Reference no: EM133836340
Assignment:
Time to event data
Read the paper by Mowery et al describing the SU2C-SARC032 trial of pembrolizumab as a treatment for soft tissue sarcoma.
(Article: Mowery et al. Safety and efficacy of pembrolizumab, radiation therapy, and surgery versus radiation therapy and surgery for stage III soft tissue sarcoma of the extremity (SU2C-SARC032): an open-label, randomised clinical trial.)
a) With reference to the primary outcome measure in the Mowery paper, explain what is meant by censoring in a trial with a time-to-event outcome measure.
b) From the Kaplan-Meier plots shown in the paper, estimate
The probability of being disease free at 12 months in the experimental group.
The time point at which 25% of patients have had the primary outcome event in the control group.
c) Explain why the Mowery paper does not present median values for any of the time-to-event outcomes assessed.
d) Compare the observed probabilities of being disease free at 2-years to those used in the sample size calculations. Why is doing this important when interpreting the results of a trial? With reference to any results in the Mowery paper, if the observed difference is smaller than the targeted difference, why might you have doubts about adopting a new treatment even if statistical significance is seen?.
e) The Mowery paper presents disease free survival as the primary outcome measure, with overall survival as a secondary outcome measure. With reference to the Mowery paper, suggest one practical reason (i.e. consider trial resourcing, not clinical justification) why it is preferable not to use overall survival as the primary outcome measure in this trial.
f) Identify and interpret the hazard ratio (and its confidence interval) obtained from the analysis of disease-free survival in the modified ITT population.
g) Explain why a 90% confidence interval is presented for the hazard ratio of disease-free survival.
h) Explain why we might want to use a hazard ratio in addition to the log-rank p-value to summarise our data.
i) What is meant by the proportional-hazards assumption? Using any information presented in the Mowery paper, do you believe the proportional-hazards assumption is justified for the analysis of disease free survival?
j) In an analysis of a dataset from a different trial, we included a time varying covariate for treatment in our survival model in order to assess the proportional hazards assumption. This time varying covariate has a p-value of 0.001. Interpret this result, and explain whether we should use an alternative analysis method for our data.
k) In a trial with non-proportional hazards, the analysts decide to analyse their data using restricted mean survival times. Explain why it is important to pre-specify the value of t*, the point we are restricting survival times to.