Rami S. Komrokji, MD, explains the diagnostic and molecular workup behind myeloproliferative neoplasm diagnoses and discusses treatment options based on the case scenario of a patient with myelofibrosis.
Rami S. Komrokji, MD
During a recentTargeted Oncologylive case-based peer perspectives presentation, Rami S. Komrokji, MD, explained to a group of physicians the diagnostic and molecular workup behind myeloproliferative neoplasm (MPN) diagnoses. In addition, Komrokji, a senior member of the Malignant Hematology and Experimental Therapeutics Program and head of the Leukemia and Myelodysplastic Syndrome Section and vice chair of the Malignant Hematology Department at the Moffitt Cancer Center, discussed treatment options based on the case scenario of a patient with myelofibrosis (MF).
A 59-year-old man presented to his physician with symptoms of fatigue, night sweats, weight loss, and abdominal pain with early satiety during the past 6 months. On physical exam, the spleen was found to be palpable 10 cm below the left costal margin.
Genetic testing was completed and he was found to have aJAK2V617F mutation and anASXL1mutation. A bone marrow biopsy revealed megakaryocyte proliferation and atypia with evidence of reticulin fibrosis, and a blood smear showed he had leukoerythroblastosis.
Laboratory findings were notable for a red blood cell count of 3.20 x 1012/L; hemoglobin, 9.7 g/dL; hematocrit, 34%; mean corpuscular volume, 93.1 fL; white blood cell count, 12.1 x 109/L; platelet count, 247 x 109/L; and peripheral blood blasts of 0%.
How do you approach a diagnosis of MF?
I always think of 3 things when I see patients: I want to verify their diagnosis, then I want to know the risk of the disease, and then I want to tailor the treatment accordingly. That is the first step in any disease that we see as oncologists. At least in the myelodysplastic syndrome [MDS] series, in around 20% of the patients, sometimes we change diagnosisand this occurs in MF, as well. We have gone from the extreme of seeing patients that do not have MDS and have been on azacitidine [Vidaza] for a year—and a lot of them have a high chance for anemia—to patients coming to see me before they go to transplant in a week for MF and they have hairy cell leukemia. I have seen this spectrum, so I always like to take a minute and step back to verify the diagnosis.
The next step is to do a disease risk assessment; then the treatment is based on that. It is getting very complicated nowadays for every myeloid disease we have. We have 6 or 7 risk models, molecular data that in real life become very difficult to calculate and capture…but are important.
Does this patient meet the criteria for primary myelofibrosis (PMF)?
Komrokji:This case is a classic case of MF. As you know, MF is a Philadelphia chromosomenegative MPN. It is not the most common MPN disease we see. Essential thrombocythemia [ET] and polycythemia vera [PV] are probably more common, but MF is the most symptomatic and the one that is generally more associated with worse outcomes. In our practice, because of the referral time bias, we actually see more MF than we see ET or PV because most of the patients with ET and PV are managed locally, but we get to see almost every case of MF in the area. The line between those diseases is sometimes not well demarcated. We see a lot of the patient features overlap, but obviously, MF is the primary de novo PMF. About 50% are coming from ET and PV. All those diseases can go into what we call an accelerated phase MPN with a blastoid increasing and eventually, it goes to active myeloid leukemia [AML]. If we look at PV and ET, roughly somewhere around 10% to 20% will end up being MF and roughly around 5% to 10% of those at one point will move to AML, which is probably the worst MPN and has the worst outcome.
Obviously, what we know about the pathogenesis of those diseases is that all of them are at the stem-cell level of clonal myeloproliferative, where we have somatic mutations that are acquired. The base of the somatic mutation and subsequent hits is where we get the disease. In certain mutations we get MDS; in certain mutations we get ET or PV. In MF particularly, the disease is characterized by cytokine release or changes that lead to fibrosis in the bone marrow, extramedullary hematopoiesis, and the manifestations of the disease that we have seen. Sometimes the sequence of those mutations is what dictates the clinical phenotype that we see. For example, looking at theTET2andJAK2mutations: If theTET2mutation happens first and it is the founding mutation, and theJAK2is the second mutation, you end in a phenotype of ET. If theJAK2mutation happens first and theTET2was the subsequent mutation, then it is a PV clinical phenotype.
