Article Text

THU0152 How should the primary endpoint be analysed in rheumatoid arthritis trials which mandate rescue prior to the clinical endpoint? Analysis of three phase 3 trials of tofacitinib
  1. R. Fleischmann1,
  2. D. Gruben2,
  3. S. Krishnaswami2,
  4. S.H. Zwillich2,
  5. B. Benda3,
  6. C.A. Mebus2,
  7. J. Bradley2
  1. 1Metroplex Clinical Research Center, Texas
  2. 2Pfizer Inc, Connecticut
  3. 3Pfizer Inc, Pennsylvania, United States


Background In the 21st century, with the availability of effective therapies for rheumatoid arthritis (RA), the conduct of double-blind placebo (PBO)-controlled studies has changed dramatically. PBO therapy is continued for shorter periods of time, especially if patients (pts) have an inadequate clinical response. In many of these trials, pts are “rescued” after a finite period of time (usually around 3 months) if they do not meet minimum response criteria, although the usual primary clinical endpoint is at month 6. The challenge is how to represent the response of pts who are treated with active medication for 3 months, but do not obtain sufficient clinical response and remain on the same medication and dose from months 3-6. Using a modified non-responder imputation method (NRI) “with advancement penalty” (AP), all pts with insufficient clinical response would be considered non-responders at month 6. Even though they continued on the same medication and dose, their response at month 6 would be “penalised” as if they had been “rescued”. That is, NRI with AP treats pts as if use of rescue led to a withdrawal of treatment with no chance of a later efficacious response on that treatment. That is a high hurdle, a useful one in drug development, but less realistic for the practising rheumatologist. The more well-known NRI (which is “without AP”) reflects responses ignoring this aspect of rescue, though it penalises pts who withdraw, making them non-responders. This might reflect a more accurate measure of effectiveness of study medication as it encompasses only tolerability and efficacy, and allows those pts who then begin active therapy after experiencing inadequate clinical response to be classified as responders under active therapy.

Methods We analysed the data from 3 tofacitinib Phase 3 RA trials and compared pts’ clinical responses, as measured by ACR20 by NRI (with and without AP), as well as with last-observation-carried-forward (LOCF). Response rates under each method were calculated for each treatment group at each visit. Group profiles were compared and contrasted.

Results In the primary analyses, tofacitinib was compared to PBO using NRI “with AP”. In comparison, the less-stringent NRI (without AP) yielded response rates that were generally greater. A Phase 3 study of tofacitinib or adalimumab versus PBO in pts with RA1 is illustrated as an example (see Table). Further, the use of NRI “without AP” as well as LOCF allows the improvement in response rate to be readily seen as pts transition from PBO to tofacitinib.

ACR20 response rate (%), month 6

Conclusions Different imputations incorporate different aspects of disease which is then reflected in the response rates: NRI “without AP” incorporates early withdrawal; NRI “with AP” adds a strong penalty analogous to use of rescue medication by imposing permanent non-response; and LOCF show results free of either aspect. Tofacitinib was consistently efficacious vs PBO using both unmodified and modified NRI analyses as well as LOCF.

  1. Arthritis Rheum 2011; 63 (Suppl 10): S153.

Disclosure of Interest R. Fleischmann Grant/Research support from: Pfizer Inc, Consultant for: Pfizer Inc, D. Gruben Shareholder of: Pfizer Inc, Employee of: Pfizer Inc, S. Krishnaswami Shareholder of: Pfizer Inc, Employee of: Pfizer Inc, S. Zwillich Shareholder of: Pfizer Inc, Employee of: Pfizer Inc, B. Benda Shareholder of: Pfizer Inc, Employee of: Pfizer Inc, C. Mebus Shareholder of: Pfizer Inc, Employee of: Pfizer Inc, J. Bradley Shareholder of: Pfizer Inc, Employee of: Pfizer Inc

Statistics from

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.