Photo Credit: Love Employee

The risk for disease progression, thrombosis, and mortality in essential thrombocythemia can be predicted with “readily available” patient information.

The risk for essential thrombocythemia-related events in patients with the disease can be predicted using currently available clinical data, researchers wrote in a letter to the editor appearing in Blood Cancer Journal.

“Although considered the most indolent myeloproliferative neoplasm (MPN), essential thrombocythemia (ET) is linked to burdensome vasomotor symptoms and potentially fatal complications that include thrombosis, hemorrhage, and disease progression to myelofibrosis and aggressive myeloid neoplasms,” Ghaith Abu-Zeinah, MD, a hematologist and oncologist at the Richard T. Silver Myeloproliferative Neoplasms Center at Weill Cornell Medicine, New York, and colleagues wrote. “Prognostic measures to identify those at greatest risk for thrombosis, progression, and death in ET [events] are important for timely risk-adapted intervention with available treatments, and for development of interventional trials to improve event-free survival [EFS]. But predicting risks of events in ET has been difficult because ET is an uncommon and clinically heterogeneous chronic disease.”

Predicting excess mortality and disease progression related to ET presents challenges because such events often occur decades after a patient is diagnosed, the researchers continued.

“Thus, retrospective analysis of large cohorts with sufficiently long follow-up is required to identify prognostic measures to stratify risk in patients with ET,” they wrote.

The researchers used several previously established prognostic models designed for predicting patients’ risk for thrombosis (International Prognostic Score of thrombosis in Essential Thrombocythemia [IPSET]-thrombosis), as well as overall survival (IPSET-survival,

Mutation-Enhanced International Prognostic Scoring System-ET, and AAA [Age, absolute neutrophil, and absolute lymphocyte]) that were used in two large, recent retrospective studies of patients with ET. These studies also confirmed various risk factors for thrombosis, disease progression, and death in ET, including elevated white blood cell count, male sex, age 60 or older, low absolute lymphocyte count, and elevated absolute neutrophil count. Dr. Abu-Zeinah and colleagues applied these models and parameters to 328 adults with ET who were treated at the Weill Cornell Medicine Silver MPN Center in New York City. The median follow-up in the Weill cohort was 6 years.

Predicting Thrombosis Risk

The researchers reported 33 thromboses in the Weill cohort, with 27 occurring in patients with JAK2 mutations. In the two previously published studies, patients’ thrombosis risk was related to JAK2 mutations, prior thrombosis, and age. Similarly, Dr. Abu-Zeinah and colleagues reported that JAK2 mutation was associated with a significantly shortened thrombosis-free survival compared with CALR mutations in the Weill Cohort. The researchers wrote that this increased risk was “mainly due to thrombotic events” compared with patients with CALR/MPL mutations.

Also, in keeping with previously published research was the finding that the IPSET-thrombosis model did a poor job of discriminating between intermediate levels of risk and performed better at distinguishing between very low risk- and very-high risk-stratified patients, the researchers wrote (receiver operator characteristic [ROC] area under the curve [AUC], 0.63). Tools like the IPSET-thrombosis model are static and cannot account for therapies known to reduce risk, like glycemic and blood pressure control antiplatelet, anticoagulant, and/or cytoreductive therapy, Dr. Abu-Zeinah and colleagues noted. However, studies have proven that “true risk is dynamic.”

Predicting Progression to Myelofibrosis

CALR mutations are associated with reduced risk for thrombosis, but both the previously published studies and the current study found that CALR mutations and MPL mutations were both associated with significantly shorter myelofibrosis-free survival than JAK2 mutations, the researchers reported. Twenty-nine patients in the Weill cohort progressed to myelofibrosis, they continued, with a Cox multivariable analysis finding that patients who had CALR mutations had progressed nearly 12 times more often than those with JAK2 mutations, whereas those with MPL mutations progressed over 16 times more often (CALR/JAK2: HR=11.5; P=0.008; MPL/JAK2: HR=16.4; P=0.04). Progression to the blast phase of myeloproliferative neoplasm reportedly occurred in only 10 patients and was unrelated to driver mutations.

Predicting Overall Survival

Earlier researchers found that the AAA model, which the other two studies previously used, performed better than IPSET-survival when predicting OS. However, Dr. Abu-Zeinah and colleagues found that both were helpful for prognosticating overall survival (ROC AUC= 0.81 for AAA vs 0.77 for IPSET-survival, P=0.18).

The researchers wrote that Cox model analysis found that absolute neutrophil count was independently linked with mortality (HR=1.12, P=0.003) when mortality was used as a continuous variable but not when used as a binarized parameter. The neutrophil-to-lymphocyte ratio also appeared prognostic of OS.

“To our knowledge, the biological link between high NLR and MPN-related complications has not been elucidated,” Dr. Abu-Zeinah and colleagues added. With this in mind, they wrote that they “hypothesize that high absolute neutrophil count, low absolute lymphocyte count, and high neutrophil-to-lymphocyte ratio are proxy measures of aggressive disease biology.”

All told, Dr. Abu-Zeinah and colleagues wrote the three studies established several key ideas:

• Patients with ET and JAK2 mutations face an increased thrombosis risk compared with those who have other mutations, and patients whose disease is CALR-mutated rarely have thrombotic events.
• CALR and MPL mutations both significantly raise the risk of progression to myelofibrosis.
• Driver mutations are not a prognostic indicator of overall survival.
• The revised IPSET-thrombosis score is good for stratifying patients with very low and very high risk for thrombosis but generally “underperforms with real-world data.”
• OS, as calculated by AAA score, “and, to a lesser degree,” IPSET-survival score, were prognostic for OS. Age had the strongest influence on mortality risk, which the researchers wrote made neutrophil and lymphocyte counts especially interesting as prognostic factors.

“Addition of our [Weill Cornell Medicine] cohort with long-term follow-up to the recently published cohorts from [the Mayo Clinic] and [Florence Center Research and Innovation of Myeloproliferative Neoplasms] CRIMM provides high confidence in the risk factors identified, and the utility of available risk models for ET in long-term assessment of thrombosis, progression, and survival,” Dr. Abu-Zeinah and colleagues concluded.

“However, these studies also support the need for accurate, dynamic, near-term predictive models. Near-term prediction of risk is a necessary step toward identifying cohorts of patients for targeted clinical trials to identify new risk-mitigating interventions. Increasing global use of electronic medical records provides access to large-scale, longitudinal datasets that can be used with health informatics resources for developing dynamic predictive models. Amidst the recent boom in AI and machine learning and the increasingly available resources for automated data retrieval, we need to collaboratively prove that life-threatening events in chronic rare disease are indeed predictable and preventable.”