Prognostic impact of bone marrow fibrosis and dyserythropoiesis in autoimmune hemolytic anemia

There is growing interest in the detrimental role of increased bone marrow fibrosis (BMF) in several hematologic conditions, including bone marrow failure and myelodysplastic syndromes as well as acute and chronic myeloid leukemia, chronic lymphocytic leukemia, and multiple myeloma.1, 2 No systematic studies on bone marrow features have been conducted in autoimmune hemolytic anemia (AIHA), where bone marrow compensatory response, usually evaluated with reticulocyte counts, is known to be one of the main mechanisms of the disease and has been shown to impact on its severity.3 Here we evaluated for the first time bone marrow histology in 47 primary AIHA patients, classified as cold agglutinin disease (CAD, DAT-positive for C), warm (WAIHA, DAT-positive for IgG or IgG + C), mixed (DAT-positive for IgG + C and high titer cold agglutinins), and atypical cases (DAT-negative, IgA-positive, warm IgM). Bone marrow trephine biopsies were revised by an expert hemopathologist and fibrosis graded (MF0 to MF3) according to WHO 2016.4 Figure 1 (left panel) shows baseline clinical and histological characteristics: median Hb, LDH, and reticulocyte values showed a great variability from highly pathological to normality. Thirty-nine patients (83%) had a bone marrow responsiveness index BMRI < 121 [patient's absolute reticulocyte count × (patient's Hb/normal Hb)], indicating inadequate reticulocytosis. Fourteen patients (30%) had normocellular, 30 (64%) hypercellular, and 3 (6%) hypocellular marrow; 12 (25%) displayed erythroid hyperplasia and 30 patients (64%) showed features of dyserythropoiesis. Notably, BMF MF1 was observed in 36% of cases (Figure 1A,B shows typical patients with MF1-fibrosis and dyserythropoiesis, respectively). Ten cases showed T-cell infiltrate, 7 B-cell, and 20 a mixed lymphoid population, without clonality, nor malignant features. Figure 1C,D shows bone marrow of a patient displaying increased B-cell infiltrate, and Figure 1E,F of another with T-cell infiltrate. Left panel: Clinical and hematologic characteristics of AIHA patients at baseline. LDH is expressed as fold upper the limit of normality (ULN); Ret: reticulocytes; IgG wAIHA: warm AIHA with direct antiglobulin test positive for IgG; IgG + C wAIHA: warm AIHA with direct antiglobulin test positive for IgG and C; CAD: cold agglutinin disease; Atypical AIHA cases were all DAT negative in this series. normo: normocellularity; hypo: hypocellularity; hyper: hypercellularity. *P < .05, **P = .005. Histogram: Cytokine serum levels in patients with (black) and without (white) bone marrow fibrosis (BMF). All cytokines were higher in patients without BMF, except for TGF-β. TGF-β = transforming growth factor beta; IL = interleukin; TNF-α = tumor necrosis factor alpha; IFN-γ = interferon gamma; MF = marrow fibrosis. Right panel: A, loose network of reticulin fibers with many intersections (MF-1, Gomori 20×). B, The hyperplastic erythropoietic series shows some grade of dyserytropoiesis in the erytrhocytes (arrows). C, Centro-lacunar lymphoid aggregate predominantly composed of CD20-positive small B lymphocytes, (D) with scattered small CD3-positive T lymphoid cells. E, Focal and interstitial lymphoid infiltrate composed mainly of CD3+ small T-lymphocytes, with very few scattered small CD79a+ small B-lymphoid cells (F) Comparing cases with BMF ≥ 1 (MF1) with those without (MF0), the former showed significantly increased reticulocyte counts (P = .02), bone marrow cellularity (P = .005), and dyserythropoiesis (P = .04). Patients with dyserythropoiesis were younger compared to those without (median 53, range 24–78, vs. 67, 31–80 years, respectively, P = .04), and showed higher prevalence of inadequate reticulocytosis, with a median BMRI of 56 (8–139), versus 88 (6–158, P = .02). To validate our findings, an independent cohort of 22 primary AIHA cases from a tertiary UK center was retrospectively analyzed (Figure 1, left panel). This cohort, comparable to the primary one concerning age, gender, and hematologic parameters, comprised a higher number of WAIHA. MF1 fibrosis was observed in 77% of cases and correlated with the presence of dyserythropoiesis and hypercellularity. A non-clonal lymphoid infiltrate was demonstrated in 82% of cases, mostly mixed (78%). To assess the role of bone marrow lymphoid infiltrate, several cytokines were investigated (Supporting Information Table 1S): TGF-β was increased in MF1 versus MF0 cases, whereas the other cytokines were reduced (statistically significant for IL-6, P = .033; Figure 1, histogram). IL-17 and IFN-γ positively correlated with LDH (r = 0.4 and r = 0.36 respectively, P < .05), and negatively with age (r = −0.38 and r = −0.37 respectively, P < .05). Finally, TGF-β (P < .001), IL-6 (P = .033), IL-10 (P = .05), and IL-17 (P = .01) levels were higher in all AIHA patients versus controls, whereas TNF-α and IFN-γ were comparable. Concerning therapy (Supporting Information Table 2S), all patients but 7 received steroids, with a response rate of 77.5%, and 28 cases (70%) required further treatment, including rituximab (24), splenectomy (3), and immunesuppressors (5). Twenty-three cases had blood transfusions, and four received erythropoietin. No differences were observed among MF0 and MF1 cases in either usage or efficacy of steroid and rituximab. Notably, MF1 cases more frequently required 2nd or further