Introduction: The first-line therapy for patients with low-risk myelodysplastic syndromes (MDSs) commonly consists of erythropoietin stimulating agents (ESAs), with a response rate ranging from 34 to 62%. For nonresponder patients, outside clinical trials, blood transfusions are the most frequent therapeutic option, with detrimental effect on the quality of life and with risks of iron-overload. Since no studies have been yet conducted on this topic, we investigated the potential predictive role of bone marrow (BM) histological evaluation in patients treated with ESAs. Materials and Methods: We performed a morphological and immunohistochemical retrospective analysis of BM biopsies of 96 patients with low-risk MDSs subsequently treated with ESAs. Results: In our series, substantial morphological overlap was found between responder and nonresponder patients. On the contrary, patients with a percentage of CD34-positive blasts >3% or with p53 protein expression <1% responded with a significantly higher frequency to ESAs. Conclusions: Our study reinforces the role of BM biopsy as diagnostic tool in MDSs, being also able to supply information related to response to ESAs and to its loss over time.
Myelodysplastic syndromes (MDSs) represent a heterogeneous group of clonal myeloid disorders charac-terized by dysplasia in at least one of the hematopoietic lineages, ineffective hematopoiesis, variable peripheral blood cytopenias, and an increased risk to develop AML. MDSs occur mainly in older adults (median age: 70 years) with male predominance and annual incidence of 3–5 cases per 100,000 population. The newest edition of WHO classification of tumors of hematopoietic and lymphoid tissue (2017) recognizes different entities based on lineage dysplasia (single or multilineage), presence of ring sideroblasts, the percentage of blasts, identification of isolated del (5q) and, besides other factors, the age of patients (refractory cytopenia of childhood). Several attempts have been made to propose scoring systems for a better stratification of patients and prediction to therapy response. In particular, the International Prognostic Scoring System (IPSS) for MDS is generally used to characterize the patient’s prognostic risk status in clinical practice. Considering the blast percentage, the karyotype, and the number of cytopenias, this score stratifies patients in 4 risk categories which can be clinically simplified as low-risk (comprising the low-risk and intermediate-1 IPSS categories) and high-risk MDS (comprising the intermediate-2 and high-risk IPSS subgroups). For patients with low-risk MDS, first-line therapy commonly consists of erythropoietin (EPO) stimulating agents (ESAs) , with or without granulocyte colony-stimulating factor, with a response rate ranging from 34 to 62% [2-4] when evaluated by using the 2000 and 2006 International Working Group (IWG) response criteria [5, 6]. For nonresponder patients, outside clinical trials, blood transfusions represent the most frequent therapeutic option, with detrimental effect on quality of life. Moreover, this therapeutic management frequently exposes patients to iron-overload with subsequent need of specific therapy . In order to predict the likelihood to response to ESAs treatment, in the recent years, attempts have been made to build score systems useful for the clinical practice. EPO levels represent the most widely accepted parameter although without a uniform cutoff when comparing different studies. To date, the European guideline recommends the administration of ESAs to patient with <200 IU/L despite representing the 86% of patients with MDS . Nowadays, the Hellström-Lindberg score represents the validated scoring system commonly used in the clinical practice to predict the response rate by integrating the transfusion need and EPO serum level . In addition to flow cytometry (FCM) and bone marrow (BM) aspirate, BM biopsy is an essential tool for the diagnosis and follow-up of patients affected by MDS. The histological report should include at least the evaluation of the overall BM cellularity in relation to the patient age, the quantity and quality description of the 3 hematopoietic lineages, the blast count, and the grade of BM fibrosis. In addition, immunohistochemical assessment of p53 has been established in subsets of MDS as prognostic and predictive of disease resistance, most notably to lenalidomide in MDS with isolated del (5q) and to progression to AML in the course of azacitidine treatment . To our knowledge, no studies have been yet conducted investigating the potential predictive role of the histological analysis of the BM toward the response to ESAs treatment in MDS patients. In this study, we hypothesize that a detailed evaluation of BM biopsy could help in the identification of MDS patients who can benefit from ESA treatment. To verify this hypothesis, we performed a morphological and immunohistochemical analysis of the BM biopsies of a large series of low-risk MDS patients aiming to describe their morphologic profiles and to identify potential predictive factors to ESAs therapy.