The hallmark of all MPN diseases is the overactivation of the JAK-STAT pathway. That is probably what drives all the clinical phenotypes that we see. The pathway is activated in different ways. One is mainly because of the increased cytokines of those diseases, so there is activation of the pathway, but [another way is] by the mutations. There are 3 mutations that we know about and talk about now: theJAK2mutation, which is the most common;CALRmutation; and theMPLmutation. Those 3 mutations are what we call the phenotype-driver mutations. They are not necessarily the disease-initiating mutations, but they dictate a lot of the clinical phenotypes that we see in those patients.
If you look at patients with PV, 95% to 98% of them will have aJAK2mutation, either the V617F mutation, which is the classic, or, in around 2% or 3%, the exon 12 mutation. It is probably important to check the exon 12JAK2mutation in PV only if theJAKV617F is negative. You will not see an exon 12 in MF or ET; it is usually just in PV. Nowadays, if patients are negative for theJAK2mutation, we start by questioning if this is really PV or not. In ET or MF, almost half of the patients will have aJAK2mutation, 20% to 30% will have theCALRmutation, and 5% to 10% will have theMPL. Then there is a group we talk about now that we call triple negative. They have none of those phenotype-driver mutations. Those types are the worst. We are actually looking at our experience in our facility, asking the question: Is triple-negative disease real, or are those more like MDS with fibrosis rather than a true myeloproliferative disease? There is a small category still called triple-negative MF that tends to be the worst acting.
In addition to those phenotype-driver mutations, there are other mutations that we see that are important prognostically. The most important [of those] are theASXL1mutation and theSRSF2. Those have been shown in several series to have worse outcomes in MF. The presence ofIDH1andIDH2mutations in MF and in PV would suggest higher risk of progression to the accelerated phase of MPN, so those patients do not do well. A very small subset of patients with MPN who haveTP53do not do as well, either. You do see a spectrum of mutations in addition to the phenotype-driver mutations in those diseases.
Would this patient then have a different characterization?
The World Health Organization now recognizes the activity called prefibrotic MF, which in many cases in practice, we deal with as ET. Those patients do not have the clear MF in the bone marrow, so the criteria are almost the same as PMF, but you lack that grade 2 or 3 fibrosis. A good hematopathologist should be able to distinguish that. The pattern of the clustering of the major karyocytes and how they look is a little bit different from ET. Clinically, the bone marrow is usually more hypercellular, and there’s more hyperplasia than what you see in ET. It’s not necessarily that we manage those patients with prefibrotic MF differently from in ET, if they just present with high platelets, but their outcome is different. Those are the patients that we sometimes see in 2 or 3 years and we say have transformed to MF. It is very unusual to have a classic ET going to MF in 2 to 3 years. Usually that happens over the span of 10, 15, to 20 years, but we do see cases where patients transform to MF within a couple of years, and many times, those are the cases that are prefibrotic MF. There is more emphasis now on trying to tease those out and classify them, maybe just for better surveillance and knowing that those patients will have higher risk of going to MF.
To my knowledge, prefibrotic MF does not have a different genetic signature from MF. From ET, I think they are very similar. You probably see more bad gene mutations in MF; it is less common to seeASXL1andSLSF2. I think that there are fewer high-risk mutations than with true ET; it is less common to seeASXL1,SLSF2,orTP53in the prefibrotic patients. Other than that, I do not know if there is a difference in the spectrum of the mutations. In the phenotypic-driver mutations, there is a little difference in how they present and in the clinical features of MF.CALRtends to be in the younger patients [with] less thrombocytosis, so you could see a little bit less ofCALR.There are subtle differences, but I do not think you can really distinguish based on that.
