Severe thrombocytopenia is a challenging situation in the treatment of myelodysplastic syndromes (MDS), with profound implications for prognosis and treatment decisions. The limit of <30 000/μL for platelet count has been proposed as an independent adverse prognostic factor, is included in the Global MD Anderson Scoring System and it may be influential on selecting therapy.1-3 This cut-point was also helpful in identifying a subgroup with very poor prognosis among International Prognostic Scoring System (IPSS) lower-risk patients2 and at the time of hypomethylating agent (HMA) failure.4 The HMAs are still the first-line option for high-risk patients and low-risk patients with transfusion dependence, or with a worsening clinical status. Their use has significantly altered the prospects of patients with MDS inducing 40%-60% of overall responses. Responses are gradual in onset, with the majority occurring by 6 months, and to continue on treatment beyond the first response improves their quality. However, these agents are non-curative and the median duration of response ranges between 7-14 months.4, 5 Our work aimed to examine the influence of the limit of ≤30 000/μL for platelet count on clinical outcomes, specifically, its relationship with treatment response to HMA in a Latin-American real-world setting. We identified 212 eligible de novo MDS patients from a multicentric retrospective database of 340 adult patients who were treated with HMA between January 2007 and January 2018. Informed consent was obtained according to institutional guidelines and following the Declaration of Helsinki. The cohort included patients from Argentina (118), Brazil (39), Venezuela (five), Colombia (27), Dominican Republic (10), Ecuador (nine), Uruguay (two) and Mexico (two). Response to treatment was evaluated according to the International Working Group criteria in 176 patients at optimum treatment (4-6 cycles for DAC-AZA) including those who failed. The median age was 70.8 years old (interquartile range [IQR] 60-77 years), 74.5% older than 60 years, and 54.2% were males. The WHO diagnosis included 39 (18.4%) cases of refractory anemia with excess of blasts (RAEB)-1 and 71 (33.5%) RAEB-2. Karyotype was abnormal in 70 (37.4%) and 28 (15.0%) with poor or very poor findings, excluding 25 not evaluable. Among the entire cohort, 120 (56.6%) scored for high risk (IPSS-R > 3.5). At treatment initiation, Hb level < 8 g/dL, absolute neutrophil count (ANC) <800/μL, and platelets <50 000/μL were present in 40.7% (median: 7.0 g/dL), 36.4% (median: 450/μL) and 42.9% (median: 20 000/μL) of patients, respectively (Table S1). Treatment with AZA (92.9%) or DAC (7.1%) started with a median of 4.0 months (IQR 1.3-10.1 months) following diagnosis (2.1 months for high risk and 8.4 months for low risk). The median number of cycles was six (IQR 4-11 cycles) during a median period of 7.2 months. Within a median follow-up from diagnosis of 18.1 months (IQR 11.2-36.3 months), 64 (30.2%) evolved to AML and 112 (52.8%) died. The median overall survival (OS) was 29.1 months (95% CI 20.7-37.5 months), from treatment initiation of 21.1 months (95% CI 14.6-27.5 months) and since last received cycle of 4.0 months (95% CI 15.6-33.0 months). According to baseline characteristics, 118 (55.7%) patients presented with platelet counts <100 000/μL, 42 (35.6%) of whom with thrombocytopenia ≤30 000/μL, and at treatment initiation those rates increased to 61.8% and 54.2%, respectively. Severe thrombocytopenia was more frequent in younger patients (64.6 vs 69.1 years old, P = .017) and with chromosome 7 abnormality or complex karyotypes (22.7% vs 10.7%, P = .033) (Table S1). Baseline thrombocytopenia ≤30 000/μL was associated with a shorter time to initiate therapy, within a median period of 2.3 months (IQR 1.7-8.1 months). Those 35 patients who presented a high count at diagnosis and lowered their counts at treatment beginning initiated the first cycle within 4.2 months (IQR 2.4-14.0 months, P = .035). Also, this parameter had a negative influence on receiving the minimum recommended cycles (53.5% vs 72.3%, P = .009). Parameters that influence worse OS in our cohort were male sex (P = .001), poor karyotypes (P < .001), hemoglobin level < 10 g/dL (P < .001), severe thrombocytopenia (P < .001), presence of peripheral blasts (P = .013), red blood cell transfusion requirement (P = .020) and bone marrow blast ≥5% (P = .024). Most of these parameters were also significant censoring at other treatments (Table S2). Particularly, patients with severe thrombocytopenia presented a reduced median OS of 10.3 months (95% CI 8.8-11.7 months) vs 30.0 months (95% CI 23.3-36.8 months, P < .001) for those with higher counts (Figure 1A). This short survival was similar whether they scored for a low (9.4 months, 95% CI 7.4-11.3 months) or a high-risk group (11.2 months, 95% CI 8.4-14.0 months) (P = .282). As expected, low-risk patients with a platelet count ˃30 000/μL showed a longer OS than those with high risk: 38.9 months (95% CI 21.1-56.6 months) vs 25.6 months (95% CI 20.4-30.8 months), respectively (P < .001) (Figure 1B). These findings were analogous when censoring at other treatments (Figures S1A,B). Based on these observations, severe thrombocytopenia was contrasted with the IPSS-R, sustaining its independency (HR 3.2, 95% CI 