A big dilemma for oncologists is to know how to treat AML in patients who are above the age of 70. Many studies include patients 60 and older and there is a big difference between 60-70 and 70 and older. So what we decided to do with this project is a systematic review of the literature on old patients in the age of 70 and above. A lot of times you have studies that include patients 70 and above but they don’t publish the subgroup analysis. So we contacted the investigators and we asked them for their subgroup data for the patients 70 and older.
In the end we had 68 trials and 13,381 patients. It’s the largest study that I know of of recently treated patients with AML. The question we asked, we said, ‘OK, there are various ways of treating AML. You can have intensive chemotherapy, 3 7 plus or minus something else; you can have a low dose chemotherapy like low dose Ara-C; you can use hypomethylating agents like 5-azacitidine and decitabine or you can simply do best supportive care.’ We wanted to compare these four approaches in patients who have either favourable cytogenetics or poor cytogenetics or a good performance status or a poor performance status or no or low level of comorbidity or a high level of comorbidity.
You cannot extract that from the literature just with the meta-analysis; the literature is too heterogeneous. But we had baseline data and then we put them in a decision analysis model to adjust the results of these treatments to the proportion of patients who had cytogenetic differences or performance status differences or comorbidity differences.
The results, it’s still the early results of the model, but they seem to indicate that you really have two winners among the regimens we gave and two regimens that do not give as good an effect. The two best results are either intensive chemotherapy or hypomethylating agents. They are really neck and neck; that’s pretty much no matter what the other aspects of the leukaemia or the patients are. Our baseline scenario if a patient is in very good condition, has low comorbidity, then they tend to do better, slightly, with intensive chemotherapy no matter what their cytogenetic is. If all the factors are favourable they have about a 59% chance of being alive one year later. For all the other patients, the ones with poor performance status, the ones with some comorbidities, it seems like hypomethylating agents have a slight advantage.
These are the baseline assumptions of our model, we still have to run a full sensitivity analysis to see what percentage of the time this is true. So for the moment I will really say you have two regimens where our patients are doing better and two regimens, low dose chemotherapy and best supportive care, where really they don’t live as long. With this, especially if patients have poor function, people are gone after six months.
Is there a place, therefore, for low dose chemotherapy?
If I wanted to test the treatments in the future I would clearly prefer hypomethylating agents. They have a toxicity that is actually a better profile than low dose chemotherapy. There are some old data that show that they tried to demonstrate that you had less hospital time with low dose chemotherapy versus intensive chemotherapy and actually they had the same amount of hospital time because of infections or bleeding or things like that. So it seems like a good idea but it’s not bringing the promise it has.
If I wanted to do a new study for AML in older patients, above 70, I would choose a backbone of either an intensive regimen or a hypomethylating agent and add a new targeted drug to these. That’s what I would choose as a regimen for future use in older patients.
When can we expect to see the analysis?
I will have that done by the end of the academic year, so next ASCO probably I will have more complete results and we have started writing the article so that should be published soon. What we are going to do is once we have a model that seems stable we are going to take the Moffitt database where we have more than a thousand AML patients above the age of 70 and we are going to validate the model. Because it’s often a step that is missing in all these models and we think it’s very important to make sure it works in a different cohort of patients.