PROJECT SUMMARY
Myelofibrosis (MF) is a severe disease with a low median survival rate of six years and a declining quality of life.
JAK2 inhibitors have revolutionized treatment for patients with MF; however, only 30%-40% of patients respond
strongly to this therapy. Identifying non-responders early is vital to clinically recommend higher-risk by effective
alternative therapies that are not suggested as first-line treatments. Histopathologic examination of bone marrow
biopsies is the current standard for identifying non-responders. However, this process is failing patients as it
takes three years to reach a definitive conclusion. We aim to reduce this assessment period to six months,
thereby improving health outcomes for non-responders and increasing their chances of recovery before the
disease causes irreparable damage. This research is essential because non-responders comprise most treated
MF patients (60%-70%).
To accurately monitor treatment response, it is necessary to identify hematopoietic BM areas and recognize
abnormal reticulin-fibers patterns within these areas. However, current methods are inadequate, as they can
only identify non-responders reliably after three years. Therefore, we propose a new technology that addresses
all the challenges with MF treatment assessment in a single instrument. This technology is observer-
independent, label-free, and quantitative, filling a critical gap in the state-of-the-art. Although histopathology is
the gold-standard method, it is susceptible to high inter-observer variability in MF due to the qualitative nature of
visual inspections. This variability is because of the difficulty in recognizing fiber-pattern trends from
immunohistochemical staining data. Our technology overcomes these limitations by providing a comprehensive
and quantitative evaluation of the evolution of fibrosis in MF.
Our proposed method using mid-infrared spectroscopic imaging (MIRSI) technologies and machine learning
addresses an unmet clinical need in assessing the treatment response of MF, benefiting thousands of patients.
Our preliminary studies show that MIRSI can distinguish between different tissue subtypes, including
hematopoietic tissue, with high accuracy and create detailed maps of reticulin fiber disorganization within
hematopoietic areas, providing a precise visualization of fibrosis progression. We propose using a MIRSI-based
treatment response score to identify early trends in disease progression. We aim to validate our approach using
678 samples spanning multiple time points for each patient. This strategy will provide a robust numerical tracking
of fibrosis progression, facilitate early identification of non-responders, allow them to transfer to novel therapies
with disease-modifying activity, and improve the current treatment paradigm.
