Comparative Effectiveness of Crizotinib Among ALK+ NSCLC Patients Across The United States, Western Europe, and Japan

VA L U E I N H E A LT H 1 9 ( 2 0 1 6 ) A 3 4 7 – A 7 6 6

median PFS and OS were 11 and 12 months, respectively. Survival analysis was not possible in patients treated with vemurafenib due to the inmaturity of data. Most common adverse events were rash and fever. Two patients had to discontinue vemurafenib due to toxicity (renal failure and peripheral neuropathy).  Conclusions: These results should be interpreted with caution due to the limited sample size. However, dabrafenib ± trametinib appears to show lower benefit than the expected taking into account the results from clinical trials. No conclusions can be obtained about the effectiveness of vemurafenib. In general, BRAFi were well tolerated. PCN16 Reweighting Rct Evidence To Better Reflect Real Life: A Case Study of The Innovation Medicines Initiative Happich M1, Brnabic A2, Faries D3, Abrams KR4, Winfree K3, Girvan A3, Jonsson P5, Johnston J3, Belger M6 1Lilly Deutschland GmbH, Bad Homburg, Germany, 2Eli Lilly, Sydney, Australia, 3Eli Lilly, Indianapolis, IN, USA, 4University of Leicester, Leicester, UK, 5National Institute for Health and Care Excellence (NICE), Manchester, UK, 6Eli Lilly, Ascot, UK On behalf of the Innovative Medicines Initiative GetReal consortium (WP1)

Objectives: Although the demonstration of improved patient and clinical outcomes within randomized controlled trials (RCTs) is widely accepted as foundational evidence of the efficacy of new treatments, concerns are frequently expressed that RCTs lack external validity. GetReal, a project under the umbrella of the Innovation in Medicine Initiative, is exploring how “real-life” clinical data can be brought in earlier in drug development.  Methods: We describe a case study that considers lung cancer the most common cancer worldwide. It investigates the generalizability of efficacy (overall survival [OS]) from the pivotal trial of pemetrexed vs gemcitabine use for the treatment of non-squamous NSCLC (Scagliotti et al 2008), using real-world data from the prospective observational FRAME study (Moro-Sibilot 2015) in a reweighting approach. Both inverse propensity scoring and entropy balancing were used to reweight RCT data based on real-world data to attempt to mirror routine clinical practice in the trial setting.  Results: Although OS differences between pemetrexed and gemcitabine appear more pronounced after reweighting, the reweighted analysis of the clinical trial yielded a hazard ratio (HR) closer to 1, with greater uncertainty: HR of 0.92 (95% CI: 0.60 to 1.33) compared with 0.81 (95% CI: 0.70 to 0.94) in a similar population in the clinical trial. Sensitivity analyses to both the methods of reweighting and the inclusion of baseline covariates gave broadly similar results.  Conclusions: < span”> The key objective of this case study was to assess the generalizability of RCT results for the treatment of non-squamous NSCLC when projected to a real-world population. Tested reweighting efforts did not seem to invalidate findings from the original RCT. Scagliotti et al J Clin Oncol. 2008 Jul 20;26(21):3543-51 Moro-Sibilot 2015 Lung Cancer. 2015 Dec;90(3):427-32

PCN18 Adjusted Comparison of Daratumumab Monotherapy Versus RealWorld Historical Control Data From The Czech Republic in Heavily Pre-Treated and Highly Refractory Multiple Myeloma Patients Diels J1, Gatopoulou X2, Besson H1, Vesela S3, Hájek R4, Jarkovsky J5, Ito T6 Health Economics & Market Access EMEA Statistics & Modelling, Beerse, Belgium, 2Janssen Health Economics & Market Access EMEA, Athens, Greece, 3Market Access Janssen – Cilag s.r.o., Czech Republic, Prague, Czech Republic, 4University Hospital Ostrava and Faculty of Medicine, Ostrava, Czech Republic, 5Masaryk University, Brno, Czech Republic, 6Janssen Health Economics & Market Access EMEA, High Wycombe, UK

