Refining Prognosis During Treatment for Molecularly Defined Lower-Risk Myelofibrosis

A New “Living” Risk Score Could Reveal Which Myelofibrosis Patients Are Truly in Danger

For patients with myelofibrosis, a diagnosis that initially appears to carry a relatively lower risk may not remain reassuring once treatment begins. A study published in Annals of Hematology suggests that prognosis can be sharpened by combining molecular information collected at diagnosis with early changes observed during therapy. The approach could help doctors distinguish patients who are likely to remain stable from those whose disease is quietly becoming more aggressive, even when conventional clinical scoring systems place them in the same broad risk category. In an analysis of 197 people with myelofibrosis, researchers found that a dynamic score called iRR6 identified sharply different survival groups after six months of treatment with ruxolitinib. In the lowest-risk groups, five-year overall survival reached 100 percent, while the highest-risk groups had five-year survival of 54.8 percent in primary myelofibrosis and 60.4 percent in secondary myelofibrosis.

Myelofibrosis is a chronic blood cancer in which abnormal stem and progenitor cells disrupt the bone marrow’s ability to produce blood cells normally. The disease is often accompanied by marrow scarring, or fibrosis, an enlarged spleen, anemia, constitutional symptoms such as fever and night sweats, and progressive abnormalities in blood counts. It can arise as primary myelofibrosis, or develop after another myeloproliferative neoplasm, in which case it is described as secondary myelofibrosis. Its clinical course is highly variable: some people live for many years with manageable symptoms, whereas others develop severe cytopenias, rapidly progressive disease or acute leukemia. That variability makes risk stratification central to treatment decisions, including the timing of stem-cell transplantation, the only potentially curative option for some patients. The challenge is that risk is not fixed. A patient’s biology and clinical condition can change after diagnosis, meaning that a score calculated only once may fail to capture the disease’s evolving behavior.

The investigators focused on patients categorized as intermediate-1 risk by molecularly informed systems. These included the Mutation-Enhanced International Prognostic Scoring System for patients with primary myelofibrosis, known as MIPSS70, and the molecularly enhanced and karyotype-enhanced MYSEC systems for secondary myelofibrosis, known as MYSEC-mPM and MYSEC-kmPM. These tools go beyond traditional clinical variables such as age, hemoglobin concentration, white-cell count, circulating blasts and constitutional symptoms. They also incorporate genetic mutations and, in some cases, chromosome abnormalities. High-molecular-risk variants were found in 56.5 percent of the patients with primary myelofibrosis and 46.7 percent of those with secondary disease. Mutations classified as “UTS,” or unclassifiable high-risk mutations in the study’s framework, occurred in 31.5 percent of primary cases and 15.2 percent of secondary cases. The findings underscore how much hidden biological diversity can exist among patients who may appear clinically similar.

At the beginning of ruxolitinib treatment, the molecular scores separated patients according to their subsequent outcomes. Among people with primary myelofibrosis, five-year overall survival was 69.9 percent in one molecular risk group compared with 40.8 percent in another. In secondary myelofibrosis, the corresponding figures were 74.3 percent and 33.7 percent. Ruxolitinib is a JAK1 and JAK2 inhibitor that targets a signaling pathway frequently overactive in myelofibrosis. The drug can reduce spleen enlargement and relieve systemic symptoms, but response is not uniform, and treatment does not eliminate the abnormal stem-cell clone driving the disease. The molecular scores therefore provide a biologically grounded baseline estimate: they indicate how dangerous the underlying disease may be before or at the start of therapy. But the central question was whether the patient’s early response and clinical trajectory could add a second layer of prognostic information.

The researchers examined two six-month response-based models, RR6 and iRR6. Although the source study does not spell out every component of the scoring algorithms in its abstract, the distinction between them is clinically important. RR6 did not successfully divide survival outcomes among patients considered molecularly lower risk. iRR6, by contrast, produced meaningful separation after treatment had begun. In primary myelofibrosis, 65 patients could be evaluated with iRR6. Their estimated five-year overall survival was 100 percent for the low-risk group, 87.7 percent for the intermediate-risk group and 54.8 percent for the high-risk group, a difference that reached statistical significance with a p value of 0.033. In secondary myelofibrosis, 73 patients were evaluable, with five-year survival estimates of 100 percent, 81.4 percent and 60.4 percent across the low-, intermediate- and high-risk categories, respectively. The separation was also statistically significant, with a p value of 0.045.

A dynamic score such as iRR6 is conceptually different from a one-time diagnostic label. Baseline molecular testing is analogous to examining the engine of a car before a long journey: it reveals the machine’s inherent risks, but not how it behaves on the road. Early treatment data provide that second perspective. Blood counts, spleen response, symptoms and other indicators can show whether the disease is responding, remaining biologically active or producing complications despite therapy. By integrating information available during treatment, a dynamic model may detect patients whose risk has changed since diagnosis. Technically, this is a form of time-updated prognostication: the estimated hazard of death is recalculated using new observations rather than assuming that the original risk score remains valid indefinitely. In myelofibrosis, where clonal evolution and treatment resistance can alter the disease course, that distinction may be especially valuable.

The results do not mean that iRR6 replaces molecular scoring, nor do they establish that the model improves survival by itself. Instead, the study supports a complementary strategy. MIPSS70 and MYSEC-based systems were useful for defining prognosis at ruxolitinib initiation, while iRR6 refined that estimate after six months of treatment. This layered approach could help physicians identify patients who need closer monitoring, earlier referral for transplant assessment or consideration of alternative therapies. It may also prevent an overly reassuring baseline classification from delaying a change in management when treatment response is inadequate. Conversely, patients whose disease remains in the most favorable dynamic category might avoid unnecessary escalation while continuing appropriate surveillance. Because the study was an observational evaluation within a specific cohort, the scores should be viewed as tools for risk estimation rather than automatic treatment directives.

The work emerged from the RUX-MF study, registered as NCT06516406, and included patients treated at participating centers under institutional review-board oversight and the standards of the Helsinki Declaration. All participants provided informed consent. The research was supported by the Italian Ministry of Health, Ministero della Salute Ricerca corrente, BolognAIL and the EPPERMED2025-134 HOPE Consortium; open-access publication funding was provided through the CRUI-CARE Agreement at the University of Bologna. The authors reported several relevant relationships with pharmaceutical companies, including advisory, consultancy, speaking and research-support arrangements. Such disclosures do not invalidate the results, but they are important context when interpreting research involving a widely used targeted therapy and decisions about treatment intensification.

The next test for this strategy will be whether the findings hold in larger, independent cohorts and across different treatment settings. The study included 197 patients overall, but the numbers available for the iRR6 analysis were smaller—65 with primary myelofibrosis and 73 with secondary disease. The reported survival differences were clinically striking, yet estimates based on modest subgroup sizes can be unstable and need external validation. Future studies could also determine precisely which clinical and laboratory changes carry the greatest predictive weight, whether the score performs similarly with newer therapies, and whether repeated assessments beyond six months improve accuracy. Even with those questions unresolved, the study points toward a broader shift in cancer medicine: prognosis is becoming less like a permanent label and more like a continuously updated biological forecast. For people living with myelofibrosis, that change could make treatment decisions more responsive to what the disease is doing now—not only to what it looked like at diagnosis.