The classic MF requires the presence of fibrosis, not meeting other criteria of other myeloid diseases. I think the classical teaching is that we always have to rule out chronic myeloid leukemia and the presence of any of those mutations, then [consider] the myeloid criteria. The ones that are really difficult to tease out for us are those of MDS or MPN with fibrosis.…Again, there is no specific signature to distinguish those. It depends if there is erythroid or myeloid dysplasia that you usually should not see in classic MF. That is also in the eye of the hematopathologist, so maybe atypia could always be labeled as dysplasia when it is not. In my experience, those are the most difficult to tease out. If it is MDS with fibrosis, typically and clinically, the patient would not have the high splenomegaly; they will have more cytopenia features than the MPNs. That is why when we see those cases of triple-negative MF, I tend to think of them more as MDS with fibrosis than MPN.
The things that could also mimic MPN that I have seen over time involve connective tissue disease. I have seen patients that have lupus; they present with fibrosis in the bone marrow and are labeled PMF. You do all the workup to see it is lupus. I have seen hairy cell leukemia, as I have mentioned. The most extreme case I have seen is somebody in his 50s who was referred to transplant and was going to go to transplant in a week, but he sought a second opinion and came to see us. It wasn’t even meit was the hematopathologist that called me and said, “You know, this is anything but PMF. I think this is hairy cell leukemia.” The guy got cladribine for 5 days, and the fibrosis went away. I have seen marginal zone lymphoma present with fibrosis in the bone marrow; with rituximab [Rituxan], it goes away. So, not every fibrosis equals MF in the bone marrow.
What’s the difference here between primary and secondary MF?
Now there are certain criteria to diagnose secondary MF. If patients have ET or PV, they may require the history of prior PV or ET, of developmental fibrosis; in PV, they stop needing phlebotomies; in ET, there is dropping hemoglobin by 2g and development of splenomegaly. Those are the criteria to diagnose secondary MF. Once we make the diagnosis, then we do the risk stratification of the patient. That is where it gets complicated. There are so many separate models that it is even more complicated than MDS, but the idea is the same: We are trying to weigh the disease factors to [determine] the estimated prognosis for this patient. Should I be thinking of allogenic stem cell transplant? In general in ET or PV, we do not think of transplant. It is mostly when we are talking about MF. If we estimate that the disease risk will affect the patient’s survival in the coming 2 to 3 years, then we are thinking of allogenic stem cell transplant for that patient.
How do you calculate risk status for such patients?
All the models use almost the same variables with different points assigned to them. I do not think it is practical to expect to memorize all those models. Basically, you look at values like constitutional symptoms, anemia, cytopenia, cytogenetics, presence of peripheral blasts, andnowadays—the molecular mutations. You get a score based on that. Patients who are symptomatic—have leukocytosis, peripheral blast; are anemic, thrombocytopenic; have bad cytogenetics and any more—will come as poor risk.
The International Prognostic Scoring System [IPSS] is the easiest to look at with 5 variables; each gets a point. If you have no points, it’s low risk. If it’s 1 point, it’s intermediate-1. If it’s 2, intermediate-2. Three points or higher, you’re at higher risk. You see the survival vary from 2 years for the high-risk patients to more than 10 years if you have low risk.
Dynamic IPSS [DIPSS] is the same, but it is supposed to be dynamic. You get more points for the anemia, but you can use it at any time. If you see the patient today after 5 years and you look at those clinical variables, you can predict the outcome from that time point.
The DIPPS-plus allows for the platelets and karyotypes. In addition to the IPPS or the DIPPS, the DIPPS-plus gives points for the platelets and the unfavorable karyotype. Again, you divide the patients into 4 risk groups, and based on that, you can predict the outcome. The outcome varies from somewhere in low risk of 15 years to 1 year, so it is really a spectrum, and it will make a big difference in how we will approach those patients.
When you look at the majority of patients with MF, <10% will be low risk. Most of the patients will be in the intermediate-risk group. It gets more complicated now that there is more that is incorporating the molecular data. There are 2 models: one is called GIPSS [genetically inspired prognostic scoring system]. That is only with somatic mutation and formation of cytogenetics. There is no clinical value, no age, nothing in that model. Obviously, if the patient has the bad mutations, like theASXL1orSRSF2or theU2AF1specifically in the hot spot Q157, lacks theCALRmutation, or has an unfavorable karyotypeusually those are like the 5, 7, or 17 complex karyotype—the patient is labeled as higher risk; and you can get the point and give the patient a state based on the GIPSS.
