Elizabeth Eisenhauer

5088 Background: The gemcitabine plus carboplatin combination (GC) is known to provide a clinical advantage compared to single-agent C for patients (pts) with recurrent PSOC. Dose reductions and delays for toxicity are common during standard GC therapy. To evaluate the potential impact of G and C dose adjustments on clinical outcomes in pts with PSOC, we retrospectively analyzed dose intensity (DI) data from the phase III trial AGO-OVAR 2.5. Methods: Data analyses were performed for 175 GC-treated pts. G and C weekly DIs were calculated as the mean dose of drug received divided by the planned weekly dose for each pt through each cycle of treatment. Progression-free survival (PFS) and response rate (RR) were compared across G DI groups for pts who received 6 cycles using the log-rank test and Fisher’s exact test, respectively To reduce the bias of time dependence, we examined subsequent PFS after cycle 2 stratified by G DI within the first 2 cycles (in pts who received >2 cycles). Results: A total of 99 pts (56.6%) completed 6 cycles. The mean weekly G DI ranged from 89.5% at cycle 1 to 73.0% at cycle 6; the mean weekly C DI ranged from 99.5% at cycle 1 to 91.5% at cycle 6. Results for pts receiving either 6 cycles or >2 cycles are shown in the table. There were no significant differences across G DI groups for either PFS (P=0.199) or RR (P=0.537) for pts receiving 6 cycles or for subsequent PFS (P=0.249) when pts were grouped by G DI within the first 2 cycles. Comparisons of G and C DI based on the numbers of cycles completed (up to 8) also showed no significant differences in PFS or RR. Conclusions: Differences in DI of both G and C for pts receiving from 2 to 8 cycles of standard GC chemotherapy had no impact on either PFS or RR in the AGO-OVAR 2.5 trial. Protocol-specified dose adjustments do not appear to impact the efficacy of GC therapy for pts with PSOC.

3090 Background: NCIC CTG conducted a phase II trial of the mTOR inhibitor TEM in endometrial cancer since this tumor is associated with frequent PTEN loss. Our earlier work did not show a relationship between response (or progression) and PTEN or other molecular markers. Because of evidence that toxicity may relate to efficacy outcomes for EGFR inhibitors (Wacker, Clin Cancer Res 2007; Jonker, NEJM 2007), we explored the correlation between toxicity and outcomes in our trial. Methods: Response results (RECIST) in 60 treated patients (pts) (33 chemonaive) were 5 PR, 32 SD, 17 PD, and 6 IN. All pts receiving > 1 cycle of TEM (n = 49) were included. Prior chemotherapy and worst cycle 1 toxicity (see Table) were evaluated as potential predictors for partial response (PR) or progression (PD), as well as maximum % decrease from baseline tumor size (TS) in a univariate analysis (UA). Multivariate analysis (MVA) assessed UA variables significant at p<0.1. Results: No significant predictor was found for PR. Only prior chemotherapy was a significant, independent predictor by MVA for PD and TS: the adjusted odds ratio (OR) for PD was 7.37 (95% CI 1.32-41.03, p=0.02) and difference in TS was 24.4 (95% CI 3.94-44.8, p=0.03) between pts with/without prior chemotherapy. Conclusions: Pts with prior chemotherapy were more likely to have early PD and increase of tumor size. Similar to the TEM renal cell cancer study (deSouza, ASCO 2008) no relationship between efficacy outcomes and TEM-induced adverse events was demonstrated in this trial. Variable UA (p value) MVA adjusted OR (95%CI, p value) PR PD TS PR PD TS mean difference Categorical variables ≥ Gr 2 ANC 0.22 0.31 0.41 — — — ≥ Gr 2 mucositis 0.18 0.68 0.26 — — ≥ Gr 2 diarrhea 0.57 1.00 0.86 — — — Prior chemotherapy (Y/N) 0.15 0.003 0.02 — 7.37 (1.32 to 41.03, p=0.02) 24.4 (3.94 to 44.8, p=0.03) ≥ Gr 2 fasting triglycerides 0.40 1.00 0.28 — — — ≥ Gr 2 fasting glucose 0.08 0.64 0.08 — — −24.7 (−51.3 to 1.9, p = 0.08) ≥ Gr 2 fasting cholesterol 1.00 0.46 0.46 — — Continuous variables Highest triglyceride 0.85 0.17 0.11 — — — Highest cholesterol 0.99 0.06 0.45 — 0.76 (0.47–1.24, p = 0.27) — Highest glucose level 0.57 0.57 0.15 — — — Lowest ANC 0.22 0.04 0.13 — 1.56 (0.69–3.51, p = 0.28) — Author Disclosure Employment or Leadership Position Consultant or Advisory Role Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration Wyeth

