Emerging Prognostic Tools for Myelofibrosis Further Refine Disease Status

Raajit K. Rampal, MD, PhD

Raajit K. Rampal, MD, PhD, hematologic oncologist, Memorial Sloan Kettering Cancer Center, reviewed the prognostic tools used to find indicators of response to treatment in patients with myelofibrosis, during a Targeted Oncology Case-Based Peer Perspective Roundtable discussion.

Targeted Oncology^TM: In your practice, do you rely on hematologic pathology to make a diagnosis, and/or do you use diagnostic criteria for myelofibrosis (MF)?

RAMPAL: I think one of the things that tends to happen is that our pathology colleagues are not always attuned to all the clinical variables in a case....It has been my experience that you have a bone marrow sent to the pathologist and they say [it is] consistent with myeloid neoplasm, MDS [myelodysplastic syndrome], or MDS/MPN [myeloproliferative neoplasm syndrome] overlap.

Do you usually use next-generation sequencing (NGS) in your management of patients?

People are using NGS as part of their standard workflow, and...now there’s an abundance of commercial vendors who can do this.

I think a lot of the academic institutions have their own platforms, but certainly I don’t think anybody has a more robust panel than another. I think most of these [panel assays] cover most of the major mutations.

With the multiple risk assessment tools currently available, how do you choose between them, and what are the things that you worry about?

We worry about who is at risk of developing AML [acute myeloid leukemia]. Without a doubt, the mutations play a role. We know this, and we’ll talk about the 4 major mutations shortly, although I think that list is growing.

Clinically, of course, there are things that we think about. For example, the peripheral blood blast count is [one of] those things. I would point out [that] there is a recently released paper [looking] at the peripheral blood blast count in terms of its over- all prognosis, because for many years in the DIPSS [Dynamic International Prognostic Scoring System] and the IPSS, 1% to 2% blasts [counted for] a point in the risk stratification scoring system, and when patients are at 10% blasts or more, [that means that] they’re in accelerated phase and on their way to AML.¹

How do you treat patients who have a blast count between 4% and 10%?

In a paper by Masarova et al, investigators demonstrated that patients who have either bone marrow blasts greater than 5% or peripheral blood blasts greater than 4% seemed to have the same prognosis as patients with 10% or more peripheral blood blasts.1 That’s an important and striking finding because I think that says to us if somebody has 4% or 5% blasts, we have to be worried that they’re more likely to have progression in the frame of 22 to 24 months. We need to think about those patients more [similarly to how] we think about patients with 10% or more blasts.

How do you differentiate between the prognostic models of MF?

The DIPSS-Plus is a bit more comprehensive. All the same variables are captured in these models according to things [such as] age, anemia, leukocytosis, peripheral blood blasts, and constitutional symptoms. That one is tricky because what is really a constitutional symptom? Do you count it or not? I think that’s sometimes a perplexing issue.

The DIPSS-Plus is helpful in that it accounts for karyotype; [it] accounts for transfusion dependence, of which there is an abundance of data...with regard to transfusion dependence in MF having a significantly negative impact on the patient’s prognosis. A platelet count that is less than 100,000/mcL is also included.

Personally, I think that’s one of the most important things because, in my experience, those patients with thrombocyto- penia are the most difficult to manage for a variety of reasons. One, their natural history does seem to be of the more aggres- sive disease [type], but the second is that it does attenuate your ability to treat them with JAK inhibitors because most of them are platelet-based dosing.

The survival ranges widely. If we think about the [patients at low risk]—particularly with the DIPSS-Plus [tool]—if you’re talk- ing about a median survival of 15 years, unless there’s some other compelling reason, you wouldn’t want to refer that patient to transplant, but...the curve falls steeply after that....We always have to...[keep in mind] the patient’s age. I think that if somebody [has a DIPSS-Plus score of] intermediate-2 and maybe they’re 70 years old, that might be a different discussion than you would have potentially with a 40-year-old patient. The clock is ticking in both cases, but...age and comorbidity make a differ- ence. I think this does lay the framework for your decision regarding transplant.

Can you discuss the newest prognostic tool—Mutation- Enhanced International Prognostic Score (MIPSS70)?

The MIPSS is the newest kid on the block, and it encompasses many of the same clinical variables—the hemoglobin, leukocytosis, platelet count, blasts, and constitutional symptoms—as the others, but it incorporates a couple other key things. One is the bone marrow fibrosis. None of the other scoring systems incorporate bone marrow fibrosis, and here you get a point for fibrosis greater than or equal to 2.

