BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A Bayesian model‐free approach to combination therapy phase I tr
 ials using censored time‐to‐toxicity data - Graham Wheeler\, Cancer Re
 search UK and UCL Cancer Trials Centre @ University College London
DTSTART:20190319T191500Z
DTEND:20190319T213000Z
UID:TALK102964@talks.cam.ac.uk
CONTACT:Peter Watson
DESCRIPTION:Many treatment regimens for cancers use several different drug
 s or treatment types. For combinations of one or more novel treatments\, f
 inding a dose/therapy combination that has a controlled and acceptable ris
 k of severe toxicity presents several challenges. In particular\, for tria
 ls involving radiotherapy\, late‐onset treatment‐related toxicities ar
 e a possibility. This means patients need to be observed for a longer-than
 -usual period to assess for severe toxicities that could lead to one or mo
 re treatments being de-escalated. When a current trial patient is being ob
 served for severe toxicity and one or more new patients are recruited to t
 rial\, waiting until the current patient has completed their follow-up cou
 ld lead to an unreasonably long trial duration and excessive costs. Also\,
  this would delay the start of treatment for the patients-in-waiting.\n\nI
 n this talk\, I present an adaptation of the Product of Independent beta P
 robabilities Escalation (PIPE) design\, an approach for dual-agent phase I
  trials that has already been implemented in practice since its publicatio
 n in 2015. The adaptation uses censored time‐to‐toxicity outcomes for 
 patients still in follow-up to aid dose escalation decisions for future tr
 ial patients. I will compare the original PIPE approach to this time-to-to
 xicity adaptation\, and briefly discuss possible extensions to these metho
 ds to deal with more complex clinical trials.
LOCATION:Cambridge Institute of Public Health. Large Seminar Room (First F
 loor)\, Forvie Site\, Robinson Way. CB2 0SR
END:VEVENT
END:VCALENDAR