumulation in G1 in contrast to the parent line. Similarly, the wild-type Rb-expressing DU-145 sublines (DU1.1 and B5) showed increased G1 arrest compared with the parent cells. We conclude that UCN-01-induced G1 arrest can occur in cells null for p53 and p16CDKN2, and that RB status influences the ability of UCN-01 to induce a G1 arrest. These data suggest that the molecular profile of cell cycle regulating genes in individual tumors may predict responsiveness and provide insight into optimal therapeutic application of this new antineoplastic agent.

E2F-1 regulates the transcription of genes required for DNA synthesis. Previously, we have reported that UCN-01 suppresses E2F-1 protein expression without any noticeable effect on its mRNA level in gastric cancer cell line SK-GT5 (Clin. Cancer Res., 4: 2201-2206, 1998). In this study, we investigated the mechanism responsible for the suppression of E2F-1 expression by UCN-01 in SK-GT5 cells. After 24-h exposure to 1 microM UCN-01, E2F-1 protein expression was decreased by >99%. The suppressive effect of UCN-01 could be reversed by ubiquitin-dependent proteasome inhibitors such as calpain inhibitor I and lactacystin. Transfection experiments using expression plasmids encoding full-length E2F-1 or truncated E2F-1 with deletion of the COOH-terminal region (which is required for eliciting ubiquitination and protein degradation) revealed that the expression of truncated E2F-1 was not affected by UCN-01. Other cell-cycle-related and ubiquitin-proteasome-regulated proteins such as p21, p27, and cyclin B1 were not repressed by UCN-01 in E2F-1-overexpressing cells. In vitro-translated, full-length E2F-1 degraded more rapidly upon incubation with extracts from UCN-01-treated cells when compared with truncated E2F-1. Taken together, these data indicate that UCN-01 suppresses E2F-1 protein expression mediated by the ubiquitin-proteasome pathway in a specific manner.

E75 is an immunogenic peptide from the HER2/neu protein, which is overexpressed in many breast cancer patients. We have conducted two overlapping E75 vaccine trials to prevent recurrence in node-positive (NP) and node-negative (NN) breast cancer patients. E75 (HER2/neu 369-377) + granulocyte macrophage colony-stimulating factor was given intradermally to previously treated, disease-free NP breast cancer patients in a dose escalation safety trial and to NN breast cancer patients in a dose optimization study. Local and systemic toxicity was monitored. Immunologic responses were assessed using in vitro assays and in vivo delayed-type hypersensitivity responses. Clinical recurrences were documented. One hundred and eighty-six patients were enrolled in the two studies (NP, 95; NN, 91). Human leucocyte antigen A2 (HLA-A2) and HLA-A3 patients were vaccinated (n = 101), whereas all others (n = 85) were followed prospectively as controls. Toxicities were minimal, and a dose-dependent immunologic response to the vaccine was shown. Planned primary analysis revealed a recurrence rate of 5.6% in vaccinated patients compared with 14.2% in the controls (P = 0.04) at a median of 20 months follow-up. As vaccine-specific immunity waned over time, the difference in recurrence lost significance at 26 months median follow-up (8.3% versus 14.8%); however, a significant difference in the pattern of recurrence persisted. E75 is safe and effective in raising a dose-dependent HER2/neu immunity in HLA-A2 and HLA-A3 NP and NN breast cancer patients. More importantly, E75 may reduce recurrences in disease-free, conventionally treated, high-risk breast cancer patients. These findings warrant a prospective, randomized phase III trial of the E75 vaccine with periodic booster to prevent breast cancer recurrences.

Altered and deregulated cyclin-dependent kinase (cdk) activity is now believed to play a major role in the pathogenesis of head and neck squamous cell carcinomas (HNSCC), thus providing a suitable cellular target for therapeutic intervention. UCN-01 (7-hydroxy-staurosporine), a known protein kinase C and cdk modulator, demonstrates antiproliferative and antitumor properties in many experimental tumor models and may represent a potential candidate to test in HNSCC. In this study, UCN-01 displayed potent antiproliferative properties (IC50 of approximately 17-80 nM) in HNSCC cells. Cell cycle analysis revealed that UCN-01 treatment of HNSCC cells for 24 h leads to a G1 block with a concomitant loss of cells in S and G2-M and the emerging sub-G1 cell population, confirmed to be apoptotic by terminal deoxynucleotidyl transferase-mediated nick end labeling analysis. Additional in vitro studies demonstrated a G1 arrest that was preceded by depletion in cyclin D3, elevation of p21(WAF1) and p27(KIP1) leading to a loss in activity of G1 cdks (cdk2, cdk4), and reduction in pRb phosphorylation. Antitumor properties of UCN-01 were also assessed in vivo by treating HN12 xenografts (7.5 mg/kg/i.p./daily) with UCN-01 for 5 consecutive days. Total sustained abolition of tumor growth (P < 0.00001) was obtained with only one cycle of UCN-01 treatment. Terminal deoxynucleotidyl transferase-mediated nick end labeling staining of xenograft samples revealed a higher incidence of apoptosis in treated tissues when compared with control. Additional tissue analysis demonstrated that elevated p27(KIP1) with minimal increase in p21(WAF1) and reduced cyclin D3 levels were readily detected in those animals treated with UCN-01, similar to those observed in HNSCC cells. Thus, UCN-01 exhibits both in vitro and in vivo antitumor properties in HNSCC models, and these effects are associated with a decrease in cyclin D3 and an increase in p27(KIP1) protein levels, thus providing appropriate surrogate markers to follow treatment efficacy in vivo and, therefore, a suitable drug candidate for treating HNSCC patients.