Lead Investigator: Christian Jonasson, NordicRWE
Title of Proposal Research: Using Norwegian real world data to emulate and reproduce results from oncology clinical trials
Vivli Data Request: 7785
Funding Source: Research Council of Norway (grant no. 327887)
Potential Conflicts of Interest: None
Summary of the Proposed Research:
Cancer clinical trials can be slow and costly, often produce results with limited external validity, and are difficult for patients to participate in. Recent technological advances and a gradually more dynamic policy landscape in the United States and Europe have created a fertile ground for the use of real-world data (RWD) to improve current methods of clinical evidence generation. In recent years, many new efforts have emerged that try to maximize the value of data collected during clinical trials (RCT). Secondary analyses of individual trials or aggregated meta-analyses of multiple comparable trials can generate additional clinical discoveries or lead to novel hypotheses. Modern trials include patient consent to generate de-identified patient-level datasets at the trial completion and making this data available for secondary research use. Thus, easier access to RCT data, in platforms like Vivli, can be used for comparison with RWD generated from ordinary clinical practice.
The company’s goal is to develop population-based external control arms (ECA) that reflect standard of care and can be used to interpret and contextualize the results from phase II-III oncology trials and potentially, at a later stage, being able to serve as a full external control arm for single-arm oncology trials. As a first step towards this goal, and the basis for this concrete data request proposal, Vivli RCT data will be used as a proof-of-concept to emulate the trial population in completed lung cancer RCTs with Norwegian RWD and investigate if the outcomes in the RCTs can be reproduced using Norwegian RWD. If successful, this proof-of-concept study will provide evidence that Norwegian RWD is fit-for-purpose and can produce trustworthy results for regulatory decision-making.
Included in this Vivli proposal are six completed lung cancer trials from Roche/Genentech. Five studies with PD-L1 immunotherapy in both small-cell and non-small cell lung cancer (PD-L1 Immunotherapy is a therapy that boosts your immune system to help it recognize and fight cancer cells that contain high levels of the protein PD-L1, which is a protein that normally helps keep immune cells from attacking nonharmful cells in the body) and one targeted therapy trial on NSCLC ALK positive patients [ALK-positive lung cancer is a type of non-small cell lung cancer (NSCLC) in which the cancer cells have mutations in the anaplastic lymphoma kinase (ALK) gene]. The research will emulate and reproduce both the standard-of-care (placebo) arm and the experimental arm in these studies. For emulation we will use nation-wide mandatory registries in Norway, including the Cancer Registry of Norway, the Hospital registry, the Drug prescription database and the Cause of death registry. For replication we will examine different endpoints, focusing first on overall and cancer-specific survival, and in later steps also progression free survival, time to treatment discontinuation and other efficacy outcomes. Statistical balancing of baseline characteristics via propensity scores will be applied as the preferred method for emulation and the Cox proportional hazard model will be used to compare event rates between groups.
Statistical Analysis Plan:
The company’s goal is to develop population-based external control arm (ECA) that reflect standard of care and can be used to interpret and contextualize the results from selected phase II- III oncology trials and potentially, at a later stage, being able to serve as a synthetic control arm for single arm oncology trials. As a first step towards this goal, and the basis for this concrete data request proposal, Vivli data will be used to emulate the trial population in some selected lung cancer RCTs with Norwegian RWD and investigate if outcomes from RCTs can be reproduced using Norwegian data.
A range of statistical models will be applied, including sensitivity and stratified analyses and use non-parsimonious models for adjustment, which includes propensity matching and weigthing (PSM/W) and inverse probability weighting (IPW). We will, in particular, study different models performance in our situation: possibility to integrate multiple sources of data (in contrast to a few), long follow-up times of individuals with many events points (in contrast to shorter and poorer time series), robustness in the statistical conclusions in the light of multiple possible control arms.
Propensity score matching (PSM) and weighting methods is a strategy to mitigate the shortcoming in single-arm trials to develop external controls based on selection of individual external patients through well-defined eligibility criteria and statistical balancing of baseline characteristics via propensity scores using patient-level data generated in real-world clinical practice. The propensity
score is most often estimated using a logistic regression model and matching on the logit of the propensity score, as this quantity is more likely to be normally distributed. Different methods exist for forming matched pairs of treated and untreated subjects when matching on the propensity score.