Materials and Methods
We enrolled 96 consecutive MDS patients with IPSS low or intermediate-1 risk diagnosed between 2000 and 2017 and treated with EPO α 20,000–80,000 IU/week or darbepoetin 150–300 μg/week. For each patient, all the clinical data, duration of treatment, Hellström-Lindberg score, and follow-up data were collected together with a BM biopsy performed within 8 weeks before starting the treatment. For each patient, the blast count evaluated by FCM analysis and on BM aspirate was also collected. The response to therapy was evaluated according to the 2006 IWG guidelines .
Bone Marrow Morphologic Evaluation
Each BM biopsy was stained with hematoxylin-eosin, Giemsa, and Gomori’s silver impregnation. Two experienced pathologists (FB and UG) reviewed all the slides at a multiheaded microscope. The following morphological variables concerning the dyserythropoiesis were evaluated: megaloblastoid changes (elements characterized by at least 1.5 times the size of a normal proerythroblast with finely dotted chromatin and increase of the nucleus-cytoplasmic ratio), left shifting, cytoplasmic vacuolization, topographic abnormalities, and nuclear alteration (including budding and multinuclearity). To define dysgranulopoiesis, nuclear hypo- or hyper-segmentation and left shifting were described; meanwhile, the presence of topographic abnormalities, micromegakaryocytes (mononuclear elements with nuclear diameter of 7–10 μm), megakaryocytes with hypolobated nuclei, and/or multinucleated ones were quantified to assess the megakaryocyte dysplasia. Each specific alteration was carefully investigated among the entire specimen and considered as “present” if clearly identifiable at least 3 times by both the reviewers. Moreover, the overall BM cellularity in relation to patient’s age, the myeloid/erythroid ratio, and the entity of marrow fibrosis (determined according to the EUMNET consensus)  were described. Percentage assessment of immunohistochemical p53 intense (3+) positive (Dako Omnis DO-7) cells, CD34-positive (Ventana, Oro Valley, AZ, USA) blasts, and CD20 (Dako Omnis L26) and CD3-positive (Dako Omnis Polyclonal antibody) lymphocytes was performed by using the automatic system BenchMark XT (Ventana Medical Systems, Oro Valley, AZ, USA). Reactions were revealed using the UltraViewTM Universal DAB, a biotin-free, multimer-based detection system, according to the manufacturer’s instruction. The evaluation was performed among the entire specimen, considering as “positive” only the cells with intense nuclear expression of p53 as previously published  and by focusing on the quantification of precursors expressing CD34 and on their distribution in clusters (>5 immature myeloid cells). The lymphocytic infiltrate was described as composed mainly by B or T lymphocytes or by a mixed population, and was then quantified and defined if present in nodules, micro aggregates or with an interstitial pattern.
The entire specimen was examined in order to obtain a representative percentage of the above antibody expression. In case of disagreement among the 2 reviewers, the specimen was reevaluated entirely until a consensus was reached. Finally, all cases were classified according to the updated 2017 WHO classification considering all together the clinical informations, BM aspirate smear, BM biopsy, FCM, and cytogenetic data .
Statistical analysis was performed with IBM SPSS Statistics, version 24.0 (IBM Corp., Armonk, NY, USA). Statistical significance was defined as p < 0.05. Distribution normality was assessed with the Shapiro-Wilk test. A multivariate logistic model was constructed considering the response to treatment as a dependent variable. The selection of univariate variables was achieved using the χ2 test for nondichotomous quantitative variables and Fisher’s exact test for dichotomous quantitative variables. For continuous variables, the ROC curve was used to calculate the Youden index and obtain a threshold value to be used for a subsequent Fisher’s exact test.
