Christophe Massard

Despite considerable progress in hematological malignancies (HM) biology during the last decades, translation into clinical benefit remains a major challenge. To improve patient selection and identify patients most likely to benefit from phase I trials, we designed and validated, in an independent cohort, a simple prognostic score. Treatment outcome, toxicity, and survival data from 82 consecutive patients enrolled in 14 phase I trials were reviewed (January 2008–February 2012). We validated these results on a prospectively collected cohort (17 phase I trials, February 2012–May 2014, 88 patients). Within a median follow-up of 19.1 months (range: 2.1–43.8 months), the median progression-free and overall survival (OS) were, respectively, 4.1 months [95% confidence interval (CI): 3.0–5.3] and 19.8 months (95% CI: 16.1–36.8). Best overall response and disease control rates were similar to HM salvage regimens (28 and 64%, respectively). Through multivariate analysis of independent prognostic factors, we designed and prospectively validated a simple prognostic score based on histological subtype, performance status, and albumin. Patients with a low-risk score experienced significantly better OS compared with patients with an intermediate or a high score (median OS: 37 vs. 17 vs. 5 months; hazard ratio=11.68, 95% CI: 4.09–33.3). Our data indicate the safety and efficacy of phase I trials in a significant number of relapsed/refractory HM patients, with clinical benefit achieved in more than half of patients. Our simple scoring system offers a valuable selection tool encouraging HM patient inclusions in phase I trials.

E14025 Background: Progression-free survival ratio (PFSr) has been proposed as a direct evaluation of treatment benefit in advanced cancer patients, based on the hypothesis that the natural history of cancer is accelerating and therefore successive lines of treatments become less efficient over time. Consequently, PFS at line +2 (PFS2) is expected to be shorter than PFS at line +1 (PFS1), whereas a PFS2/PFS1 ratio > 1.3 might reflect treatment benefit. However, this hypothesis has been poorly documented, especially in the context of early phase trials where determining treatment benefit is becoming key. We therefore proposed to evaluate PFSr in a large cohort of cancer patients enrolled in early phase clinical trials at Gustave Roussy Drug Development Department. Methods: Patients enrolled in at least two phase 1 studies for advanced solid tumors or lymphomas were retrospectively identified. Demographical, clinical and therapeutic data were collected. Time to progression (PFS) was measured from treatment initiation to progression, using radiological evaluations. Patients who had gone off-trial for reasons other than progression were censored. PFSr was defined as the ratio of the PFS under line 2 divided by the PFS under line 1 and might be censored in presence of censored PFS2. Ratio distribution in this population was estimated using Kaplan Meier. Calibration of the ratio was proposed for hypothesis testing. The influence of therapeutic class and combination versus monotherapy was studied via non parametric Gehan-Wilcoxon tests or Mick tests. Results: 212 patients enrolled between 2009 and 2016 in at least two phase 1 clinical trials were included, corresponding to 113 different clinical trials. PFSr distribution was described and correlated with demographical, clinical and therapeutic data. The relevance of the usual 1.3 PFSr cut-off to determine treatment benefit was evaluated. Final analysis results will be presented at the time of the congress. Conclusions: This study, one of the first to establish PFSr in a large cohort of patients treated in early phase clinical trials, provides guidelines for using PFSr to assess the impact of treatment sequences in advanced cancer patients.

4554 Background: We conducted a national phase II multicenter trial that aimed at evaluating the efficacy and tolerance of Paclitaxel plus Ifosfamide followed by high-dose carboplatin plus etoposide treatment (TICE) in previously treated germ cell tumors. The particularity of our study (in comparison with the standard protocol [Motzer RJ, et al. J Clin Oncol 2000 Mar; 18(6): 1173-1180.]) is that the carboplatin dose was individualized for each patient according to therapeutic drug monitoring (TDM) in order to reach the target AUC of 24 mg.min/ml over 3 days. Methods: In total, 89 patients were evaluable for pharmacokinetic study. Blood samples were taken on day 1 to determine the carboplatin clearance using a Bayesian approach (NONMEM 7.2) and to adjust the dose on day 3 to reach the target AUC of 24 mg.min/ml over 3 days. On days 2 and 3, samples were taken for retrospective assessment of the actual AUC and the intra- and inter-cycle clearance variability. Secondly, a population pharmacokinetic analysis was also performed on 59 patients using NONMEM to develop a covariate equation for carboplatin clearance prediction adapted for future patients treated with the TICE protocol. The performance of this new equation was then prospectively evaluated on the other 30 patients along with different methods of carboplatin clearance prediction. Results: TDM allowed us to control the carboplatin exposure with a mean actual AUC of 24.5 mg.min/ml (22.2 and 28.0 for 5^th and 95^thpercentile respectively) per cycle. We observed a modest but significant decrease of carboplatin clearance over cycles (median value of change of -11.8% from cycle 1 to cycle 3, maximum value of -36%). The new covariate equation allows unbiased and more accurate prediction of carboplatin clearance in the prospective validation cohort compared to other equations. Conclusions: Carboplatin TDM allowed the target AUC to be accurately reached and thereby avoid over- or under-exposure. We propose a new equation to predict carboplatin clearance more adapted to these particular patients (young males) that could be used as an alternative if the TDM cannot be organized. Clinical trial information: 2008-005068-14.