We will also apply inverse probability of treatment weighting (IPTW) using weights based on the propensity score to create a synthetic sample in which the distribution of measured baseline covariates is independent of treatment assignment.
The SAP involves the analysis of each study separately. Since we are doing a PSM, PSW and IPTW of the data elements, we can analyze the data elements, looking at key factors relating to the type and amount of data separately in each platform. This analysis will give us an understanding of the common and uncommon data elements by doing a comparative analysis of the different platforms which we have analyzed removing the need to actually physically combine the data.
When substantial missing data for key endpoints or other variables are present, the pattern of missing data will be identified. Statistical modeling techniques (eg, multiple imputation chained equation) based on the missing data pattern will be applied if needed as complementary analyses to a complete-case analysis.
Requested Studies:
A Phase I/III, Randomized, Double-Blind, Placebo-Controlled Study of Carboplatin Plus Etoposide With or Without Atezolizumab (Anti-PD-L1 Antibody) in Patients With Untreated Extensive-Stage Small Cell Lung Cancer
Data Contributor: Roche
Study ID: NCT02763579
Sponsor ID: GO30081
Randomized, Multicenter, Phase III, Open-Label Study of Alectinib Versus Pemetrexed or Docetaxel in Anaplastic Lymphoma Kinase-Positive Advanced Non Small Cell Lung Cancer Patients Previously Treated With Platinum-Based Chemotherapy and Crizotinib
Data Contributor: Roche
Study ID: NCT02604342
Sponsor ID: MO29750
A Phase III, Open-Label, Randomized Study of Atezolizumab (MPDL3280A, Anti-PD-L1 Antibody) in Combination With Carboplatin+Paclitaxel With or Without Bevacizumab Compared With Carboplatin + Paclitaxel + Bevacizumab in Chemotherapy-Naïve Patients With Stage IV Non-Squamous Non-Small Cell Lung Cancer
Data Contributor: Roche
Study ID: NCT02366143
Sponsor ID: GO29436
A Phase III, Open-Label, Multicenter, Randomized Study to Investigate the Efficacy and Safety of Atezolizumab (Anti-PD-L1 Antibody) Compared With Docetaxel in Patients With Non-Small Cell Lung Cancer After Failure With Platinum Containing Chemotherapy
Data Contributor: Roche
Study ID: NCT02008227
Sponsor ID: GO28915
A Phase III, Open-Label, Multicenter, Randomized Study Evaluating the Efficacy and Safety of Atezolizumab (MPDL3280A, Anti-PD-L1 Antibody) in Combination With Carboplatin+Paclitaxel or Atezolizumab in Combination With Carboplatin+Nab-Paclitaxel Versus Carboplatin+Nab-Paclitaxel in Chemotherapy-Naive Patients With Stage IV Squamous Non-Small Cell Lung Cancer
Data Contributor: Roche
Study ID: NCT02367794
Sponsor ID: GO29437
A Phase III Multicenter, Randomized, Open-Label Study Evaluating the Efficacy and Safety of Atezolizumab (MPDL3280A, Anti-PD-L1 Antibody) in Combination With Carboplatin+Nab-Paclitaxel for Chemotherapy-Naive Patients With Stage IV Non-Squamous Non-Small Cell Lung Cancer
Data Contributor: Roche
Study ID: NCT02367781
Sponsor ID: GO29537
Summary of Results:
We’ve thoroughly explored the challenge of not being able integrating individual-level data from completed Randomized Controlled Trials (RCTs) and Real-World Data (RWD) within the same platform. Throughout this discussion, we’ve carefully weighed different other options, e.g., the potential of utilizing summary statistics and the option of using the distribution of distinct patient characteristics from RCT data to the emulation in RWD.
After meticulous evaluation, our decision is not to proceed with Vivli for our ongoing research. This choice has been influenced, in part, by the impending conclusion of our courtesy billing period on August 31. Regrettably, we find ourselves in the position of having to close the project. It’s disheartening that this research, which is both interesting and relevant, cannot be performed due to the constraints inherent in data sharing.
In the interim, it’s important to note that NordicRWE is actively exploring alternate avenues for obtaining individual-level data from RCTs.