These historical comparisons are particularly difficult here because there is a substantial delay after diagnosis and surgery before the drug is actually administered. So the question is whether their comparables are truly comparable or not - maybe they are, but you'd have to look at how they did this very carefully to be sure. Basically you want an ITT-type analysis starting at the comparable point in time.
(Imagine that a patient has surgery, and then suddenly deteriorates and they decide not to proceed with the drug. My guess is they wouldn't include that patient in their data, which means you have to exclude patients like that from the historical controls as well).