1Janssen

Objectives: To perform an adjusted comparison of overall survival (OS) for daratumumab monotherapy versus physician’s choice [PC]) as observed in a real-world historical control cohort from Czech Republic, using patient-level data.  Methods: Patient-level data were pooled from daratumumab monotherapy clinical studies (patients treated with 16 mg/kg in the SIRIUS/GEN501 studies) and from the Czech Registry of monoclonal gammopathies (RMG), representing treatments observed in real-world cohort collected of MM-patients with at least 2 prior therapy lines and/or double refractory to a PI and an IMiD from 18 centres. For the Czech cohort, longitudinal follow-up in subsequent treatment lines was available for patients in 3rd (n= 206), 4th (n= 256), 5th (n= 203), and 6th+ (n= 307) lines; individual patients could contribute information to the analysis for multiple lines of therapy, with baseline defined as the date of initiation of the actual treatment line. A multivariate proportional hazards regression model was developed to compare OS between treatments, including age, gender, B2M, albumin, line of therapy, refractory status and prior pomalidomide/carfilzomib-exposure as co-variates.  Results: Patients treated with daratumumab (N= 148) and the CZ cohort (N=  463) differed in median age (64 vs. 62), median prior therapy lines (5 vs. 4), prior exposure to carfilzomib (41.2% vs 0.3%) and pomalidomide (55.4% vs 0.6%) and > = triple refractory status (64.2% vs. 5.3%). Median OS) for the DARA-patients was 20.1 [95% confidence interval: 16.6-NE] months versus 11.9 [11.2 - 13.1] months in the Czech database. The adjusted OS-HR for daratumumab versus PC was 0.35 [0.22-0.56] (versus unadjusted HR of 0.61 [0.48-0.78]). Impact of adjustment was mainly driven by refractory status and prior pomalidomide/carfilzomib-exposure.  Conclusions: This adjusted treatment comparison suggests improved OS for daratumumab compared to real-world historical control data in heavily pre-treated/refractory MM patients. Such comparisons can provide useful insights to clinicians and reimbursement-decision makers on relative treatment efficacies in the absence of head-to-head comparison studies. PCN19 Comparative Effectiveness of Crizotinib Among ALK+ NSCLC Patients Across The United States, Western Europe, and Japan DiBonaventura M1, Higginbottom K2, Meyers A3, Morimoto Y4, Ilacqua J2

A711

1Ipsos Healthcare, New York, NY, USA, 2Ipsos Healthcare, Mahwah, NJ, USA, 3Ipsos Healthcare, Washington, DC, USA, 4Ipsos Healthcare, Tokyo, Japan

Objectives: Up to 7% of patients with non-small cell lung cancer (NSCLC) have a rearrangement of the anaplastic lymphoma kinase (ALK) gene. Crizotinib has been identified as a potent ALK inhibitor and has demonstrated superior efficacy in a number of clinical studies. With its recent approval, the objective of this study was to investigate crizotinib’s real-world effectiveness among ALK+ NSCLC patients across several countries.  Methods: A multi-country retrospective medical chart-review of NSCLC patients was conducted by cancer-treating physicians in the United States (N= 7,334), 5EU (France, Germany, Italy, Spain, UK; N= 16,549), and Japan (N= 3,492) between Q2 2015 and Q1 2016. Physicians randomly selected patient charts currently on an anti-cancer regimen and abstracted data on patient demographics, disease status, treatment patterns, and biomarker status. Only patients who were ALK+ were included in the analyses. Treatment response between those who had used crizotinib in their prior treatment line and control patients who had used a non-crizotinib treatment in their prior treatment line was examined using a generalized ordered logit model controlling for baseline differences.  Results: A total of 1,471 patients were ALK+ and were included (crizotinib: N= 336 vs. controls: N= 1,135). Compared with controls, patients using crizotinib were more likely to come from Germany and Spain, to be younger, and to be passive/non-smokers (all p< .05). Patients using crizotinib were also more likely to have been diagnosed in stage IIIb/IV and have a lower ECOG score (both p< .05). Controlling for baseline differences, patients using crizotinib were significantly less likely to experience recurrence/progression (odds ratio (OR) =  0.38, 95%CI: 0.24, 0.59; p< .05) and significantly more likely to experience a complete response (OR= 2.65, 95%CI: 1.69, 4.15; p< .05).  Conclusions: As more ALK inhibitor treatments become available, it will be increasingly important to demonstrate the real-world effectiveness of these treatments. The results suggest a significantly better response for ALK+ patients using crizotinib. PCN20 Sunitinib Dosing Schedules in The Management of Metastatic Renal Cell Carcinoma: A Meta-Analysis Abogunrin S1, Ashaye AO2, Fahrbach K2, Cappelleri JC3, Sandin R4, Ramaswamy K5 Ltd, London, UK, 2Evidera Inc., Lexington, MA, USA, 3Pfizer Inc, Groton, CT, USA, 4Pfizer AB, Stockholm, Sweden, 5Pfizer, Inc., New York, NY, USA 1Evidera