Or we have the MIPSS70, and there are different versions of it that are hybrids between clinical and molecular data. It is particularly designed to assess transplant decision for younger patients. I even started with this in the past; we always weigh in age and prognosis since we always know that if somebody is 75 years, they are not going to be like somebody who is 50. For example, if someone has constitutional symptoms, I give them 1 point. But if I say, “He’s above age 65,” I’m giving him 2 points on the IPSS, and I’m going to say, “This is an intermediate-2 risk.” I will start thinking of transplant because of the age, which is counterintuitive for me. I always decide on the transplant based on disease-related factors rather than the patient-related factors.
How do these prognostic models help guide treatment decisions?
That’s what the 2 models are trying to dotake the age outside of the decision for treating the patients. The same exists in some of the MDS models, as well. Those models are supposed to be decisive for transplant. We tried to look at those models to see if there is any discrepancy. For example, 1 of our faculty looked at our database; we have almost 400 or 500 patients with MF in our database, and we looked at when there was a discrepancy of 2 stages between the clinical and the molecular ones. The genetic drive model does better in predicting the outcome. It’s getting so complicated, but the idea is to try to calculate the risk of the disease. Any of the bad clinical variables we know—presence of symptoms, leukocytosis, separated blasts—are all bad players. Mutations likeASXL1,SRSF2,U2AF1,P53all of those are bad players. It could be more complicated than this so that even the type of mutation could matter. Even with theASXL1, if you have a frameshift, they do worse than with a missense mutation, and they differ in terms of the outcome, so it’s very complicated. At the end, you try to get a sense of the risk of the disease to say, “Am I going to think of a transplant? Am I going to justify a 20% transplant-related mortality for this patient?” If I have a patient that has a survival of 15 years, there’s no way to have the patient go for a transplant with a 20% transplant-related mortality becoming 6 months. If I have a patient with a 1-year survival of disease, I think it’s very justifiable to take them to allogeneic stem cell transplant for those patients.
Once we put the patient in the risk group of MF, if they are very high or high risk, we are thinking of transplant. If patients are very low risk, I think observation is very acceptable. If they are intermediate risk, then we think of treatment based on symptoms. In reality, we don’t transplant more than 10% of the patients with MF, not necessarily because of the disease risk but also because of the comorbidities and other things that might exclude those patients from transplant. Anybody who is at high or very high risk should be considered for transplant. Whichever model you think of, that’s what we treat patients with.
What is the trigger to initiate therapy for a patient with MF?
If the disease risk is high, patients are intermediate-2 or higher, we do consider transplant. If patients are not candidates for transplant, then the treatment is based on symptoms. You try to categorize or put the patients into a category, and what you see in MF is almost like 2 buckets: somebody with main symptoms of splenomegaly or constitutional symptomssometimes those go hand in hand—or cytopenias. The most difficult groups are the ones that share the features of both. They have cytopenias and constitutional symptoms and splenomegaly.
For patients with constitutional symptoms and splenomegaly, obviously JAK2 inhibitors have become the standard of care. Patients with cytopenia don’t benefit from that treatment. Actually, the main side effect of ruxolitinib [Jakafi] is myelosuppression. If they are purely anemic, about 20% of patients with MF come just with predominantly cytopenia and will be given a 6 or 7, and unfortunately, those patients we start now on ruxolitinib, those patients typically within 6 to 8 weeks will drop the hemoglobin by 2 gm on treatment, so it’s really not a treatment for anemia.
For that group [with both anemia and symptomatic splenomegaly], sometimes we have a discussion with the patient based on what is most symptomatic. If the patient’s predominant symptoms are from splenomegaly and constitutional symptoms and are better than cachexia, we may do ruxolitinib, even if it would sometimes render them transfusion dependent. The group that has cytopenia and splenomegaly is probably the most difficult to treat.