Clinical Features of “Responders” and “Nonresponders”
According to the 2006 IWG guidelines , patients were classified as “responders” (n = 65) and “nonresponders” (n = 31) to the ESA therapy. The main clinical features of the patients in the 2 categories are reported in Table 1. Among “responders,” 36 patients (55%) were male and 29 (45%) were female with 50 (78%) were alive at last follow-up (median 35 months, min–max: 1–192 months). According to the Hellström-Lindberg score, 55 (86%) were stratified as “good” and 9 (14%) as “intermediate.” Moreover, the mean serum EPO level was 81.8 mU/mL (median value 53.85 mU/mL). The “nonresponders” patients were 21 (68%) male and 10 (32%) female with 20 (65%) patients alive at last FU. According to the Hellström-Lindberg score, 19 (61%) were stratified as “good” and 12 (39%) as “intermediate.” In this subgroup, the mean serum EPO was 139.1 mU/mL (median value 73.3 mU/mL). Significant differences between the 2 groups were found considering the Hellström-Lindberg score and IPSS score. In detail, a “good” Hellström-Lindberg score was found in 55 (86%) responders patients and 19 (61%) nonresponders (p = 0.009 and p = 0.034 on univariate and multivariate analysis, respectively), while IPSS low-risk patients category comprised 40 (63%) responders and 19 (48%) nonresponders (p = 0.029 and p = 0.013 on univariate and multivariate analysis, respectively).
Morphologic Features of “Responders” and “Nonresponders”
“Responders” and “nonresponders” patients did not differ significantly in their morphological profile (Table 2). Hence, “responders” more frequently displayed increased total cellularity (52%), megaloblastic changes (63%), micromegakaryocyte (78%), and hypolobated megakaryocyte (92%). Likewise, “nonresponder” patients presented increased cellularity (61%), megaloblastic changes (77%), and nuclear alteration (68%) in the erythroid precursors, hypolobated megakaryocyte (90%), and topographic abnormalities (61%) in the megakaryocytic lineage.
The CD34-Positive Blast Percentage and the p53 Expression Are Associated with the Response to Erythropoietin Stimulating Agent Therapy
The percentage of CD34-positive blasts showed a mean value of 3.6% among “responders” (median 3%) and 2.6% among “nonresponders” (median 2%) (range: <1–9% for both subgroups). The threshold value of 3% CD34-positive blasts obtained by the ROC curve analysis correlated with the response to ESA therapy: patients with a CD34-positive blast count >3% responded with a significantly higher frequency to the ESA therapy (univariate analysis: p = 0.021 and multivariate analysis: p = 0.003) irrespectively from the other clinical features including also WHO subgroup, IPSS and Hellström-Lindberg score, and blast percentage on flow cytometric or smear.
Similarly, a “cut-off” value in the expression of p53 was defined with ROC curve analysis (mean value: 1, median value: 0.5, and range: 0–5% for “responders”; mean value: 1.8, median value: 1, and range: 0–6% for “nonresponders”). Patients with p53 expression <1% of the nucleated cells were associated with higher rate of response to treatment (univariate analysis: p = 0.009 and multivariate analysis: p = 0.005). The qualitative or quantitative assessment of lymphocytic infiltrate evaluated with CD20 and CD3 antibodies did not correlate with the response. No statistical relation was found considering CD34-positive cells and p53 expression in the same patients. Neither CD34-positive blasts by flow-cytometric evaluation (p = 0.412) nor morphological blast count obtained on BM aspirate (p = 0.079) resulted informative for the response to ESAs, despite showing a similar trend when considering the patients according to their different risk categories (Table 3).
