Case Study Analysis: Four Ovarian Cancer Trials
The ovarian dataset combines the data that were collected in four double-blind randomized clinical trials in advanced ovarian cancer (for details, see Section 2.2.3). Two treatments were considered, i.e., cyclophosphamide plus cisplatin (control treatment) and cyclophosphamide plus adriamycin plus cisplatin (experimental treatment). The candidate surrogate endpoint S is progression-free survival time, defined as the time (in years) from randomization to clinical progression of the disease or death. The true endpoint T is survival time, defined as the time (in years) from randomization to death of any cause. The results of the case study are discussed without reference to the software tools that can be used to conduct the analyses. Section 13.3 provides full details on how the analyses can be conducted in R.
The estimated LRF was 0.7446, with 95% confidence interval [0.7152; 0.7720]. When the censoring mechanism is taken into account, the estimated LRFa was 0.8193, with 95% confidence interval [0.7928; 0.8433]. Overall, these results indicate that the amount of uncertainty in T that is expected to be removed when the value of S becomes known is quite high, suggesting that progression-free survival time is a relatively good surrogate for survival time.
Ovarian Cancer Data. Plot of the estimated LRFa (and their 95% confidence intervals) per cluster.
Figure 10.1 provides the estimated LRFa and their 95% confidence intervals in each of the clusters. As can be seen, the point estimates for LRFa were similar in most clusters (units) and they were typically above 0.65. In cluster 59, the point estimate of the LRFa was only 0.0639, though its 95% confidence interval overlapped with those of many other clusters. This indicates that the individual-level association between progression-free survival and overall survival is of the same magnitude across the different clusters.