If patients have huge splenomegaly and cytopenias, we treat. We don’t do a great job with gauging the symptoms of patients. There are those that have forms where you try to quantify predictively some of the symptoms that could be related to quality of lifelike the fatigue, itching, night sweats, and bone pain, and I don’t think we have done a good job with that. The new studies do that—we now have patients with an iPad or something that they check in with every day. I don’t know how much value they have, but I think it’s important to look at them at one point. There are some studies that suggest that if you have a total symptom score more than 5, patients should be getting treatment, but the FDA actually would approve drugs based on symptom score improvement, including when they looked at ruxolitinib and the new JAK2 inhibitors; they do approve based on that.
What is your experience with treating splenomegaly with radiation?
In my experience, although patients respond, it is a very short-lived response.…I’d say on average, if their response is 2 to 3 months, then the spleen will start growing back again. They can get profoundly cytopenic. I do it, but I do it if someone is very symptomatic, as palliative [care]as a last option. In the past, they used to say also if you radiate, it’s contraindicated to do splenectomy; I’m not a big fan of splenectomy, so I don’t care about that point per se.
We’ve tried to do splenic embolizations.…We’ve had limited success with that. I’ve been able to get patients off paracentesis because after a while, with the spleen enlargement, they develop portal hypertension and ascites, and we have to do paracentesis on those patients on a regular basis. I’ve been able to get a couple of patients off that with splenic embolization, but they get sick in the beginning. So we will not be able to do classic splenic embolization in many patients with MF.
Splenectomyin 10 years, I’ve probably done 2 in patients with MF. There’s a lot of mortality, morbidity with it. You get a level of compensation. I’ve used radiation in extramedullary hematopoiesis; it’s the most effective treatment. The standard treatment for extramedullary hematopoiesis, if it’s symptomatic, is radiation—that’s a very effective treatment.
What other treatment options are there for such patients?
Ruxolitinib is the treatment approved by the FDA for patients with splenomegaly and constitutional symptoms. There were 2 trials: the COMFORT-I and COMFORT-II.1,2COMFORT-I was done in the United States against placebo. COMFORT-II was done in Europe against best available therapy. Both had primary endpoints of the spleen and symptom response. One of them looked at 24 weeks; the other looked at 48 weeks. They looked at spleen reduction by MRI [magnetic resonance imaging] by 35%, which we don’t do in real life; we do either physical exam or ultrasound. The reason for reduction of 35% is because in phase I, they correlated that to 50% reduction by physical exam, so the FDA had required this MRI for approval of every drug.
There is no doubt that the drug is effective. More than half of the patients would meet the primary endpoint of the studies. But actually, almost 90% of the patients would have some degree of spleen reduction with those treatments, in addition to symptom improvement. Now, the studies allowed crossover between the 2 arms, and they were never designed for survival, But both of them actually showed survival advantage. I would not start treatment in those patients by saying “Take ruxolitinib because it is going to improve survival.” However, patients that are intermediate-2 and high risk, if they had splenomegaly and constitutional symptoms, derive survival advantage from the treatment. My intent of treatment is symptom management, but compared with using hydroxyurea [Hydrea], they do better. It is not that those drugs alter the natural history of the disease; there is no proof that ruxolitinib will decrease that late AML transformation. Even the reversal of fibrosis is very controversial. Those drugs do not affect theJAK2allele disease burden much. What happens is that those patients’ spleens are shrinking, the symptoms get better, the performance status gets better. We get a much higher percentage of patients to transplant, whereas in the past, they were excluded because of very poor performance status prior to transplant. It’s indirectly having a survival advantage but not necessarily by affecting the natural history of disease.
Do you use ruxolitinib in patients with platelets below 50,000?