therapies (70% vs. 53%). Moreover, cases with hypercellularity and dyserythropoiesis required higher number of therapy lines (67% vs. 47%, for both), and showed shorter median relapse-free survival [RFS, 600 days for hypercellular vs. 1250 for normo/hypocellular cases; 5 years RFS 17% (95% CI 6–35) vs. 35% (95% CI 15–61); and 750 days for dyserythropoietic cases vs. 1250 in patients without; 5 years RFS 20% (95% CI 8–39) vs. 29% (95% CI 11–56)]. Splenectomized cases were all MF0 with 100% response, and cytotoxic immunesuppressors induced a response in MF0 patients only (67% vs. 0% in MF1). Finally, patients with CD3 + T-cell infiltrate showed 67% response rate to immunesuppressors, compared to 0% in those without; moreover, cases with CD20+ B-cell infiltrate showed 100% response rate to rituximab, versus 89% in the other patients. As regards AIHA-related complications (Supporting Information Table 2S, 6 infections, 7 thrombosis, 1 acute renal failure, and 4 cases of concomitant thrombocytopenia), thrombosis were more common in MF0 cases (20% vs. 6%). In conclusion, we firstly report the presence of BMF ≥ MF1 in more than 1/3 of AIHA patients, along with increased marrow cellularity and dyserythropoiesis in at least 2/3 of them, and confirmed the findings in an independent cohort. The higher prevalence of reticulin fibrosis in the validation cohort shows that this finding is not infrequent and excludes any center bias. Concerning hypercellularity, it was increased in total and not only in the erythroid lineage, and didn't correlate with increased reticulocytes, possibly reflecting a broader immune activation beyond marrow compensation of anemia. Interestingly, patients with marrow fibrosis, hypercellularity, and dyserythropoiesis more frequently required 2nd or further therapy lines, confirming the negative prognostic value of fibrosis already observed in other malignant and non-malignant hematological conditions, like hypoplastic-myelodysplastic syndromes (MDS).1, 2 Hypercellularity correlates with acute myeloid leukemias progression, and reduced overall survival in high risk MDS, possibly reflecting a higher proliferating and inflammatory marrow milieu.1 Accordingly, in our series, hypercellularity correlated with shorter RFS compared to hypo/normocellular patients. The same was observed for the presence of dyserythropoiesis, which correlated with reduced bone marrow responsiveness, a known predictor of treatment refractoriness and high risk of fatality in AIHA patients.3 Here we show that clinical–morphological features of chronic refractory AIHA resemble those of hypoplastic-MDS and of the novel idiopathic cytopenia/dysplasia of uncertain significance. As lately reported, refractory/relapsing autoimmune cytopenias may lose their main peripheral pattern (hemolysis, thrombocytopenia, neutropenia) and shift toward refractory cytopenia/dysplasia, possibly because of autoimmunity against marrow precursors, increased apoptosis, and unfavorable cytokine microenvironment.5 Consistently, TGF-β was the only cytokine increased among those tested, in MF1 cases. TGF-β belongs to a superfamily of soluble mediators that play a crucial role in reticulin and collagen fibers deposition both in autoimmune models (eg, systemic sclerosis) and in hematologic diseases (myeloproliferative diseases, myelodysplastic syndromes, etc). Moreover, activin, member of TGF-β superfamily, has recently been targeted with luspatercept in MDS with ring sideroblasts and SF3B1 mutation, leading the way to new molecular approaches.6 Contrarily, MF0 cases present with a burden of inflammatory cytokines (IL-6, TNF, IFN-γ, IL-17) along with a florid hemolytic disease, and more frequent thrombosis. Although the physiopathology of thrombosis in AIHA is largely unknown, it may complicate up to 10% of cases,3 mostly with marked intravascular hemolysis and complement activation. Cytokine storm might contribute to endothelium/platelets/coagulation activation in these patients with hemolytic/immune hyperactive disease. These cases are more responsive to immunesuppressors, as also observed for those with prevalent CD3 + T-cell infiltrate, whereas patients with CD20 + B-cell infiltrate showed better response to rituximab. Although preliminary, our data suggest to carefully evaluate fibrosis, cellularity, dyserythropoiesis, and lymphocyte infiltrate in AIHA. These features may help in harnessing therapy either avoiding excess immunesuppression or suggesting new cytokine-modulating agents. Whether bone marrow characteristics may predict response to therapy and/or disease evolution from autoimmunity to bone marrow failure syndromes is an interesting hypothesis to be confirmed in larger prospective studies. BF and WB followed patients, collected and analyzed data, wrote the paper and revised it for intellectual content. AZ performed laboratory analysis, collected and analyzed data, wrote the paper and revised it for intellectual content. UG revised bone marrow histology samples, prepared the histologic figures and revised the paper for intellectual content. AK provided data for the validation cohort, and revised it for intellectual content. AC and AZane revised the paper for intellectual content. No conflict of interest to declare. No funding sources to disclose. Additional Supporting Information may be found online in the supporting information tab for this article. Supporting Information Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.