In this study, we retrospectively analyzed the BM histology of 96 patients with low-risk MDS (low and intermediate-1 risk, according to IPSS), to evaluate if their morphological and immunophenotypic features at diagnosis could be informative of the response to treatment with ESAs. Our detailed analysis documented that the clinical features and the morphologic profiles of the BM are similar in both “responder” and “nonresponder” groups of patients. The Hellström-Lindberg score resulted the only clinical features significantly able to predict the response to ESA treatment in our cohort, showing that this cohort is representative of real-life population. Interestingly, important predictive information can be obtained from the histological evaluation of the CD34-positive blast and the evaluation of the immunohistochemical expression of p53 by hematopoietic precursors. In detail, patients with a percentage of CD34-positive blasts >3% respond with a significantly higher frequency to ESA therapy. To date, no data are available in literature concerning the specific role of CD34-positive blasts as predictive value of response to ESAs. In their study, Park et al.  suggested a better response to ESAs in patients with <5% of blasts rated on BM aspirate, but their series included, compared to ours, also 6% of patients with >10% of blast in the BM aspirate, which cannot be diagnosed as low-risk MDS. On the other hand, it is well known that receptors for EPO are present on the cell surface of blast cells and that their expression decreases in more mature and differentiated elements with erythroid phenotype [14, 15]. Consequently, a higher proportion of CD34-positive precursors at diagnosis could also represent a higher proportion of therapeutic “targets,” potentially able to respond effectively to the maturational stimulus of ESAs treatment. The same result cannot be achieved considering, in the same cohorts, the blast count evaluated by FCM or by the morphological analysis of the BM aspirate. This evidence is possibly related to the eventuality of an imprecise blast count due to peripheral blood dilution, occurring in the preanalytical phase of these 2 techniques [16, 17]. This hypothesis could also be supported by the overall lower percentage of blast detected by these 2 procedures compared to BM biopsy (Table 3). Another explanation could be found in the irregular distribution of the blasts in the BM, an evidence frequently observed in our series supporting, once more, the importance of the histological evaluation which should always be performed on satisfying specimen in order to obtain a representative picture of this disease (Fig. 1). The role of p53 in MDSs has already been widely analyzed in some studies. p53 is normally expressed at low levels in phenotypically normal cells, as it serves as a tumor suppressor gene, mastering pathways involved in blockade of cellular cycle and promotion of DNA damage. Its abnormal expression can be observed in hematopoietic cells in several conditions, ranging from viral infections (i.e., parvovirus) to congenital condition to, most notably, myeloid neoplasms. Underlying mechanisms involve physiologic response to cellular stress, ribosomal haploinsufficiency, and, most importantly, TP53 mutation and/or 17p locus aberration (deletion). In MDS with isolated deletion of chromosome 5q, an in vitro study pointed out that a nuclear accumulation of p53 within erythroid cells is responsible for cell death by induction of apoptosis . Another study evaluated how, in cell cultures deriving from patients with MDS with deletion of the 5q chromosome, the suppression of the p53 gene leads to an increase in erythropoiesis for a blockage of the cell apoptosis . The prognostic significance of p53 expression related to ESA treatment is still unclear, and our study shows how patients with p53 expression in <1% of nucleated elements respond with a significantly higher frequency to EPO therapy. We can speculate that this cut-off value could identify a subgroup of patients with a sufficient erythroid population not subjected to the p53 apoptotic stimulus, but able to complete the maturation process if stimulated by the treatment. Moreover, a nonhomogeneous distribution of the p53-positive elements was observed during the review of the specimens with clusters of positive cells more frequently located in the centrolacunar site and the context of erythroid colonies. To confirm the nature of the suspected erythroid colonies p53 positive, we confronted the area with these clusters with the corresponding area on hematoxylin-eosin and Giemsa slides in order to verify the presence of the typical morphological aspect of the erythroid precursors. This observation also seems in line with other studies reported in literature, in which it has been demonstrated that p53 protein accumulates more frequently in the immature erythroid precursors than in cells of other hematopoietic series  (Fig. 2). The data presented need further and wider confirmations, since our study, while providing new interesting and significant data, is limited by the retrospectivity of the survey and the number of samples. In conclusion, our study reinforces the concept of the BM biopsy as a fundamental investigation for the assessment of patients with MDS from the first diagnosis. In fact, in addition of being an important diagnostic aid, this analysis allows to obtain parameters more frequently correlated with response to ESAs (CD34 > 3% and p53 < 1%). As a future goal, it would be useful to evaluate the proposed results in integrated prognostic scores in order to optimize and customize the therapeutic strategies for each patient.
This study was supported by departmental funding and by Beat Leukemia Foundation (www.beat-leukemia.org). We would like to thank Roberta Tacchi for running the immunohistochemical analysis conducted on this series.
Statement of Ethics
All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the Declaration of Helsinki and its later amendments or comparable ethical standards. All subjects have given their written informed consent. All identifying information was removed from patient samples prior to their analysis to ensure strict privacy protection. For this type of study, the institute’s committee approval is not required.
Conflict of Interest Statement
All of the other authors have no potential conflicts of interest to disclose.
The authors have no funding sources to disclose.
F.B. and U.G. conception and design of the work and revising the work critically for important intellectual content. F.B., U.G., A.D.G., M.B., G.C., and R.C. interpretation of data, conducting of experiments, and drafting the work. G.R., L.R., L.B., E.B., and M.R. data acquisition and analysis. All authors approved the submission of this work.