3029 Background: The evaluation of patient’s life expectancy is crucial to select patients who may benefit from phase I studies. The Royal Marsden Hospital (RMH) prognostic score, based on 3 objective variables (number of metastatic sites, LDH and serum albumin) was validated in patients treated with cytotoxic and targeted therapies. We aimed to determine if those factors were applicable to immunotherapy (IT) phase I trials. Methods: A retrospective analysis for risk factors of survival was conducted in a discovery cohort of 155 patients enrolled into IT phase I trials at our institution. We computed univariate and multivariate analysis (MVA) of demographics, clinical and biological data to assess their prognostic value for overall survival (OS). MVA results were used to build a prognostic score for OS. A validation cohort of 113 patients enrolled in IT phase I trials was used to prospectively re-validate this score. Results: 155 patients receiving an experimental IT between March 2012 and January 2016 were included in the discovery cohort. A MVA assessing the RMH Score variables showed that albumin < 35 g/dL (HR 1.73, 95% CI 1.05-2.86) and LDH > upper normal limit (1.88, 95% CI 1.12-3.15) were independent negative prognostic factors for OS. As opposed to the RMH score, number of metastases was not associated with a poorer outcome for this IT cohort (0.83 95% CI 0.51-1.35). Interestingly, a neutrophil to lymphocyte ratio (N/L) > 6 (1.75, 95% CI 1.04-2.95) was associated with a worse OS. A risk analysis based on the results of the MVA showed that patients presenting a high score (2-3) had a significantly shorter OS (20.4 weeks; 95% CI 5.7-35.2) compared to those with a low score (0-1) (68.9 weeks; 95% CI 50-83.7) (HR = 2.9 95% CI 1.87-4.64). In the validation cohort of 113 patients, the patients presenting a high score showed an inferior OS (HR = 6.3, 95% CI 2.7-14.8). Conclusions: In phase I trials of IT, traditional prognostic variables included in the RMH Score are suboptimal to determine patient’s prognosis. In this context, the N/L ratio, which reflects the immune contexture, is a significant prognostic variable. The new Gustave Roussy Score, based on albumin, LDH and N/L ratio allows to better select patients for IT phase I trials.

148 Background: Molecular characterization of metastatic castration resistant prostate cancer (mCRPC) is limited by tumor tissue availability. The analysis of circulating tumor cells (CTC) offers an attractive noninvasive surrogate option to analyze molecular alterations. We report whole exome sequencing (WES) of CTCs at the single cell level in mCRPC patients. Methods: Blood samples were drawn from 11 enzalutamide or abiraterone pre-treated mCRPC patients enrolled in the clinical program MOSCATO (NCT02613962). CTC enrichment, immunofluorescent detection and single cell isolation were performed using three methods (ISET filtration, CellSearch and the VyCap puncher system and RosetteSep enrichment) to obtain pools of 1-10 CTCs with distinct epithelial or mesenchymal phenotypes. After Whole Genome Amplification (WGA), WES was performed on the Illumina HiSeq 2000 platform. GATK Haplotype Caller enabled identification of germline polymorphisms from each patient in normal DNA, metastatic sample and CTCs in order to consider WGA induced bias. The detection of sSNV in tumor biopsies and CTCs was assessed with Mutect and IndelGenotyper respectively. Results: 189 WGA of CTC pools were performed. 34 pools of phenotypically different CTCs from 7 patients were selected and sequenced. Mean coverage of 51% was obtained at a sequencing depth of 10X. Allelic drop out was lower for CTC pools containing 5-10 cells. 17/34 (50%) CTC samples had shared sSNV with the paired tumor sample (range 0.35%-68%) Epithelial CTCs had more shared sSNV with metastatic biopsies than CTCs of other phenotypes but shared sSNV were also detected in large non epithelial CTC pointing out a high level of genetic heterogeneity between CTC. Overall, 89 deleterious protein-coding mutations were found only in pools of CTC, including mutations affecting oncogenic drivers such as MAPK1, HSP90AB1 or KDM5B. Conclusions: We present single cell WES of CTCs harboring distinct phenotypes. The detection of shared sSNV between CTC pools and corresponding biopsy could validate the use of CTCs as a liquid biopsy. The finding of sSNV specific to CTCs could offer additional data on tumor heterogeneity.

