Observational Registries Applications in Safety Evaluation 8 th
Observational Registries: Applications in Safety Evaluation 8 th FDA/Industry Statistics Workshop Washington, D. C. September 14 -16, 2005 Wei Dong MD Ph. D, Katie Miller MSc, Pavel Napalkov MD MPh Epidemiology Genentech, Inc.
Acknowledgements • Patients, physicians and others participating in the studies • Investigators at Genentech and many other institutions
Today’s Objectives • Overview of registries – Opportunities and challenges • Examples of safety evaluations using registries @ Genentech
Registries vs. RCT: Generalizability Target Population: All Patients with Disease X Older, multiple comorbid diseases Younger, multiple comorbid diseases Older, otherwise healthy Younger, otherwise healthy Study Population Registry Sampling Phase III Sampling (few or no exclusion criteria) (multiple exclusion criteria)
Registries vs. RCT: Scope of Study Less 1, 000 s 10, 000 s Amount of data collected More years Registries # of patients enrolled Duration of followup RCT 10 s-100 s months
Two Types of Registries Enrollment base Disease-based Patients with specific disease Product-based Patients treated with specific product Active treatment Yes Control Group Yes No Risk Estimates Absolute risk Relative risk Absolute risk
Registries at Genentech
NRMI: National Register of Myocardial Infarction • Multi-center, observational cohort study • Eligibility: disease-based – Patients hospitalized for acute myocardial infarction – Treated at physician’s discretion • Initiated in 1990 • >2 million patients • >1600 hospitals
NRMI: Intracranial Hemorrhage (ICH) 1. 2. Gurwitz J. Risk for intracranial hemorrhage after tissue plasminogen activator treatment for acute myocardial infarction. Ann Intern Med 1998; 129: 597 -604. [Data thru 09/1996] Gore JM. Stroke after thrombolysis. Mortality and functional outcomes in the GUSTO-1 trial. Circulation 1995; 92: 2811 -8.
Stroke Studies: ICH Risk and t. PA • To evaluate the risk of ICH in special populations • Pooled analysis of 4 prospective cohorts of Alteplase treated patients – Standard Treatment With Alteplase To Reverse Stroke [STARS]1 – Epidemiology Study Of Ischemic Stroke [ESIS] – University Of Texas Houston Stroke Study [UT] – Canadian Activase For Stroke Effectiveness Study [CASES]2 • Systematically collected stroke treatment and outcomes • Symptomatic ICH was ascertained per head CT scan or MRI and a decline in neurological status – within 72 hours for STARS/ESIS, or 24 hours for UT and CASES 1. 2. Albers GW. et al. Intravenous tissue-type plasminogen activator for treatment of acute stroke: the Standard Treatment with Alteplase to Reverse Stroke (STARS) study. JAMA. 283(9): 1145 -50, 2000. Hill MD. Buchan AM. Methodology for the Canadian Activase for Stroke Effectiveness Study (CASES). CASES Investigators. Canadian Journal of Neurological Sciences. 28(3): 232 -8, 2001
% of patients Incidence Of Symptomatic ICH Among Alteplase-Treated Stroke Patients 1 1. NINDS Dong W et al. Safety Outcomes of Alteplase in Ischemic Stroke Patients with Special Characteristics. International Stroke Conference 2005.
NCGS: National Cooperative Growth Study • Multi-center, prospective, observational cohort study • Eligibility: product-based – Treated with growth hormone • Initiated in 1985 • >24, 000 patients • >400 sites in US and Canada 1. Allen D, et al. Risk of leukemia in children treated with human growth hormone: Review and analysis. J Pediatr 1997; 131: S 32 -6.
NCGS: Leukemia risk among patients with no risk factors by “person-time at risk” vs. “person-time on therapy” 1. Allen D, et al. Risk of leukemia in children treated with human growth hormone: Review and analysis. J Pediatr 1997; 131: S 32 -6. [Data thru 12/1995]
BRi. TE: Bevacizumab Regimens Investigation of Treatment Effects and Safety • Multi-center, prospective, observational cohort study • Eligibility: treatment-based – Metastatic colorectal cancer – Treated with bevacizumab (with chemo) as 1 st-line therapy • Select Study Outcomes – Safety: GI perforation – Effectiveness: overall survival • Data collection – Baseline and every 3 months for up to 3 years – No study-specific visits or evaluations
BRi. TE: GI Perforation 1. Kozloff M. et al. Safety of bevacizumab among patients receiving first-line chemotherapy for metastatic colorectal cancer- preliminary results from a large registry in the US (BRi. TE). ASCO 2005. 2. Hurwtz H et al. Bevacizumab plus Irinotecan, Fluorouracil, and Leucovorin for Metastatic Colorectal Cancer. NEJM. 2004; 350: 2335 -2342. *Bevacizumab labeling [2004] reports 2% GI perforation
EXCELS: Epidemiologic Study of Xolair (omalizumab) in Patients with Moderate to Severe Asthma • Multi-center, prospective, observational cohort study • Eligibility: disease-based – Moderate to severe persistent allergic asthma – 2: 1 Xolair-treated vs. Non-Xolair treated • Selected Safety Outcomes – All malignancies • Current status – Enrolled 3826 of planned 7500 [as of 09/2005] – 1500 pys accumulated (with <12 month f/u for most pts)
EXCELS: Approaches to Minimizing Biases • Potential for channeling bias – Collecting data on confounders • Demographics, asthma severity, history of and risk factors for cancer, etc – Statistical methods to adjust for confounders • Multivariate regression analysis adjusting for covariates • Propensity scores to assess prescribing decisions • Sensitivity analysis for residual non-measurable confounders
Discussion • Sampling – True random sample unlikely – Key is to sample a broad range of sites – “Every patient counts” – Don’t cherry pick patients • CRF – Necessary to revise over time – Try to maintain consistency in key outcome measures • Comparator cohort – Collection of confounding factors on CRF (i. e. , before SAP) – Examine potential bias (extent and direction) – Adjustments for bias • Potential for nested case-control studies
Conclusions • Observational registries are a significant component of safety (and benefit) evaluation • Given the absence of randomization, these studies need closer attentions to patient enrollment plans, data collection methods and choices of study endpoints in order to minimize potential bias.
Thank You!
Overview of Registries • Large, non-experimental cohort studies • No randomization of treatment • Minimal or no exclusion criteria in order to capture real-world patients and practices • Safety endpoints and duration of follow up selected according to disease and product MOA • Disease-based or product-based registries
ICH risk in Activase-treated Stroke Patients 1 30 -day Intracranial Hemorrhage STARS N=389 ESIS N=236 UT N=241 CASES N=1100*** 3. 3 (1. 8 -5. 6) 6. 8 (3. 9 -10. 8) 5. 8 (3. 2 -9. 6) 4. 6 (3. 4 -6. 0) Overall ICH: 4. 7% (3. 8 -5. 9) 30 -day complete recovery (i. e. , MRS≤ 1) (%) 34. 6 (29. 8 -39. 6) 40. 0 (33. 3 -47. 0) Not Recorded 32. 5** (29. 1 -36. 1) 30 -day mortality (%) 12. 9 (9. 7 -16. 6) 14. 8 14. 5* (10. 6 -20. 0) (10. 3 -19. 6) 23. 3** (20. 2 -26. 5) 1. Dong W et al. Safety Outcomes of Alteplase in Ischemic Stroke Patients with Special Characteristics. International Stroke Conference 2005. *. UT reported in-hospital mortality. **. CASES reported outcomes at 90 -day follow up. ***. Sample size and results as August. 2001.
- Slides: 22