2.1-4.7, P < .001, and HR 1.1, 95% CI 0.8-1.7, P = .525, respectively). All parameters that showed a P-value <.2 in the univariate analyses (Table 2S) were included in a Cox regression analysis. Independent predictive factors for a worse outcome along with severe thrombocytopenia (HR 3.4, 95% CI 2.2-5.1, P < .001) were male sex (HR 1.9, 95% CI 1.3-2.9, P = .002), hemoglobin level < 10 g/dL (HR 2.8, 95% CI 1.6-5.1, P = .001), age ˃60 years old (HR 1.6, 95% CI 1.0-2.6, P = .047) and presence of chromosome 7 abnormality/ complex ≥3 abnormalities (HR 1.7, 95% CI 1.0-2.9, P = .034) (Figure S2 and Table S3). A total of 174 patients were evaluated for response to treatment at 4-6 cycles showing an overall response rate of 50.6% (complete response (CR)/ medular CR (mCR)/ partial response (PR): 29.9%, hematological improvement (HI): 20.7%) during a median follow-up period of 12.2 months. The outcome of patients with stable disease (SD) (16.1%) at 4-6 months was similar to those with CR/mCR/HI (P = .765), therefore, grouped as responders. The median OS for responders was 30.0 vs 7.7 months for those who failed, progressed or died (P < .001). Among those parameters that were predictive for survival, a failure to achieve response was associated with hemoglobin level < 10 g/dL (OR 7.9, 95% CI 2.3-27.4, P˂.001) and platelet count ≤30 000/μL (OR 3.0, 95% CI 1.5-6.1, P = .002). Moreover, thrombocytopenia ≤30 000/μL adversely impacts on the OS of responders, censoring (24.4 months vs NR, HR 3.8, 95% CI 1.9-7.6, P < .001, Figure S1C) or not at other treatments (15.1 vs 39.0 months, HR 3.3, 95% CI 1.8-5.8, P < .001, Figure 1C). Severe thrombocytopenia has been recognized as a strong independent risk factor for prognosis, with direct implications for therapeutic interventions. Its frequency varies among 7% to 26%, and from 3% to 17% in lower-risk patients.1, 2 Our results are consistent with a greater incidence in those who require therapeutic interventions, in the evolution of the disease and the high-risk group. The rate of one fifth at baseline was increased to one third before the onset of HMA and more frequent in higher-risk patients with a borderline statistical significance. And, as previously reported, it was concomitant with higher-risk karyotypes1, 2 and a lower age at diagnosis. The cut-point of 30 000/μL has also been independently associated with a shorter OS of 7 months, between 12-16 months or 4 months depending on their IPSS risk score in untreated series.1, 2 Our data also confirmed its independency on predicting a short OS of 10 months, however, without differences in risk distribution. It may suggest some benefits of HMA on high-risk patients. Nevertheless, the reduced survival observed in the lower-risk group advises considering this finding as a warning at the time of selecting therapy. Consistently, allogeneic HSCT for candidates or the use of new approaches have been recommended.2 The OS of de novo MDS patients from treatment initiation up to last follow up was 21 months: 16 months for high risk and 26 months for low risk, are comparable with previously published data.4-6 Also, achieving an SD at 4–6 months was beneficial to our patients with a median OS of 30 months, as preceding reports.3 The current debate is how to select the candidates who will achieve a response and its duration, and optional treatments for no responders. This last group of patients has a poor prognosis with a median OS of 7.9 months. In both HMA, the raise of platelet count after treatment initiation has been associated with better outcomes. The rate of increase, however, depends on platelet counts at baseline with a survival impact, and levels <100 000/μL were adverse for a clinical response. Our results complement these observations pointing that severe thrombocytopenia was associated with not reaching an optimal treatment and with three times lower chances to achieve a response. Even in responders, a platelet count ≤30 000/μL depicted a 2.5-fold shorter overall survival. This collaborative effort by the GLAM for collecting the large data from eight Latin-American countries to overview the experience on HMA during the last decade has some limitations. Despite the retrospective nature of the study and the heterogeneity of the cohort, the present series represents our real-world patients, and this may be a trigger to develop more regional studies.6 In summary, a platelet count ≤30 000/μL is an independent adverse factor in MDS under HMA influencing a lower response rate and short survival, even in responders. The integration of a lower cut-point of thrombocytopenia into risk scores highlights its importance, and severe thrombocytopenia should be recognized as a warning at the time of selecting therapy. The authors thank the investigators of the Argentinean MDS´s Study Group organized by the Argentinean Society of Hematology for the use of the MDS Registry database and to GLAM members. This work was not specifically supported, but C.B.B. receives grants from the Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) [PID 0044 and PICT 0480] and from the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) [PIP 0056]. The authors declare no conflict of interest. Data available on request from the authors. Appendix S1: 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.