Examination of survivor functions from SOCRATES and THALES trials for kinetics of stroke recurrence

Lead Investigator: James Brorson, The University of Chicago
Title of Proposal Research: Examination of survivor functions from SOCRATES and THALES trials for kinetics of stroke recurrence
Vivli Data Request: 6550
Funding Source: None
Potential Conflicts of Interest: None

Summary of the Proposed Research:

Study Summary for Lay Audience:
Stroke is a common and often devastating condition in which blockage of blood flow to a part of the brain leads to its destruction, with corresponding loss of function, producing a variety of symptoms such as paralysis of limbs or face, loss of speaking or comprehension, loss of peripheral vision, or loss of coordination and balance. Stroke is the leading cause of long-term disability in many societies, as well as a frequent reason for death. In the United States, some 800,000 persons suffer strokes each year, and over 100,000 die.
After an initial mild stroke, patients are at increased risk of having a second stroke. Such a recurrence can be devastating, producing severe disability. A number of clinical trials have tested medical treatments aimed at preventing these secondary strokes, and in testing these effects, have followed groups of stroke patients closely from the time of enrollment soon after the first stroke through 1 to 3 months following the stroke. Careful examination of the time course of outcomes in these groups of patients for the timing and rates of stroke recurrence can give us insight into the underlying conditions and states of patients following initial stroke. Such insights may help us devise better ways to prevent these recurrent strokes. The requested data from two recent stroke treatment trials has not been examined with this kind of kinetic analysis, and doing so will provide information about whether the timing of stroke recurrence is similar across different trials, pointing to shared underlying conditions.

The research proposed here will consist of quantitative analysis of the rates of events occurring in the subjects of the trials, both those in control groups and those in the active treatment groups. Survivor functions, representing the proportion of study subjects remaining free of an outcome event, will be estimated with a standard method called the Kaplan-Meier method. A mathematical formula derived from a theoretical kinetic model of the condition of the patients following stroke will be fitted to the observed survivor functions by adjustments of parameters in this model, using a standard mathematical approach called non-linear regression. The resulting values of the adjusted parameters will provide estimates of the kinetic rates affecting stroke recurrence in patients following initial stroke.

Study Rationale (Technical language):
Following incident stroke, stroke recurrence is front-loaded, perhaps indicating short-lived conditions of the blood vessels, heart, or brain that confer risk of recurrence of stroke. Kinetic modeling of stroke recurrence may provide insight into the number and duration of such states. Detection of short-lived states requires data on early recurrence, soon after the index event. Trials like Platelet-Oriented Inhibition in New TIA and Minor Ischemic Stroke (POINT), Clopidogrel in High-Risk Patients with Acute Non-disabling Cerebrovascular Events (CHANCE), Acute Stroke Or Transient IsChaemic Attack TReated With Aspirin or Ticagrelor and Patient OutcomES (SOCRATES) and Acute STroke or Transient IscHaemic Attack Treated With TicAgreLor and ASA for PrEvention of Stroke and Death (THALES), that enrolled and treated acute ischemic stroke or transient ischemic attack (TIA) patients in the early hours following index event occurrence, provide an accurate source of such information regarding the timing of recurrence of stroke after an initial TIA or minor stroke. Both control groups and groups treated with the active intervention can illuminate these kinetic features. We have chosen methods of quantitative analysis of this existing data to provide testing of the kinetic model’s mathematical predictions of survivor functions, and to determine the underlying kinetic parameters, as a direct approach to testing this theory of kinetic modeling.

Statistical Analysis Plan:

Summary of Statistical Analysis Plan:
Primary de-identified study data will be used to construct Kaplan-Meier survivor functions. Constructed Kaplan-Meier survivor functions, including all outcome event (failure) points but excluding censored points, weighted by surviving subject numbers, will be fit by non-linear regression. These procedures will be conducted in SAS Studio version 9.4.
Goodness-of-fit of predicted theoretical survivor functions from two alternative kinetic models, or from two different sets of kinetic parameter value assignments, will be compared using the F test. Specifically, the F statistic comparing the success of data prediction produced by alternative models will be computed from the residual sum of squares for each model, the number of data points, and the degrees of freedom for each model, and the calculated F statistic will be compared to the critical value of the F-distribution for alpha = 0.01.

In the revised project, we plan to extend our two-state kinetic model for stroke recurrence adjusting for categorical risk factors such as (a) demographics (race, age group, sex), (b) metabolic disease biomarkers (diabetes, renal disease, obesity, cholesterol levels), (c) stroke recurrence biomarkers (history of stroke recurrence, type/location of stroke, blood thinner medication). To this end, we will first extend the model by allowing the kinetic rates parameters (k0, k1, and k2) to vary across the groups defined by the individual risk factors. Then, we will fit the model adjusting for each categorical risk factor, one at a time, and then we rank the risk factors according to a goodness of fit criterion such as AIC/BIC (Aikaike’s/ Bayesian information criteria). Then we will fit a sequence of multivariate models on this sequence of submodels and select the model with the lowest information criterion. This approach will, in addition, provide us with a prediction model for stroke recurrence based on our two-state kinetic model.

Requested Studies:

Data Contributor: AstraZeneca
Study ID: NCT01994720
Sponsor ID: NCT01994720

Data Contributor: AstraZeneca
Study ID: NCT03354429
Sponsor ID: NCT03354429

Public Disclosures:

  1. James R Brorson, Shyam Prabhakaran, S Claiborne Johnston, Mihai Giurcanu. Abstract WMP61: Vulnerable And Stabilized States After Cerebral Ischemic Events: Implications Of Kinetic Modeling In The POINT, SOCRATES, And THALES Trials. Stroke. 2023;54:AWMP61. doi: 10.1161/str.54.suppl_1.WMP61
  2. Brorson, J.R., Giurcanu, M., Prabhakaran, S. and Johnston, S.C., 2023. Vulnerable and Stabilized States After Cerebral Ischemic Events: Implications of Kinetic Modeling in the SOCRATES, POINT, and THALES Trials. Neurology. Doi: 10.1212/WNL.0000000000207904