Adaptive Decision Making Based on Interval-Censored Data in a Clinical Trial to Optimize Rapid Treatment of Stroke - Interval-Censored-Time-to-Event Data: Methods and Applications - page 329

12.1 Introduction

Acute ischemic stroke (AIS) is a major cause of disability and death in adults (John-

son et al., 2009). Intra-arterial (IA) fibrinolytic infusion, a new treatment for AIS,
delivers a thrombolytic agent to dissolve the clot that caused the stroke via two

telescoping catheters. The catheters are inserted into the femoral artery and moved
through the carotid artery to the site in the brain artery of the clot that caused
the AIS. Tissue plasminogen activator (tPA) is a thrombolytic agent approved by

the FDA for intravenous (IV) treatment of AIS. While the behavior of IV tPA in
adult stroke patients is well understood, IA tPA is still an experimental regime.

This chapter describes a new design (Thall et al., 2011) that aims to optimize IA
administration of tPA for treating AIS.

The treatment regime is as follows. A concentration c; in mg/kg body weight, of
tPA is given using a xed maximum volume V: A proportion q is given as an initial

bolus of size qV: If the clot is not dissolved immediately, the bolus is followed by
continuous infusion (ci) of the remaining amount (1 q)V at a constant rate for a
maximum of 120 minutes. Ecacy is the time Y E to dissolve the clot (response). If
the bolus dissolves the clot, then Y E = 0; otherwise, the ci is begun and response
is evaluated at 15-minute intervals, so only the time interval in which Y E occurs is
known. The ci is stopped if and when response is observed within 120 minutes. Thus,
Y E is interval-censored from 0 to 120 minutes and administratively right-censored
at 120 minutes. For example, if a patient's clot was not dissolved by the 60-minute
evaluation but was found to have dissolved at 75 minutes, then it is only known that
60 < YT E 75. Toxicity is the binary indicator, YT T ; of symptomatic intracerebral
hemorrhage (SICH), characterized by neurological worsening compared to baseline,
evaluated by brain imaging at 48 hours. SICH is associated with high morbidity

rates and a death rate of approximately 50%. Because the ci is stopped when the
clot is dissolved, the amount of agent that a patient actually receives depends on

Y E as well as c and q: In the above numerical example, using standardized time s
= t=120 and V = 1; the actual amount of tPA given would be Y E c = 0.625c: Thus,
the probability of SICH depends on c; q; and Y E :