What we do know is they did 1) review for safety only, and then at some point 2) they did a blinded review for the paper.
Yes... and in fact, the blinded review was also specified in the protocol:
In addition to the planned interim analyses for the primary outcome, a sample size re-estimation will be performed after 80% of the 348 patients have been randomized. An independent statistician (IS) will be provided a dataset with all available primary outcomes (i.e., times to progression, death, or censoring) for the randomized cohort, without treatment group assignments. The IS will calculate the Kaplan-Meier curve for the combined randomized cohort (DCVax- L-treated and placebo treated patients). The median time to PFS (M) will be estimated from this curve. The IS will use M as the assumed median for the placebo group. The IS will then calculate a sample size assuming a median PFS of M for the placebo group, and the same hazard ratio, Type I error rate and power used in the original sample size calculation.
And it's obviously from this review that the blinded blended data came that was the basis for the JTM article from 2018.