463 Background: Bone metastases (BMs) are associated with significant morbidity and shorter survival in metastatic renal cell carcinoma (mRCC). Our purpose was to identify prognostic factors for mRCC patients (pts) with BMs. Methods: Data from mRCC pts with BMs, treated at Gustave Roussy between April 1992 and March 2016, were retrospectively collected. Age, sex, ECOG-Performance Status (PS), Memorial Sloan-Kettering Cancer Center (MSKCC) score, histology, number and site of BMs, concomitant metastases (presence and sites), therapy for BMs (radical resection or palliative surgery, radiotherapy, other local and systemic treatments), time to BMs, and outcome were analyzed. Synchronous solitary bone metastasis (SSBM) was defined as a single bone metastasis without concomitant visceral lesions at the initial diagnosis of RCC. Overall survival (OS) was calculated from the date of BMs diagnosis to death or last follow-up using Kaplan-Maier method and modelled with Cox-regression analysis. Results: Three hundred pts were identified. Median time to BMs was 32.4 months (range 0–324 months). In 64 pts (21%), bone was the only metastatic site and 22 of them (7%) had a SSBM; 236 pts (79%) had concomitant metastases in other sites. Median OS was 23.22 months. SSBM pts had better OS then those with concomitant metastases (40 vs. 20 months; p<0.001). At univariate analysis, number of BMs (p<0.0001), spinal column as site of BMs (p<0.005), concomitant metastases (p<0.0001), Fuhrman grade (p<0.001), non-clear cell histology (p<0.003), and MSKCC score (p1) and radical resection are prognostic factors. Additionally, in presence of solitary bone metastasis without other concomitant metastases at the initial diagnosis of RCC, bone surgery should be considered to achieve local tumor control and increase survival.

420 Background: Late relapse in patients with germ cell tumors (GCT) are rare events. As there is no consensus for follow-up (FU) protocols, it is unclear if shorter duration of CT scanning is safe while reducing unnecessary exposure to radiation. Methods: At the completion of their treatment, all patients with Stage I, metastatic good or intermediate prognosis seminoma and non-seminoma (NSGT) underwent physical exam, serum tumor markers and imaging at regular intervals. We retrospectively identified 2 groups of patients with different FU strategies: those with regular CT scan of the chest, abdomen and pelvis for 3 years or less, then followed by abdomino-pelvic ultrasound (Group 1); and those with CT scanning continued beyond 3 years (Group 2). Results: Between December 1997 and March 2016, 203 patients were included: 80 (39.4%) in Group 1 and 123 (60.6%) in Group 2. The mean duration of FU was 6.7 ± 2.4 years in Group 1 and 8.1 ± 2.5 years in Group 2 (p < 0.001). After year 3 the median number of abdomino-pelvic ultrasound in Group 1 was 5 (4-6) and the median number of CT scanning in Group 2 was 6 (5-8). Five (2.4%) relapses were identified: 1 in Group 1 and 4 in Group 2. Relapse was revealed by symptoms in 2 patients with NSGT (1 in each group) who had been lost for FU. Three patients underwent local treatment (2 surgeries for teratoma relapses and 1 radiotherapy for seminoma relapse) and are still disease-free. International prognostic factors score was low and intermediate for the first 2 patients, very low for seminoma relapse and low for teratoma relapses. Conclusions: Patient follow-up by regular abdomino-pelvic ultrasound instead of CT scan appears a feasible alternative after the third year in patients with GCT, with no excess risk of incurable relapses. Prospective validation is required.

