Although I think the flat-lining of the survival curve for the early transplant group is interesting, I would be hesitant to draw too many conclusions based on it.
I think that, by the time you are getting out to overall survival times of 5 or 6 years in a relatively small study like this, the researchers have lost track of a lot of the patients. So the survival estimates for those sorts of survival times are relying on a limited number of results.
Also, I wouldn't interpret the flat line for the one group as indicating that the line will never go down. The observation period just hasn't gotten long enough to pick up the eventual drop in the line at longer survival times. People see plateaus in curves like these and think they'll go on forever. Then, an update comes out in a year or two, and lo and behold, the plateau is no longer a plateau, just a temporary flattening.
I would also mention that there is some difference between the groups in what the initial treatments were that they got. As the article mentions,
Both groups of patients had similar exposure to induction therapy with Revlimid (33 percent of the early transplant group and 28 percent of the late transplant group), but the early transplant group had more patients who had received Velcade-based therapy (63 percent versus 49 percent) and the late transplant group had more patients who had received thalidomide-based therapy (66 percent versus 38 percent)."
Some might argue that the difference between the groups in how many people got Velcade and thalidomide might have affected the results to the advantage of the early transplant group.
The study is a retrospective study, meaning that patients did not get the same upfront treatments, and we really don't know why some got early transplants and some got late transplants.
