Now if it's 0 at risk, it means *all* the KM curve for that point are made of LTFU. All of them.
You keep asserting that, and it is still wrong.
The vast majority of censors are patients who were alive when the final data point in that trial was recorded. For almost all, this happened prior to 60 months. So they are censored while known alive..
In the external trials, all but one did not last 5 years. So they had no chance to contribute any to the total at-risk at 5 years. The only trial that did report 5 year data was only 2 years after enrollment ended.
The K/M analysis is not "conservative". It is the way this type data is handled.