Granted there is some degree of subjectivity here. We’ve all seen cases where we get a vague answer in terms of whether there’s a lot of fibrosis or not.

The genetics are what distinguishes this tool from the other tools. The 4 mutations that are known to have a prognostic impact in terms of overall survival [OS] and leukemia-free survival are ASXL1, which is not in any disease that I know of, EZH2, SRSF2, and IDH1/2.

I think it is worth recognizing that this is not likely to be a static list of mutations, meaning that these mutations came about in an analysis that was performed in 2013 when our gene panels were definitely smaller than they are now.

Recent data that have come out demonstrate that RAS pathway mutations also have a high association with leukemic transformation and with disease progression. I think this list [of mutations] is going to grow, and I think that’s the only point of caution.

Nonetheless, within the framework of this scoring system itself, the absence of calreticulin gets you points, ie, calreticulin is somewhat protective here. The presence of HMR [high-molecular risk] mutations, which are the 4 we talked about, gets a point, and the presence of 2 of these or more gets 2 points.

Can you explain the significance of the study by Paolo Guglielmelli and coinvestigators?

The investigators in that trial evaluated low-risk patients.² What’s striking to me is that the low-risk patient’s median OS was 27 years. What does that patient look like? That’s the patient who has calreticulin mutant MF, whose counts are preserved, in all likelihood.

Reviewing the Kaplan-Meier curves, the intermediate- and high-risk [patients] fall off of that curve, and you’re now talking about 2.3 years of median OS in patients with the highest risk. I think it is a powerful tool to stratify.

In the validation cohort, the intermediate-risk...patients trended a bit closer to the low-risk patients. Again, the intermediate group is a bit problematic, but the high-risk patient population is [where the survival drops].

I think this has been the most useful in patients who are, according to DIPSS, intermediate-1, maybe intermediate-2, but then you see they have an ASXL1 and/or ASXL2 [mutation, and] suddenly that shifts them from being intermediate- to high-risk.... That’s something I’ve noticed going back in a number of patients’ charts, looking at how I stratified them [based on] DIPSS and then reapplying their data to the MIPSS....It does shift—in some cases, not all—their risk profile, which could be clinically significant. It is something to think about.

I think this tool is really user friendly. I think it’s a nice tool... [and] a comprehensive tool, as well.

What’s the difference between MIPSS70 and MIPSS70-Plus?

Risk scoring stratification is [similar to] iPhones: There’s a new one every year, and it’s hard to keep track of all of these. There is now the MIPSS70-Plus model that also incorporates other mutations, as I mentioned earlier.³

This one incorporates U2AF1. The U2AF1 Q157 mutation has specifically been associated with anemia in [patients with] MF. Anemia, particularly transfusion-dependent anemia, gets you an additional point. This also incorporates more scoring categories, from very low risk to very high risk, and it includes karyotypes that are deemed high risk, such as inversion 3, or changes in 7 or 17q. This model does give you a bit more granularity.

At the end of the day, I think part of the challenge is that this doesn’t tell you how to manage the patient. If somebody is very low risk or somebody is very high risk, you know what to do with that. What do you do with the patient who is intermediate-1? There’s ambiguity there because there are a lot of clinical arguments in terms of whether you might want to take that patient to transplant. I think you have to be careful. Again, on the edges, it’s clear what to do, but in the middle, it is not always so clear. It gives you insight about the transplant and transplant timing, but it doesn’t tell you how to treat the patient in terms of nontransplant medical therapies.

What are the advantages and disadvantages of the MYSEC-PM model?

This is a tool that’s been used for post–PV [polycythemia vera] or post–ET [essential thrombocythemia]-MF.⁴ That’s what it was developed for or validated in. Again, it uses most of the same clinical variables we’ve talked about. However, it is not very robust from a genomic [perspective], nor is it very robust in terms of cytogenetics. There are no cytogenetics accounted for here. The only thing that’s accounted for is the calreticulin genotype.

I found this scoring system not well utilized by most people, and...what I’ve seen and observed and what I’ve also done is to use the MIPSS70 [model] even for patients with post-PV or post–ET-MF, recognizing that the higher-risk mutation profile was established in primary MF. I also think that it is not likely that the presence of an ASXL1 mutation is going to be favorable in a patient with post–PV-MF.

Part of the data that validate that idea [show] there is an emerging MIPSS-ET in the MIPSS-PV model. There is only 1 paper on this right now in the literature. Some of the same mutations are emerging in patients with PV and ET as risk factors for progression to MF. This suggests a continuum or spectrum of disease progression, and the same mutations that we know to be of negative prognostic implication in MF are likely to play some role in prognostication of PV and ET as they do in AML and other myeloid diseases. That’s something to keep in mind.

