Fat Fraction MRI for Longitudinal Assessment of Bone Marrow Heterogeneity in a Mouse Model of Myelofibrosis

Abstract

Background/objectives: Myelofibrosis (MF) is a myeloproliferative neoplasm characterized by the replacement of healthy bone marrow (BM) with malignant and fibrotic tissue. In a healthy state, bone marrow is composed of approximately 60-70% fat cells, which are replaced as disease progresses. Proton density fat fraction (PDFF), a non-invasive and quantitative MRI metric, enables analysis of BM architecture by measuring the percentage of fat versus cells in the environment. Our objective is to investigate variance in quantitative PDFF-MRI values over time as a marker of disease progression and response to treatment.

Methods: We analyzed existing data from three cohorts of mice: two groups with MF that failed to respond to therapy with approved drugs for MF (ruxolitinib, fedratinib), investigational compounds (navitoclax, balixafortide), or vehicle and monitored over time by MRI; the third group consisted of healthy controls imaged at a single time point. Using in-house MATLAB programs, we performed a voxel-wise analysis of PDFF values in lower extremity bone marrow, specifically comparing the variance of each voxel within and among mice.

Results: Our findings revealed a significant difference in PDFF values between healthy and diseased BM. With progressive disease non-responsive to therapy, the expansion of hematopoietic cells in BM nearly completely replaced normal fat, as determined by a markedly reduced PDFF and notable reduction in the variance in PDFF values in bone marrow over time.

Conclusions: This study validated our hypothesis that the variance in PDFF in BM decreases with disease progression, indicating pathologic expansion of hematopoietic cells. We can conclude that disease progression can be tracked by a decrease in PDFF values. Analyzing variance in PDFF may improve the assessment of disease progression in pre-clinical models and ultimately patients with MF.

Keywords: bone marrow; hematopoietic stem and progenitor cells; myelofibrosis; myeloproliferative neoplasms; proton density fat fraction; quantitative MRI.

Figure 1

Study population for disease groups non-responsive to administered treatments. Disease group 1 received ruxolitinib, fedratinib, navitoclax, or vehicle control. Disease group 2 received ruxolitinib, balixafortide, both, or vehicle control. We randomly selected mice from each disease cohort, and among treatments, with the sole inclusion criterion being failure of treatment to reduce splenomegaly or hypercellular bone marrow as determined by MRI. All mice from the healthy group were at pre-treatment time points and are not represented in this figure.

Figure 2

Effective therapy increases PDFF values in the distal tibia. Axial mean trajectory graphs for two different mice undergoing effective (top) and ineffective (bottom) treatment. The baseline time point (day 0) is in blue, day 64 in yellow, day 69 in green, and day 74 in red. The grey lines denote the BM region: left is proximal, middle is transition, and right is the distal region. Note that the slight differences in the y-axis are due to the biological variability in the distal region of each mouse. The proximal region on the left of the graph shows minimal change due to the proximal tibia containing normal hematopoietic cells indistinguishable from malignant cells, as determined by PDFF-MRI. We did not analyze the middle transition region due to the high noise. The distal region on the right side of the graphs demonstrates an increase in PDFF values for effective treatment and a decrease in PDFF values for ineffective treatment. We focused this study on the distal region because the expansion of malignant hematopoietic cells in MF replaces normal BM fat as the disease progresses.

Figure 3

Ineffective treatment results in low variance in PDFF values in the distal tibia of mice with MF. Heatmap of the mouse tibia over the study period of 74 days. Values of 0 and 100 define no and 100% fat cells, respectively. Effective treatment (top) shows an increasing number of fat cells as the BM progresses towards a healthy baseline. Ineffective treatment (bottom) shows almost no heterogeneity with low amounts of fat as measured by PDFF. Muscle tissue may appear different for each scan date due to the positioning of the tibia in the coil, but the same slice is displayed for each time point. The PDFF ROI and heatmap are overlayed onto the T1-weighted anatomic image for display clarity.

Figure 4

Histology shows the replacement of normal bone marrow fat in the distal tibia BM in mice receiving ineffective treatment for MF. Pink areas represent collagen in bone; white shows fat cells in healthy mice; and purple areas show hypercellular bone marrow. Mice with MF show hypercellular bone marrow, replacing the normal fat cells and other stromal cells seen in healthy mice.

Figure 5

Disease 1 and disease 2 cohorts of mice show no difference in PDFF variance. Nonparametric two-sided permutation test to determine significant difference between disease 1 and disease 2 groups. p-value of 0.8092 at a significance level of 0.05. There is no significant difference between the two diseased cohorts of mice, allowing us to combine these groups into one disease cohort for analysis.

Figure 6

Mice with MF significantly differ from healthy mice by variance in PDFF MRI. Nonparametric one-sided permutation test between pre-treatment MF disease and healthy group baseline values. Variance in PDFF values differs significantly between groups (p = 0.0001).

Figure 7

Principal component analysis reveals no difference in variance in PDFF MRI values for disease 1 and 2 groups. We performed PCA to assess variance patterns across time points for each mouse within the disease 1 and disease 2 cohorts. The lack of clustering or distinction between the two treatment groups shows that the disease groups can be grouped into one cohort for further analysis (p = 0.135).

Figure 8

One-dimensional principal component analysis shows a large difference in PDFF variance between the combined disease and healthy groups. Variance in PDFF values for the combined disease groups spans the entire x-axis. The healthy values cluster on the negative x-axis, exhibiting a significant difference (p = 0.0014).

Figure 9

Hypercellular BM in mice with MF shows minimal variance in PDFF values. For each mouse, we calculated the variance of every voxel at each time point, presented as violin plots to show variability of these values. The y-axis is capped at 300 to maintain a consistent scale across all groups; however, note that some outliers in disease groups 1 and 2 exceed this limit. The dashed lines represent the interquartile range. The bottom line is Q1, the middle line is median and the top line is Q3. Thelarge spread in day 0 in the disease groups is due to the irradiated BM, with a drastic drop in variance as disease state progresses. The healthy group shows a spread in variance from 0 to 100 (mean 49.0%) fat cells in the BM.

Figure 10

Variance in PDFF decreases in mice that are non-responsive to treatment for MF. We generated lines of best fit by combining the linear regressions of each mouse’s variance over time. The healthy single-time-point study shows an average variance value across all samples of 49.0%. The large variability in the disease groups can be seen by the high initial value and rapid negative slope. The disease BM surpasses the healthy baseline as fat cells are replaced by hypercellular bone marrow with unregulated proliferation of hematopoietic cells.