There have been studies looking at 50,000 to 100,000 dose escalated, where you start at the lower dose and then go up. Sometimes, if patients are very symptomatic with splenomegaly and constitutional symptoms, we do. The hope is that if pacritinib gets approved, that would be the niche for it. The other trick I have done sometimes nowadaysand there are phase I/II data—is use combination ruxolitinib plus thalidomide [Thalomid] because you do get platelet response with thalidomide, then you add the ruxolitinib for those patients. That’s what I have done in some patients, so you definitely could do it. The problem is, how much can you push the dose? In terms of symptom improvement with ruxolitinib, you really see symptom improvements at 5 or 10 mg. Patients will feel amazingly better in a week or 2 because it’s all related to cytokines. The spleen response is dose dependent. You rarely will see a meaningful spleen response beyond a 10-mg dose. If you cannot push the dose, it becomes not useful. I have tried it sometimes if they have very symptomatic splenomegaly. I do use it, but you just have to proceed with caution. You may not be able to get the meaningful dose to have a spleen response with that.
What about some of the newer JAK2 inhibitors? Would you consider any of them if they were approved?
There are a few other JAK2 inhibitors we’ve been looking at. The median duration of response to ruxolitinib was around 3 years. After ruxolitinib failure, the outcome is not the best. We looked at our series, which was around 17 months, and [The University of Texas] MD Anderson Cancer Center published their series of around 17 months. It’s really an unmet need that when patients stop responding to ruxolitinib, especially when they have a huge spleen, there are not many options outside there. As mentioned, sometimes we do radiation, splenic embolization, and sometimes I have done even palliative chemotherapy like low-dose cytarabine, but all of those are very short-lived.
There are 2 JAK2 inhibitors further in development. One is called pacritinib, which originated in studies at Moffitt and is being developed particularly in patients with thrombocytopenia, so it looks less myeloid suppressive. So for patients with platelets less than 100, that could become the choice therapy. It’s gone through different phases of studies; the phase III was positive, but there was a question about increased mortality, so the FDA had requested more data. It’s still ongoing in a dose-finding study for approval.
The other is called fedratinib, which again finished phase III studies. This has shown survival and met the primary endpoint of spleen response. It was going to the FDA for approval, but then there were reports of several cases of encephalopathy showing up, so the company that was doing the studies pulled the application from the FDA, and then they sold the drug for a very low price to the small company that had originally sold them the drug. That company collected those cases of Wernicke encephalopathy and went into further detail to find out that only about 4 or 5 cases were true Wernicke encephalopathy; the rest were not. Some of those were patients that were thiamine deficient with nausea and vomiting. They sold the drug again to Celgene. It’s now in front of the FDA, and in 2 or 3 months, they will have to decide on the approval.
Momelotinib is the third one that has some anemia response. The studies were not as strong meeting the primary endpoint, so it is still up in the air regarding whether or not [it’ll go ahead]. The ones that will get approved are pacritinib and fedratinib.
Fedratinib [was studied] in the JAKARTA-1 and JAKARTA-2 trials. JAKARTA-2 looked at patients that had ruxolitinib failure, so after ruxolitinib failure, they used fedratinib and reported a 55% response, which is almost as a first-line response.3The catch on this study is that their definition of ruxolitinib failure was a little bit loose. They had said that anybody who has 14 days or more of exposure to ruxolitinib was allowed to go on the study. When you look at the median time on ruxolitinib as 2 to 3 months, there is a group that may not have reflected the classic ruxolitinib failure of someone who had been on for 2 to 3 years. My guess is that in real life, when we start to use this, it’s after 2 to 3 years of ruxolitinib. We’re not going to see the 50% response. We still could see some response, and I think it’s going to get approved for patients with ruxolitinib failure.
What are some of the differences between these newer JAK2 inhibitors?
The toxicity profile between all those JAK2 inhibitors is really dependent on theJAK2. It’s dependent on the other targets they hit. Ruxolitinib has a lot of JAK1 or JAK2 activity. Pacritinib has FLT3 activity, and momelotinib has some of the activity against the TGF β-receptors. That’s why we think it has some anemia responsebecause it works toward the active receptor. Pacritinib has many more gastrointestinal adverse effects than fedratinib because it hits FLT3. With FLT3 you get that. It is a mixed thing. Sometimes the efficacy could also be off target. You definitely see a difference in the toxicity profile based on the other targets that the drugs are affecting.