What happens next after you risk-stratify these patients with MF using these tools?

According to the National Comprehensive Cancer Network guidelines, we can risk-stratify these patients with MF.⁵ In higher- risk patients, we think about transplant referral if the patient is clinically appropriate for transplant. Regarding timing of treatment, there’s a tendency to treat earlier unless there are a few comorbidities or age....Clinical trial [enrollment] is something that everybody considers or at least thinks about at the time of treatment start.

Is there a utility in measuring the allele burden as a measurement when initiating treatment?

That’s an interesting question. I guess there are 2 things there. One is that JAK inhibitors, in the data to date, have a fairly modest effect on the allele burden. There are a couple of case reports of people having the JAK allele burden go down to nearly undetectable, but that’s the exception.

What has been known is that you see an increase in the variant allele frequency as the disease progresses in general. Patients have loss of heterozygosity where the variant allele frequency goes up to 90%. Those [patients] tend to be sicker and tend to have more proliferative disease. I think it tends to correlate with the clinical course. I think the problem is that there isn’t a well- defined demarcation as to when the severity happens. It tends to be a spectrum. It’s also influenced by the presence of the other mutations, and for that we have no perspective on variant allele frequency of things [such as] ASXL1 mutations in these diseases. That has not been studied to my knowledge. I think it does correlate with severity.

How do you interpret the poll results?

It looks like we have about as close [to a] consensus as we’re going to get. Ninety-three percent saying to initiate ruxolitinib [Jakafi] and refer to stem cell transplant. Seven percent are saying to refer straight to stem cell transplant. The fedratinib [Inrebic] label is quite broad, and it can be used in first-line [treatment] as well as a salvage after ruxolitinib failure.

I think Wernicke encephalopathy has concerned people, and I also think it’s just comfort and utilization. The trials for fedratinib, the JAKARTA trials, were carried out from 2012 to 2014 or so, so even [individuals] who have used the drug in clinical trials have not used the drug in years. It certainly seems [as though] there is a comfort factor there, maybe [an adverse effect (AE)] profile, but it’ll be interesting to see what the utilization of that drug is going forward.

How would you describe the spleen responses observed in COMFORT-I and COMFORT-II?

In COMFORT-I [NCT00952289], patients were randomized to placebo or ruxolitinib.⁶ COMFORT-II [NCT00934544] evaluated ruxolitinib versus BAT [best available therapy], which most commonly was hydroxyurea.⁷

The outcome data in COMFORT-I showed spleen size reduction of 35% or more, which is what the FDA had mandated as the response criteria. With ruxolitinib, it was 41% in COMFORT-I, and in COMFORT-II it was close to 28%, which is by itself interesting that there is some variability in response, but it was statistically significant.

Looking at the waterfall plots in terms of the spleen volume changes, in the patients treated with ruxolitinib, the vast majority had some degree of spleen size response. Now, I think one of the things that we have to take away from data [such as] these is a large proportion of patients are still getting a reduction in their spleen size.

My experience has been that it doesn’t take much spleen size reduction to make a patient [who is symptomatic] feel better. It is not uncommon to see a patient have an ultrasound that shows reduction in spleen size of 2 or 3 cm and for them to feel better from it. I think there is a broad range of utility of the JAK inhibitors in terms of spleen size reduction.

In COMFORT-II, the percentage of patients who had a spleen volume reduction was 97% with ruxolitinib and about 56% with BAT. Does that translate into all 97% of those patients having a clinical benefit? That’s unclear because the data are not granular enough to tell you that.

How do JAK inhibitors improve patients’ symptom response?

The number of patients who [had] some kind of a response was high. There are different symptom assessment forms. There’s the MPN symptom assessment form. About 50% or so of patients are getting a 50% reduction in the total symptom burden.

I think most patients in general get some degree of benefit. Again, most patients get some degree of reduction in their symptom burden, and, looking at the data more granularly in terms of the types of symptoms that improved, a lot of it was things [such as] early satiety and night sweats, but also things [such as] muscle pain and fatigue to some degree, which [are] not reflected [in these data]. Symptom burden and spleen size reduction are the mainstays of the JAK inhibitor therapeutic profile.

In COMFORT-II, a different scoring system was used; it was a more generalized, quality-of-life scoring metric. I think the major take-home message is that ruxolitinib improved patient’s quality of life by these validated scores. I think that’s useful information.

