Q: Is it possible to identify clinical features in patients with ITP who are at risk of life-threatening bleeding?

I am a little bit skeptical about this. We have these crude predictors that we use right now. Of course, we use the platelet count. We also use age, bleeding history, and whether the patient is on concomitant medications that increase bleeding risk as possible risk factors. They are very crude tools, and they may ultimately, over a large number of patients, help to stratify risk a little bit. Crude predictors are certainly not a crystal ball that is going to identify who is going to bleed and who is not going to bleed. Frankly, I am somewhat skeptical that we will ever be able to identify predictors or biomarkers that identify bleeding risks. That is because there are just too many stochastic factors at play in why a patient with ITP bleeds or does not bleed. Some experts and investigators have thought about whether we should be measuring platelet function, for example, in patients with ITP, which is not a straightforward thing to do in somebody with a low platelet count. Maybe that has some utility, but I, frankly, am skeptical that we are ever going to find biomarkers and/or clinical predictors that add a lot to the variables that I listed.

First of all, people who are bleeding are at risk of life-threatening bleeding, so the best way to predict if somebody is at risk of life-threatening bleeding is to know if that patient was already bleeding. Then, patients who are older are probably at a higher risk. When you get down into lower platelet counts of less than 10,000, a patient may be at risk for life-threatening bleeding, and, certainly, when you start getting down to less than 5000. There are a lot of data, although not in patients with ITP, that suggest that when platelet counts get down to less than 5000, bleeding really starts to take off. And the longer that you have severe thrombocytopenia—meaning platelet counts of less than 10,000 or 5000—the more at risk you are of having a bleed. Part of this is really just knowing your patient. Therefore, early on in treatment when I do not really know a patient, I am much more nervous about someone who is running in the low teens or less than 10 than I am about a patient I have been following long term who is aware of his or her symptoms and what to look for, and also it seems that these patients have their own set point. Clearly, if patients have some other morbidity that adds to their risk of bleeding, such as uncontrolled hypertension, you are going to worry more about them, but I do not think that there is anything else that I could really add that would help me to identify patients whom I am really worried about being at risk for severe bleeding episodes.

In ITP, it is all of the clinical features that would mark a patient for risk of bleeding in other circumstances, such older patient age, hypertension, and patients who are taking aspirin. It would include those clinical features that we know about from clinical studies in other disease state, such as the warfarin atrial fibrillation trials, that will indicate if someone might be at risk for more significant bleeding, which is intracranial. There is a suggestion that has been raised in the medical literature about wet purpura, specifically how extensive does mucosal bleeding and cutaneous bleeding have to be to indicate a bleeding phenotype. In my own experience, I saw a relative refractory female patient who had an excellent workup and had a spontaneous intra-abdominal bleed. It was really an unusual circumstance, and when I saw the patient she had 23,000 platelets. She is now being treated with the highest dose of a thrombopoietin receptor agonist. She is totally covered with ecchymoses, and she says that her gums bleed when she brushes her teeth. This is a bleeding phenotype patient. Those are the types of patients that really get you very anxious about severe bleeding. You need to know your patient and consider the factors, and then work as best you can to maintain a safe platelet count with treatment.