Adding Patient-Specific Comorbidities May Improve Risk Evaluation in Myelofibrosis

Adding patient-specific comorbidities improved the prognostic effect of risk prediction models for patients with primary myelofibrosis (PMF) or secondary myelofibrosis (sMF), according to findings from an assessment of data collected in Vanderbilt’s Synthetic Derivative and BioVU Biobank comprehensive electronic health record (EHR).

Discrimination power was significantly higher using the extended Dynamic International Prognostic Scoring System (DIPSS) model that incorporated renal failure/dysfunction, intracranial hemorrhage, invasive fungal infection, and chronic encephalopathy (C-index, 0.81; 95% CI, 0.78-0.84) compared with the original DIPSS model (C-index, 0.73; 95% CI, 0.70-0.77).

All 4 of the comorbidities were individually associated with poorer survival (TABLE 1).In particular, renal failure may play a greater-than-expected role in patient outcomes.

“PMF remains a complex and challenging disease that will require a continued effort to improve patient outcomes. BioVU is unique as a fully annotated deidentified patient record of millions of patients, and to our knowledge, a similar deidentified data source, with this level of necessary annotation, is not available,” investigators wrote. “Still, we have demonstrated reliable identification of myelofibrosis within an EHR, and further implementation of natural language processing and data extraction algorithms are actively being pursued to leverage our ability to identify hematologic malignancy in these databases.”

There are several prognostic systems available for myelofibrosis. Each includes validated disease-specific parameters such as high molecular risk, peripheral blood counts, cytogenetics, and disease-specific clinical characteristics. However, none of the current models take comorbidities into account. Previous data have shown that increased comorbidity burden is associated with reduced overall survival (OS). These previous studies, however, included only a limited selection of selected comorbidities in each validated tool and failed to account for the broad assortment of comorbid conditions that affect prognosis and treatment decisions.

Investigators at Vanderbilt set out to build an unbiased EHR evaluation based on genotypic risk scores that makes use of DNA collected in the BioVU Biobank to determine the role comorbidities play in survival. In this analysis, investigators evaluated data collected from approximately 300,000 patients treated at Vanderbilt University Medical Center from 1995 to 2016. They reviewed patient phecodes to evaluate overall comorbidity burden for each patient, excluding the codes related to PMF, such as acute myeloid leukemia (AML) and variables dependent on Dynamic International Prognostic Scoring System (DIPSS), such as leukocytosis.

Investigators evaluated peripheral blood DNA collected in Vanderbilt’s biobank using established PMF risk assessors including DIPSS, DIPSS plus, the Genetics-Based Prognostic Scoring System (GPSS), and the Mutation-Enhanced International Prognostic Scoring System 70+ (MIPSS70+) along with comorbidities using EHRs and next-generation sequencing (NGS).

Age, race, gender, and clinical and laboratory parameters were automatically excluded.

In a cohort of 193 patients with PMF or sMF, investigators identified 374 phecodes at diagnosis. Investigators conducted risk score recapitulation for these patients using DIPSSA and DIPSS plus.

There was biobanked DNA for another 140 patients that was available for NGS. Investigators used this data to conduct risk score recapitulation for all 140 using the GPSS and MIPSS70+.

These methods included prognostic predictors such as score-specific cutoffs for age, circulating myeloid blasts, leukocyte count, hemoglobin, and platelets. Investigators used phenome-wide association study (PheWAS) to determine the correlation between OS and comorbidity burden. Investigators conducted PheWAS to evaluate how each phecode related to survival.

To further investigate how comorbidity influences survival in myelofibrosis, we developed a simple prognosis model that accounts for comorbidity burden at PMF diagnosis. In this model, Investigators segregated patients into quartiles based on cutoff values for the number of distinct phecodes at PMF diagnosis. The low-risk quartile included patients with 0 to 2 phecodes; the intermediate-low risk group included patients with 3 to 7 phecodes; the high-risk quartile included patients with 8 to 17 phecodes; and the very high-risk quartile included patients with more than 18 phecodes at diagnosis.

The median patient age at diagnosis was 59 years (range, 24-87) and women made up 42% of the population. The median OS was 39 months (range, 1-265). Twenty-three patients developed AML at a median of 37 months (range, 1-265). Forty patients received allogeneic hematopoietic stem cell transplantation at a median of 30 months from diagnosis. Of the 193 patients in the study, 158 were treated at Vanderbilt within 1 year of diagnosis and 35 were treated there at least 1 year after diagnosis.

Most patients (80.8%) had PMF, 15% had polycythemia vera, and 13.5% had essential thrombocythemia. Median follow-up was 4 years (range, 1-22).

As determined by DIPSS, the 5-year OS was 94% in the low-risk group, 91% in the intermediate-low risk group, 78% in the high-risk group, and 0% in the very high–risk group. DIPSS plus, the 5-year OS was 94% in the low-risk group, 93% in the intermediate-low–risk group, 81% in the high-risk group, and 0% in the very high–risk group. As assessed by MIPSS70+ (n = 113) the 5-year OS was 96% in the low-risk group, 93% in the intermediate-low group, 79% in the high-risk group, and 32% in the very high–risk group. (P ≤.0010).

Using GPSS (N = 140), the 5-year OS rate was 100% in the low-risk and intermediate low-risk quartiles, 83% in the high-risk group, and 77% in the very high–risk group (P = .03; TABLE 2).

“Our repurposing of an institution-wide biobank for hematologic malignancy evaluation, with the potential for clonal evolution assessment, is a novel use of a tool in a manner to study relatively large populations of otherwise rare myeloid disease,” lead study author Andrew L. Sochacki, MD, and coinvestigators wrote in the paper. “In aggregate, our findings suggest that a more objective measurement of patient-specific comorbidities is needed to best individualize therapy in this highly comorbid patient population.”

Reference

Sochacki AL, Bejan CA, Zhao S, et al. Patient-specific comorbidities as prognostic variables for survival in myelofibrosis. Blood Adv. 2023;7(5):756-767. doi:10.1182/bloodadvances.2021006318