Fatigue is the No. 1 symptom among patients with MPN, and that’s also reported in the literature. You do see some improvement in fatigue in patients who are taking JAK inhibitors.

How would you describe the survival data in these trials?

In the analysis of COMFORT-I, there was a survival benefit that was observed. Neither COMFORT-I nor COMFORT-II [was] designed to evaluate patients’ OS, so there was a pooled analysis from COMFORT-I and COMFORT-II.⁸ That analysis shows that in some patients there seemed to be a reduction risk in death that is statistically significant for patients treated with ruxolitinib versus controls. Again, controls are both BAT and placebo. The hazard ratio was 0.7.

In general, what do the patients die from with this disease?

We worry about acute leukemia, but these patients are still susceptible to thrombotic events. They are susceptible to infections. One of the things that has emerged from these types of analyses is that it does appear that patients with MF have some increased risk of infection that appears to be innate to their disease....Other factors to consider are bleeding, portal hypertension, and other causes.

There are relatively recent data that have come out from a number of centers that identified pulmonary hypertension as a relatively frequent event in patients with MF, and it can be quite severe. We’re talking end-stage pulmonary hypertension. The etiology of this is also not clear at this point, but I think that is something we should always think about in our patients with MF, [especially] those who are exhibiting signs of heart failure. This seems to be particular to the disease and not so much to the treatment.

Does spleen size have an effect on outcome?

The reduction in spleen size appears to make a difference in terms of how patients do. I don’t think anybody would argue that you are saving lives by shrinking spleens, but I think it’s more a marker of responsiveness to disease. If you look at patients who have a spleen size response at 6 months versus those with no response at 6 months, there’s a statistically significant improvement in their OS. That being said, the same thing is true of the durability. If you look at responders, even those who don’t have a sustained response versus those who are nonresponders, it’s clear there is a survival benefit.

I would argue...that this is just a marker of disease responsiveness to JAK inhibitors rather than the spleen being part and parcel of a patient’s survival.

Please describe factors that reduce the response to ruxolitinib.

Some of the factors that have been identified in terms of a reduced chance of response to ruxolitinib include having a higher-risk disease but also having a spleen size greater than 20 cm. I think that’s important, and [it’s] the reason that you start treatment earlier on.

As the spleen gets larger, the likelihood that you’re going to get a substantial reduction in the spleen size decreases, and that can affect a patient’s survival. There is certainly an argument to starting treatment on the earlier side.

The dose of ruxolitinib matters, so using doses less than 10 mg twice daily can also be associated with a lower likelihood of response.

If you look at spleen volume responses as a function of the dose of ruxolitinib, there’s clearly a dose response. It flattens out around 20 mg, but if you look at under 10 mg twice daily versus 10 mg twice daily, there is a substantial response difference.

What’s more striking, though, is when you look at that dosing difference with regard to the symptom score. It’s quite striking that you’re getting much more of a benefit once you get to that 10-mg twice-daily dose. Again, this does have implications for potentially how much spleen reduction you’re going to get, and by extension what the impact on the patient’s survival is. Using the maximal dose that you can and that is tolerable for the patient is important.

How does platelet count factor in your decision about dose for ruxolitinib?

The current label of ruxolitinib for patients under [100,000] platelets, between [50,000] and [100,000], is 5 mg twice a day. There have been 2 studies that have looked at increased doses.

Verstovsek et al looked at using a gradual dose escalation in patients in 2 cohorts.⁹ One was patients who had a platelet count of 75,000/mcL to 99,000/mcL. The other group was patients with a count of 50,000/mcL to 74,000/mcL. In both of these cohorts, patients were started on the FDA label dose of 5 mg twice daily, and then they were titrated up to 5 mg and 10 mg and then titrated up to 10 mg and 10 mg. Investigators looked at the maximal tolerable dose by way of hematologic toxicity. What they found in both of these cohorts was that 10 mg twice daily seemed to be a tolerable dose in both of these cohorts.

How would you describe the safety profile of ruxolitinib in the COMFORT-I and COMFORT-II trials?

Grade 3 and 4 AEs showed that 35% of patients in stratum 1, which is 75,000 to 99,000 platelets, had a grade 3 or 4 throm- bocytopenia. [However,] 78% in the 50,000 to 74,000 range had a grade 3/4 AE. Granted, you had to consider the baseline that you’re starting from.

Anemia was roughly 20% in both. It does appear that careful up-titration is something that you can reasonably do with patients with thrombocytopenia of 50,000 to 100,000 [platelets]. I think it is something worth considering again because we know that the dose of ruxolitinib does matter in terms of patients’ outcomes.