Objectives: Metastatic renal cell carcinoma is traditionally managed with a four-week-on, two-week-off (4/2) sunitinib (SU) dosing schedule. To reduce toxicity associated with the 4/2-schedule, alternative schedules like the two-week-on, one-week-off (2/1) schedule, have been considered. We conducted a meta-analysis to estimate the comparative effectiveness and safety of alternative and traditional SU dosing schedules.  Methods: Articles were identified from a published systematic literature review (Guida et al. 2014), and complemented by a targeted search in MEDLINE (conducted in December 2015) and by bibliographic searches. We included randomized controlled trials (RCTs) and comparative observational studies that evaluated a 4/2-schedule, 2/1-schedule or a switch from 4/2- to 2/1-schedule. Fixedand random-effects Bayesian network meta-analyses were conducted to estimate relative effects for overall survival (OS), progression-free survival (PFS), progressive disease, and CTCAE grade 3+ adverse events (AEs).  Results: Seven of 13 studies identified (1 RCT and 6 observational) met the inclusion criteria. Six studies evaluated 4/2- and 4/2-to-2/1-schedules and four studies evaluated the 2/1-schedule. Relative to the 4/2-schedule, the 2/1-schedule was associated with better PFS, OS, and higher rates of response. For this comparison, no Bayesian estimates were statistically different, but a classical meta-analysis for PFS limited to the 4/2-schedule vs. the 2/1-schedule showed significance due to the homogeneity in those effects (HR: 1.35 [95% CI: 1.03, 1.79]). PFS was statistically lower in patients on the 4/2-schedule vs. the 4/2-to-2/1-schedule (HR: 2.30 [95% CrI: 1.07, 4.99]) and the odds of grade 3+ diarrhoea (OR: 5.64 [95% CrI: 1.9, 20.13]) and fatigue (OR: 4.67 [95% CrI: 1.88, 13.17]) were statistically higher. There was no statistical difference for other AEs. There was insufficient evidence to address observed heterogeneity.  Cclusions: Our findings indicate that the alternative SU dosing schedules may have better effectiveness and safety profiles than the 4/2-schedule. The results should be interpreted with caution given data limitations.

PCN21 Nintedanib Plus Docetaxel as Second-Line Therapy in Patients with Non-Small-Cell Lung Cancer (NSCLC): A Network Meta-Analysis vs. New Therapeutic Options Popat S1, Mellemgaard A2, Reck M3, Hastedt C4, Griebsch I4 1Royal Marsden Hospital, London, UK, 2Herlev Hospital, Herlev, Denmark, 3LungenClinic Grosshansdorf, Grosshansdorf, Germany, 4Boehringer Ingelheim GmbH, Ingelheim, Germany

Objectives: The relative efficacy of nintedanib + docetaxel versus other secondline agents approved in patients with adenocarcinoma histology NSCLC was evaluated using network meta-analysis (NMA) in 2015. Here we provide an update including new therapeutic options.  Methods: A systematic literature review found 3 new trials which fit to the previously published inclusion criteria or present subgroup results for the population of interest. The NMA of progression-free survival (PFS) and overall survival (OS) was updated following the previously published methods.  Results: Nintedanib + docetaxel was superior vs. docetaxel, erlotinib and gefitinib (Nintedanib + docetaxel vs. docetaxel: OS HR= 0.83 (95% CI: 0.70–0.99), PFS HR= 0.77 (0.6–0.96); vs. pemetrexed: OS HR= 0.82 (0.60–1.11), PFS HR= 0.84 (0.61– 1.15); vs. erlotinib: OS HR= 0.64 (0.46–0.90), PFS HR= 0.70 (0.50–0.99); vs. gefitinib: OS HR= 0.59 (0.36–0.96), PFS HR= 0.45 (0.28–0.72)). Results comparing nintedanib vs. ramucirumab and nivolumab confirmed overall similarity. However, hazard ratios vs. nivolumab pointed into a direction in favour of nivolumab for OS (HR= 1.20 (0.92– 1.58)) but in favour of nintedanib for PFS (HR= 0.91 (0.68–1.21)). Because significant interaction p-values indicate that PD-L1 expression is a relevant effect modifier