401 Background: HDCT is a valid option of treatment for relapsed advanced GCTs pts. The results of the TICE regimen as salvage therapy in poor risk pts demonstrated a 50% complete response (CR) but with very high variability of measured carboplatin area under the curve (AUC) (between 10.9-36.7 for target AUC = 24 mg.min/mL). We initiated a phase II trial of TICE with therapeutic drug monitoring (TDM) for individual carboplatin dosing in order to target the 3-day AUC to 24 mg/min/mL. Methods: Were included pts with unfavorable relapsed GCTs , secondarily classified according to the International Prognostic Factors Study Group. Pts were treated according the TICE regimen with two cycles combining paclitaxel and ifosfamide followed by three cycles HDCT: carboplatin plus etoposide with stem cell support. Carboplatin dose was adapted on day 3 based on carboplatin clearance (Cl) at day 1, in order to reach the target AUC. The primary endpoint was the CR rate (cCR sCR pCR). A Simon Minimax design was performed using the following hypothesis: p0 = 50%, p1 = 65 α = 5%, β = 10%. Results: Between 03/2009 and 11/2015 101 pts were accrued, 60 pts were treated in first relapse (34 classified as high and very high-risk) and 41 in 2^ndor more relapse. 72 pts (71%) received the whole treatment, 12 pts (10%) did not receive any cycle of HDCT. 29pts (26%) stopped treatment earlier, 8 for toxicity. Three pts died on treatment. 35/79 (44.3%) evaluable pts achieved a CR. 19 pts (24%) presented a PRm-. The mean observed carboplatin AUC was 24.5 mg.min/mL (between 18.9 and 28.8) at C1.Following TDM, the modification of the total carboplatin dose during C1 was comprised between -33% and +44%, showing the benefit of TDM in comparison with individual dosing based only on carboplatin Cl predicted according to patient’s renal characteristics. Conclusions: The CR rate observed in this very poor prognosis population was 44.3% and 69.6% CR + PRm-. Carboplatin dose individualization based on TDM allowed to reach more accurately the target AUC compared to previous reports. Clinical trial information: NCT00864318.

246 Background: Castration-resistant prostate cancer (CRPC) is characterized by high androgen receptor (AR) expression, various AR mutations and persistent activation of AR signaling axis by residual tissue androgens and other steroids. Inhibiting both the AR and androgen biosynthesis may be more effective than inhibiting either alone to treat CRPC. ODM-204 is a potent, orally administered investigational non-steroidal dual inhibitor of CYP17A1 and the AR that has shown activity in nonclinical tumor models. Methods: We report early data of the DUALIDES phase I dose escalation part (NCT02344017), in which patients with metastatic CRPC (mCRPC) were enrolled into ODM-204 dose levels of 50, 100, 200, 300, 500 mg twice daily. Data cut-off was Oct 14^th2016. Results: In the trial 23 patients with progressive mCRPC received increasing doses of ODM-204. At baseline, the median age of patients was 70 yrs (57-84 yrs), and median PSA was 46.5 ng/mL (0.7-249.6 ng/mL). Median time on treatment was 11.6 weeks (0.3-60.4 wks). Nine (39%) patients continued treatment for more than 12 weeks, and 4 (17.4%) patients were still receiving treatment at the time of data cutoff. ODM-204 was generally well tolerated, and the most commonly reported adverse events (AEs) were fatigue (n = 6, 26%), nausea (n = 6, 26%), decreased appetite (n = 5, 22%,) diarrhea (n = 5, 22%), and vomiting (n = 5, 22%). Two patients discontinued treatment due to an AE. The AUC_t and C_max values of ODM-204 in plasma increased with dose up to 300 mg after a single dose. Steady state values on day 8 were unexpectedly decreased in most patients, especially at higher doses, contrary to what was observed in the nonclinical monkey studies. Further, decreased predose concentrations were observed on day 29. PSA decreases were seen in 7 (30%) patients, and the median decrease was 47% (2-99%). At 12 weeks, 3 (13%) patients had a PSA response ( ≥ 50% reduction from baseline). Conclusions: Although evidence of anticancer activity was provided with ODM-204 in mCRPC, decreasing steady state concentration was observed. Clinical trial information: NCT02344017.