AUTHORS
John O. Mascarenhas, MD, Justin Johnson, BS2, Josef Prchal, MD3, Jennifer O'Sullivan4, Alessandro Vannucchi, MD5, Jyoti Nangalia, MBBS, PhD, FRCPath, FRCP6, Christian Salib, MD7, Marina Kremyanskaya, MD, PhD8, Abdulraheem Yacoub, MD9, Jean-Jacques Kiladjian, MD, PhD10, Mary Frances McMullin, MD11, Adam Mead, MBBChir12, Sonia Fox13, Anna Green14, Gillian Sanchez15, Amylou C Dueck, PhD16, Heidi Kosiorek, MS17, Alessandro Rambaldi, MD18, Ruben Mesa, MD19, Tiziano Barbui, MD20, Ronald Hoffman, MD21, Claire Harrison, DM22, Mohamed Salama, MD23
AFFILIATIONS
1 Icahn School of Medicine at Mount Sinai, New York, New York, United States
2 New York University Grossman School of Medicine, NYU Langone Health, New York, United States
3 University of Utah & Huntsman Cancer Center, Salt Lake City, Utah, United States
4 Guy's and St Thomas NHS Foundation Trust, London, United Kingdom
5 University of Florence, Firenze, Italy
6 Wellcome Sanger Institute, Hinxton, United Kingdom
7 Icahn School of Medicine at Mount Sinai, New York, New York, United States
8 Icahn School of Medicine at Mount Sinai, New York, New York, United States
9 University of Kansas, Kansas City, United States
10 Clinical Investigation Center, Saint-Louis Hospital, Paris, France
11 Queens University, Belfast, Belfast, Ireland
12 University of Oxford, Oxford, United Kingdom
13 University of Birmingham, Birmingham, United Kingdom
14 Guy's and St Thomas NHS Foundation Trust, London, United Kingdom
15 Icahn School of Medicine at Mount Sinai, New York, New York, United States
16 Mayo Clinic Scottsdale, Scottsdale, United States
17 Mayo Clinic Scottsdale, Scottsdale, United States
18 Foundation for Research Ospedale Papa Giovanni XXIII (FROM), Bergamo, Italy
19 Atrium-Levine/ Wake Forest Cancer Center / Advocate Health, Charlotte, North Carolina, United States
20 FROM - Fondazione per la Ricerca Ospedale di Bergamo ETS, Bergamo, Italy
21 Icahn School of Medicine at Mount Sinai, New York, New York, United States
22 Guy's and St Thomas' NHS Foundation Trust, London, United Kingdom
23 University of Texas, SanAntonio, San Antonio, United States
ABSTRACT
Background
Molecular response (MR) to cytoreductive therapy in polycythemia vera (PV) and essential thrombocythemia (ET) patients is influenced by treatment type, baseline JAK2 variant allele frequency (VAF), and mutational context. However, whether pretreatment bone marrow histology predicts treatment-specific MR remains unknown. We evaluated baseline paratrabecular adiposity (PTA= paratrabecular fat %) and trabecular abundance (TA= trabecular volume) as candidate predictors of MR, hypothesizing that these spatial features reflect biologically relevant differences in disease burden and may predict treatment responsiveness.
Methods
Pretreatment bone marrow core whole-slide images from 118 patients with JAK2 V617F-positive PV or ET enrolled in MPN-RC-111 (n=16), MPN-RC 112 (n=23), or MAJIC-PV (n=79) were analyzed. MPN-RC 111 included patients with hydroxyurea (HU) failure treated with pegylated interferon-α (PEG-IFN). MPN-RC 112 included treatment naive patients receiving PEG-IFN (n=13) or HU (n=10). MAJIC-PV patients had failed HU and received either ruxolitinib (RUX) (n=46) or best available therapy (BAT) (n=33). CALR, MPL mutated and triple negative patients were excluded.
PTA and TA biomarkers were evaluated by AI processing of images across four feature frameworks families: shell, tile-based, composite, and global, including a PTA-TA index. Associations with continuous and binary JAK2 VAF response endpoints were assessed using adjusted linear regression and Firth logistic regression, adjusting for age, sex, disease type, baseline JAK2 VAF, and treatment. False-discovery rate was controlled using q values.
Results
Global TA varied according to prior treatment history: MAJIC-PV and MPN-RC-111 cohorts had similar global TA (10.48% vs 9.45%; p=0.359), whereas treatment-naive MPN-RC-112 cohort had higher TA than both MAJIC-PV (12.45% vs 10.48%; p=0.024) and MPN-RC-111 (12.45% vs 9.45%; p=0.0062). Baseline PTA inversely correlated with baseline JAK2 V617F VAF, with stronger correlations in the MPN-RC cohorts (Spearman ρ = −0.61 to −0.71) than in MAJIC-PV (ρ = −0.20 to −0.37), indicating that lower PTA correlated with greater baseline clonal burden.
In the combined MPN-RC-111/112, each 1-standard deviation (SD) decrease in PTA was associated with a 43 percentage points greater maximum JAK2 VAF reduction (p=0.0195, q=0.0325). The strongest association involved the composite PTA-TA index, for which each 1-SD lower index associated with a 70.8-percentage-point greater maximum VAF reduction (p=0.00023, q=0.0014). PTA-TA remained significant among PEG-IFN-treated patients (62 percentage points; p=0.0022, q=0.022), but not in the smaller MPN-RC-HU-treated subgroup.
In MAJIC-PV overall, higher PTA was nominally associated with greater durable/last-assessment JAK2 VAF reduction (β=+20.0 percentage points per 1-SD, p=0.0094, q=0.13), while higher TA was nominally associated with durable 20% MR (OR=2.56, p=0.021, q=0.10). Among MAJIC-PV BAT-treated patients, greater tile-based TA was associated with greater durable VAF reduction (β=+36.6 percentage points per 1-SD, p=0.0010, q=0.024). This association was directionally consistent in both HU-based BAT (β=+40.0 percentage points, p=0.025, q=0.175) and non-HU BAT (β=+40.0 percentage points, p=0.035, q=0.069). MR was frequent among RUX-treated patients, but PTA and TA did not discriminate response.
Conclusion
Pretreatment bone marrow histotopography demonstrated treatment-context-specific associations with MR. Lower PTA and a lower composite PTA-TA index strongly predicted greater JAK2 V617F reduction in the MPN-RC 111/112 cohorts, particularly among PEG-IFN-treated patients, whereas greater TA predicted improved MR among BAT-treated patients in MAJIC-PV. The lower baseline TA observed in the HU-failure-enriched cohorts may reflect differences in prior treatment exposure and/or disease stage. These findings support further development of computational histotopographic biomarkers for MR stratification and biomarker-informed treatment selection in PV and ET. Validation is ongoing in a larger prospective cohort of IFN-treated patients.
DISCLOSURES
Justin Johnson, BS: No relevant conflicts of interest to declare.