5018 Background: There are no validated molecular biomarkers in EOC which correlate with prognosis or chemosensitivity. Breast cancer 1 (BRCA1) protein inactivation in sporadic EOC is common and reduced BRCA1 expression has been linked with platinum sensitivity. However, BRCA1 has not been clinically validated as a prognostic or predictive marker in EOC. Methods: The NCIC CTG-led Gynecologic Cancer Intergroup phase III trial, OV.16, compared topotecan/cisplatin followed by carboplatin/paclitaxel to standard carboplatin/paclitaxel in advanced EOC patients. With a median follow-up of 62 months (mo), there was no improvement in progression free survival (PFS) in the topotecan arm (Hoskins P, ASCO 2008). BRCA1 protein expression was determined by immunohistochemistry (IHC) on tissue microarrays of tumour samples from 251 Canadian patients (pts). IHC score was correlated with response rate (RR) and PFS using Cox proportional modeling adjusting for three independent prognostic factors identified in the main study (FIGO stage, residual disease (RD), and ECOG performance status). Preplanned subgroup analysis was performed for pts with minimal RD (microscopic or < 1 cm after debulking). Results: For the whole study group, BRCA1 score analyzed as a continuous variable showed no significant correlation between BRCA1 protein expression and PFS (adjusted HR = 1.15 [0.96, 1.37]; p = 0.12) or RR (HR = 0.89 [0.70, 1.12]; p = 0.32). In 116 pts with minimal RD, higher BRCA1 expression analyzed as a continuous variable was significantly associated with shorter PFS (HR = 1.40 [1.04, 1.89]; p = 0.03), but no significant correlation was found with RR. Exploratory analysis suggested that patients with minimal RD could be divided into low (BRCA1 ≤ 2.5) and high (BRCA1 > 2.5) expression groups. Pts with low BRCA1 expression had a longer PFS compared to those with high expression (median 24.7 and 16.6 mo, respectively; HR = 0.56 [0.35, 0.89]; p = 0.01). Conclusions: This study suggests that BRCA1 protein is a prognostic marker in sporadic EOC patients with minimal RD. Further research is needed to evaluate BRCA1 as a predictive biomarker and to target BRCA1 expression to enhance chemotherapeutic sensitivity.

3074 Background: The commonly used Fleming (FL) and Gehan phase II designs make recommendations on activity or early termination based on response rate (RR). The MN design, by incorporating early progression (PD) and RR, has been shown to perform more efficiently than FL or Gehan as assessed by appropriate early stopping for inactivity in phase II trials of cytotoxic agents (Dent S, 2001). Because RR is a controversial endpoint for targeted agent trials, we compared the performance of MN versus FL designs in a series of NCIC CTG phase II trials. Methods: We applied two MN stopping rules (MNA and MNB) to 15 trials evaluating single agent targeted therapies. All trials were conducted using a FL design (Ho: proportion responding p_r≤5% vs. Ha: p_r≥20%). The hypotheses were as follows: for MNA: Ho: p_r≤5% OR proportion with PD p_p≥60% vs. Ha p_r≥20% and p_p≤40%; for MNB: Ho: p_r≤5% AND p_p≥60% vs. Ha p_r≥20% or p_p≤40%. The trial would stop at stage 1 by MNA if either the response OR the PD criterion for stopping were met and by MNB only if both the response AND the PD criteria for stopping were met. Sequential patient data were reviewed in each trial to tabulate a stop/go decision at the end of stage 1 based on FL, MNA and MNB and also to determine if each agent was considered active by that design at end of study. Results: FL stopped 8/15 trials: of 7 trials completing stage 2, only 2 met FL criteria for activity. The MNA rule would have stopped 12/15 trials, including 2 that the same design would have accepted as active based on the final data. FL and MNA disagreed on early stopping in 4/15 trials, all of which were eventually rejected by FL. MNB did not recommend early stopping in any trial. Conclusions: Using the PD hypotheses noted: MNA, which considered both RR and PD in assessing activity, stopped trials early more often than FL. MNB, which required only one of RR or PD to assess activity, never stopped a trial early. We plan to repeat this exercise using additional PD hypotheses to determine levels that would change MNB performance. No significant financial relationships to disclose.

We read the article by Trotta et al. [1] with interest and agree that reporting of randomized controlled trials (RCTs) before meeting protocol-defined criteria for conducting a final analysis is an important and complex issue. However, we differ from some of their conclusions.

Background: We evaluated the activity and toxic effects of bortezomib in patients with mantle cell lymphoma. Patients and methods: Thirty patients, including 29 eligible patients, were enrolled; 13 had received no prior chemotherapy. The dose of bortezomib was 1.3 mg/m² given on days 1, 4, 8 and 11 every 21 days. Response was assessed according to the International Workshop Criteria for non-Hodgkin's lymphoma and toxicity graded using the National Cancer Institute Common Toxicity Criteria version 2.0. Results: There were 13 responding patients (46.4%; 95% confidence interval = 27.5% to 66.1%), including one unconfirmed complete remission. The median response duration was 10 months. Response rates were similar in previously untreated (46.2%) and treated (46.7%) patients. Neurological toxicity and myalgia led to treatment discontinuation in 10 patients after two to seven treatment cycles. Five serious adverse events (including two deaths) associated with fluid retention were observed in the first 12 patients. We subsequently excluded patients with baseline effusions, dyspnea or edema; no further events were seen. Conclusions: Bortezomib is active in treating patients with mantle cell lymphoma. While cumulative neuromuscular toxic effects limited therapy duration and specific issues related to fluid retention require further evaluation, continued study of this drug in combination regimens is warranted.

Purpose: A multi-centre phase II study of SarCNU—a novel chloroethylnitrosourea (CNU)—in patients with recurrent malignant glioma to assess response rate, survival and effects of treatment. Patients and methods: Ten patients with histologically proven malignant glioma (seven with glioblastoma multiforme (GBM) and three with anaplastic astrocytoma) received SarCNU (860 mg/m²) orally on days 1, 5 and 9 on a 6 week schedule. Results: A total of ten patients were treated on protocol before accrual was suspended for a high rate of pulmonary toxicity. Of eight evaluable patients, five demonstrated at least one grade deterioration in DLCO from baseline. This necessitated premature closure of the trial. Stable disease was seen in five of seven evaluable patients (median duration 4.8 months; range 0.8–9.2) with progressive disease in the remainder. Conclusion: Despite promising preclinical data, SarCNU caused pulmonary toxicity in patients with recurrent malignant glioma and we plan no further studies in this indication.