Please describe the spleen response in the JAKARTA trial.

In JAKARTA [NCT01437787], the spleen response was a 35% volumetric reduction at week 24 with fedratinib 400 mg daily.¹⁰ It is similar to what we see with ruxolitinib. Spleen volume in the placebo arm resulted in a reduction of 1%. When stratified by platelets, the spleen response with patients under 100,000/ mcL was 36%, and [for those whose] platelets [were] greater than 100,000/mcL, [it was about] about 49%. You are getting a bit more response in the [patients with] higher platelets.

It’s important to remember there is no dose reduction in the label for fedratinib in patients with 50,000 to 100,000 platelets; it’s still 400 mg daily. Symptom responses were roughly the same, 31% and 42%.

How would you describe the AEs in the trial?

They are a bit different when we’re talking about the nonhematologic AEs. Diarrhea was observed in 66% of patients treated with fedratinib at 400 mg daily. That’s all grades. Five percent or so were grade 3 or 4. Vomiting was in 42% and nausea was in 64%. You see more gastrointestinal toxicity with fedratinib versus what has been reported historically with ruxolitinib.

With regard to cytopenias, 99% of patients had some degree of anemia and some degree of thrombocytopenia, but the grade 3/4 thrombocytopenia was 17% and anemia was about 43%. Maybe [there was] a bit less thrombocytopenia, but [it was] the same degree of anemia, roughly speaking.

There were several cases [of Wernicke encephalopathy] reported. Retrospective analysis was presented at [the American Society of Hematology annual meeting] about 2 years ago on the reported cases, and at least there was the potential that attribution for many of the cases was to something else, or the diagnosis couldn’t be confirmed. Nonetheless, this is a black box warning on the drug label, so that needs to be discussed with patients.

_References:

_{1. Masarova L, Bose P, Pemmaraju N, et al. Prognostic value of blasts in periph- eral blood in myelofibrosis in the ruxolitinib era. Cancer. 2020;126(19):4322-4331. doi:10.1002/cncr.33094}

_{2. Guglielmelli P, Lasho TL, Rotunno G, et al. MIPSS70: Mutation-Enhanced International Prognostic Score System for transplantation-age patients with primary myelofibrosis. J Clin Oncol. 2018;36(4):310-318. doi:10.1200/JCO.2017.76.4886}

_{3. Tefferi A, Guglielmelli P, Lasho TL, et al. MIPSS70+ version 2.0: Mutation and Karyotype-Enhanced International Prognostic Scoring System for primary myelofi- brosis. J Clin Oncol. 2018;36(17):1769-1770. doi:10.1200/JCO.2018.78.9867}

_{4. Passamonti F, Giorgino T, Mora B, et al. A clinical-molecular prognostic model to predict survival in patients with post polycythemia vera and post essential thrombocythemia myelofibrosis. Leukemia. 2017;31(12):2726-2731. doi:10.1038/ leu.2017.169}

_{5. NCCN. Clinical Practice Guidelines in Oncology. Myeloproliferative neoplasms, version 1.2020. Accessed December 14, 2020. https://bit.ly/2Wcczfa}

_{6. Verstovsek S, Mesa RA, Gotlib J, et al. A double-blind, placebo-controlled trial of ruxolitinib for myelofibrosis. N Engl J Med. 2012;366(9):799-807. doi:10.1056/ NEJMoa1110557}

_{7. Harrison C, Kiladjian JJ, Al-Ali HK, et al. JAK inhibition with ruxolitinib versus best available therapy for myelofibrosis. N Engl J Med. 2012;366(9):787-798. doi:10.1056/ NEJMoa1110556}

_{8. Verstovsek S, Gotlib J, Mesa RA, et al. Long-term survival in patients treated with ruxolitinib for myelofibrosis: COMFORT-I and -II pooled analyses. J Hematol Oncol. 2017;10(1):156. doi:10.1186/s13045-017-0527-7}

_{9. Verstovsek S, Gotlib J, Gupta V, et al. Management of cytopenias in patients with myelofibrosis treated with ruxolitinib and effect of dose modifications on efficacy outcomes. Onco Targets Ther. 2013;7:13-21. doi:10.2147/OTT.S53348}

_{10. Pardanani A, Harrison C, Cortes JE, et al. Safety and efficacy of fedratinib in patients with primary or secondary myelofibrosis: a randomized clinical trial. JAMA Oncol. 2015;1(5):643-651. doi:10.1001/jamaoncol.2015.1590}