Introduction: Primary myelofibrosis (PMF) is hallmarked by stromal bone marrow (BM) changes that include reticulin fibrosis, collagen deposition, osteosclerosis, and microvascular proliferation. Accurate evaluation of BM fibrotic changes is key for MF diagnosis and deserves prognostic implications. Gianelli et al. ( Histopathol 2017;71:89) developed a comprehensive score, named reticulin (MF), collagen (Co), osteosclerosis (Ost) score (RCOs), which was prognostically informative and more accurate than the 2005 European Consensus on Grading of Bone Marrow Fibrosis (ECGMF) in identifying high-risk patients (pts). Aims&Methods: This retrospective, single-center study aimed to validate the RCOs and explore its clinical, genetic and prognostic correlates in a large cohort of clinically and molecularly annotated PMF pts diagnosed according to 2022 ICC criteria. The RCOs was assessed by two independent experienced histopathologists (U.G, R.S.). Results: The study included 222 pts with a diagnosis of PMF: 121 (55%) prefibrotic PMF (pre-PMF), 101 (45%) overt PMF. Median age was 66 years (range, 18-91), 127 (57%) were male. BM histological analysis according to the RCO algorithm provided the following results: MF-0 n=20 (9%), MF-1 n=102 (46%), MF-2 n=47 (21%), and MF-3 n=53 (24%); Co-0 n=130 (59%), Co-1 n=33 (15%), Co-2 n=31 (14%), and Co-3 n=28 (13%); Ost-0 n=146 (66%), Ost-1 n=45 (20%), Ost-2 n=18 (8%), and Ost-3 n=13 (6%). Next, we looked at the distribution of the Co and Ost variables according to MF grade and PMF diagnosis. Overall, morphological features evolved harmonically, with pre-PMF showing the less severe stromal changes and higher reticulin fibrosis being associated with the highest values of Co and Ost. Computation of the RCOs returned the following results: RCOs 0 n=20 (9%), RCOs 1 n=94 (42%), RCOs 2 n=15 (7%), RCOs 3 n=19 (9%), RCOs 4 n=9 (4%), RCOs 5 n=16 (7%), RCOs 6 n=13 (6%), RCOs 7 n=17 (8%), RCOs 8 n=11 (5%), and RCOs 9 n=8 (4%). Median RCOs for pts with pre- and overt PMF was 1 (range, 0-2) and 5 (range, 2-9), respectively. We investigated the correlations of RCOs with clinical and molecular variables. Higher RCOs was associated with overt PMF (median, 5 vs 1, P<.0001), male gender (median, 3 vs 1, P=.0005), lower hemoglobin (Pearson's r, -0.5, P<.0001), lower platelets (Pearson's r, -0.4, P<.0001), higher blasts (Pearson's r, 0.4, P<.0001), higher LDH (Pearson's r, 0.5, P<.0001), and higher prevalence of splenomegaly (median, 3 vs 1, P<.0001) and constitutional symptoms (median, 5 vs 1, P=.0004). As to molecular variables, a higher RCOs was associated with abnormal karyotype (median, 3 vs 1, P=.0003), absent JAK2 mut (median, 2 vs 1, P=.0224) but higher JAK2 allele burden in JAK2-mut pts (Pearson's r, 0.3, P<.0006), triple negativity (median, 5 vs 1, P=.0223), EZH2 mut (median, 6 vs 1, P=.0033), ASXL1 mut (median, 6 vs 1, P<.0001), KRAS mut (median, 8 vs 1, P=.0059), and HMR status (i.e., ≥1 mut in ASXL1, EZH2, IDH1/2, SRSF2) (median, 5 vs 1, P<.0001). Median overall survival (OS) was 96 months (95% CI 81-134). In univariate Cox regression analysis, RCOs directly correlated with risk of death (HR 3.1, 95% CI 1.5-6.3, P=.0023). MF and Co variables correlated with OS (respectively: HR 4.0, 95% CI 1.8-8.6, P=.0003; HR 2.7, 95% CI 1.5-5.0, P=.0008), unlike Ost. ROC analysis with death as endpoint confirmed 5 as the optimal cut-off value ( Histopathol 2017;71:89). Pts with RCOs >5 (n=65 overt PMF) had significantly worst OS compared to pts with RCOs <5 (n=121/36 pre-/overt PMF), with median values of 64 vs 134 months (HR 2.3, 95% CI 1.4-3.6, P=.0004) (Fig. 1A). Considering only overt PMF pts, ROC analysis confirmed 5 as the more accurate cut-off value (univariate survival analysis: HR 2.1, 95% CI 1.1-4.2, P=.0257). We finally computed the MIPSS70 by replacing the original BM fibrosis grade ≥2 variable with the dichotomous RCOs 5 ( >/<5) in 177 (80%) pts with full molecular information, and calculated the C-index, Brier score, and time-dependent area under the curve (AUC) (Fig. 1B). The highest values for performance and accuracy in predicting death were achieved by the RCOs 5-integrated MIPSS70 at 12, 36, and 48 months. Conclusions: Our findings confirm the clinical and prognostic consistency of RCOs, supporting the importance of a comprehensive and accurate evaluation of BM stromal changes in PMF. Integration of RCOs within the MIPSS70 might improve the predictive performance of the model.
Read full abstract