Article Text

PDF
THU0162 Multiple Approaches for Implementation of Long-Term Efficacy Interpretation of Certolizumab Pegol Data: RAPID1 and RAPID2 CASE Study
  1. E. Keystone1,
  2. J.S. Smolen2,
  3. V. Strand3,
  4. T. Kumke4,
  5. S. Walker5,
  6. I. Mountian6,
  7. R. Landewé7
  1. 1Mount Sinai Hospital, Toronto, Canada
  2. 2Medical University of Vienna and Hietzing Hospital, Vienna, Austria
  3. 3Biopharmaceutical Consultant, Portola Valley, United States
  4. 4UCB Pharma, Monheim, Germany
  5. 5UCB Pharma, Raleigh, United States
  6. 6UCB Pharma, Brussels, Belgium
  7. 7Academic Medical Center, Amsterdam and Atrium Medical Center, Heerlen, Netherlands

Abstract

Background Use of imputed data vs observed data in long-term studies, as well as patient (pt) population evaluated (those completing randomized controlled trials [RCTs] vs all pts initiating treatment), can have considerable impact on interpretation of long-term efficacy data. In statistical analysis of data from RCTs, particularly long-term studies, consideration must be given to the impact of missing values, resulting from pt drop-out due to lack of efficacy or treatment-related adverse events. Imputation of missing data must link such reasons with assumptions made for missingness. In last observation carried forward (LOCF) and non-responder imputation (NRI), it is assumed that pts would sustain the same status inferred by the method from point of discontinuation for the entire trial duration. Both LOCF and NRI are part of the missing completely at random (MCAR) approach. In contrast to MCAR, likelihood-based models, such as mixed models with repeated measures (MMRM), are based on the assumption that missing data are missing at random (MAR). The MMRM-based imputation is frequently used and closely follows mean treatment response.

Objectives To evaluate the impact of the use of imputed vs observed efficacy data in intent-to-treat (ITT) and RCT Completer populations, using clinical trial data as a case study.

Methods Data from pooled analysis of Rheumatoid Arthritis Prevention of Structural Damage (RAPID) 1 and 2 RCTs and open-label extensions (OLEs) (NCT001523861, NCT001758772, NCT001606023 and NCT001606414) were used. RAPID11 and 23 evaluated safety and efficacy of certolizumab pegol (CZP) with methotrexate. Efficacy data were collected up to 256 weeks (wks) of CZP exposure for clinical measures, including DAS28(ESR) (LOCF), HAQ-DI (LOCF) and ACR20/50/70 (NRI). Observed and imputed data are presented for CZP Completer population (pts randomized to CZP, who completed RCT and reconsented into OLE) and CZP ITT population (all pts randomized to CZP in RCT).

Results Improvements from baseline (BL) in DAS28(ESR) and HAQ-DI were evident in CZP Completer and ITT populations at 256 wks (Table). The use of LOCF imputation gave results consistent with observed data (Table). Long-term CZP exposure resulted in sustained ACR response. Response rates determined by NRI are, per definition, lower than observed data. This particularly shows for the ITT population (Table), but is in line with data observed in blinded periods.1,3

Conclusions Multiple populations and approaches to imputation provide more reliable long-term efficacy data interpretation given inherent bias from withdrawals and imputations. Pooled RAPID1 and 2 results revealed similarities between observed and imputed data for ACR, DAS28(ESR) and HAQ-DI. Analysis of a CZP ITT population gives a more conservative estimate of efficacy compared with CZP Completers. Nonetheless, results were consistent with maintained improvements in RA signs and symptoms following 256 wks of exposure to CZP.

References

  1. Keystone E. Arthritis Rheum 2008;58:3319-3329.

  2. Keystone E. Ann Rheum Dis 2013; epub.

  3. Smolen J.S. Ann Rheum Dis 2009;68:797-804.

  4. Smolen J.S. Arthritis Rheum 2013;65:S988.

Acknowledgements The authors acknowledge Costello Medical Consulting for writing and editorial assistance which was funded by UCB Pharma.

Disclosure of Interest : E. Keystone Grant/research support: Abbott, AstraZeneca, Biotest, BMS, F. Hoffmann-La Roche, Genentech, Janssen, Lilly, Merck, Nycomed, Pfizer, UCB Pharma, Speakers bureau: Abbott, Amgen, AstraZeneca, BMS Canada, F. Hoffmann-La Roche, Janssen, Pfizer, UCB Pharma, J. Smolen Grant/research support: UCB Pharma, Consultant for: UCB Pharma, V. Strand Consultant for: UCB Pharma, T. Kumke Employee of: UCB Pharma, S. Walker Employee of: UCB Pharma, I. Mountian Employee of: UCB Pharma, R. Landewé Grant/research support: Abbott, Amgen, Centocor, Novartis, Pfizer, Roche, Schering-Plough, UCB Pharma, Wyeth, Consultant for: Abbott, Ablynx, Amgen, Astra-Zeneca, Bristol Myers Squibb, Centocor, Glaxo-Smith-Kline, Novartis, Merck, Pfizer, Roche, Schering-Plough, UCB Pharma, Wyeth, Speakers bureau: Abbott, Amgen, Bristol-Myers Squibb, Centocor, Merck, Pfizer, Roche, Schering-Plough, UCB Pharma, Wyeth

DOI 10.1136/annrheumdis-2014-eular.1813

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.