Abstract
Background: Splenomegaly due to extramedullary hematopoiesis (EMH) is a hallmark of myelofibrosis (MF). Previous studies of splenectomized MF patients highlight the distinct pathological changes found in MF spleen, including nodular vs diffuse tri-lineage hematopoiesis, clusters of atypical megakaryocytes (MKs), and variable reticulin/collagen fibrosis. Recently, we and others have shown that mesenchymal stromal cells (MSCs) and clonally-derived fibrocytes contribute to BM stromal remodeling in MF; however, the role of fibrocytes/MSCs in MF splenic EMH/remodeling, and correlation with clinical outcome, has not been evaluated. We performed multiplexed imaging along with targeted mutational profiling on a large, unpublished cohort of MF spleen samples to investigate fibrocyte/MSC abundance and correlation with clinicopathological outcomes, including in response to JAK inhibitor (JAKi) therapy.
Methods: Spleens from N=31 MF patients who underwent splenectomy at MD Anderson (1989–2015), along with matched BM (N=22), were compared with healthy donor BM/spleens (N=7/N=8; commercially obtained). Liver biopsies were also available on a subset of MF patients (N=12). Standard H&E, Gomori silver, and Masson's trichrome stains were performed on all samples. Fibrocyte (CD45⁺/CD68⁺/Procollagen-I⁺) and MSC (CD90⁺/CD105⁺) numbers, along with reticulin/collagen density, were determined using a 7-color Opal multiplex immunofluoresscence assay and quantitation performed using deep-learning software (Visiopharm) trained on red/white pulp spleen parenchyma. Matched targeted mutational profiling of 26 MF-associated genes on both BM and spleen was available for all patients.
Results: Primary MF (N=23), post-PV (N=4), and post-ET (N=4) patients were included in this cohort. BM fibrosis grade at time of splenectomy included MF-1 (N=4), MF-2 (N=6), and MF-3 (N=19). Driver mutations included JAK2V617F (N=20), CALR (N=6), MPL (N=1), and triple-negative (N=4) MF. Review of spleen histology confirmed prominent EMH within red pulp in all patients, with diffuse tri-lineage hematopoiesis, focal erythroid–myeloid nodules (N=3) and/or dysplastic MK clusters (N=6) among a subset of individuals as observed with previous studies. Expanded CD34⁺ immature blast populations were also seen in N=3 patients. In paired samples, JAK2V617F variant allele fraction (VAF) was 1.4-fold higher in BM than in spleen (65.4% vs 47.4%; p=0.008).
Surprisingly, and in contrast to what we previously observed in BM, both fibrocyte (26.4 vs. 85.8 cells/mm²; p=0.002) and MSC frequencies (77.2 vs 120.7 cells/mm²; p=0.03) were significantly reduced in MF spleen compared to healthy controls. Similarly, quantification of reticulin fibers as determined by Visiopharm showed no significant difference in overall reticulin (6.0% vs 6.1%) or collagen (1.3% vs 0.9%) density within spleen parenchyma despite dense EMH. In contrast, fibrocytes were inversely increased in matched BM (193.8 vs 28.5 cells/mm²; p=0.001) and correlated with increased BM reticulin fibrosis (3.9% vs 0.5%; p<0.001). Liver biopsies, not previously evaluated in MF, also demonstrated greater fibrocyte fractions (13.9 vs 3.9 cells/mm²; p=0.034) and collagen fibrosis (4.5% vs 0.8%; p=0.004) compared to healthy liver. While absolute fibrocyte numbers were reduced in spleen, their presence still strongly correlated with BM reticulin fibrosis grade (p<0.05) and other markers of advanced disease, including transfusion dependence, leukocytosis, and increased circulating blasts.
We then evaluated spleen histology in response to JAKi. Of 31 patients, 9 (29%) had received a JAKi for at least 12mo prior to splenectomy. All demonstrated persistent/progressive splenomegaly despite therapy. Significant reductions in spleen parenchyma reticulin fiber density (6.9% vs 3.7%; p<0.001) were observed in patients who had received JAKi compared to those who had not. Interestingly, however, absolute fibrocyte numbers in these patients were significantly increased, suggesting that while splenic fibrocytes may not contribute substantially to stromal remodeling, they may influence spleen response with JAKi.Conclusions: MF spleens display reduced fibrocyte/MSC numbers compared to healthy spleen, in contrast to what we observe in BM and liver. Splenic fibrocyte numbers correlate with advanced MF and are increased following JAKi exposure. Additional studies are needed to better understand the cell types contributing to splenic stromal remodeling in MF.
