1. Introduction
Acute promyelocytic leukemia (APL) is a distinct subtype of acute myeloid leukemia (AML) characterized by the PML::RARA fusion gene resulting from the t(15;17)(q22;q21) translocation, which induces maturation arrest at the promyelocytic stage [ 1 ]. Historically considered the most fatal AML subtype due to severe coagulopathy, the introduction of all-trans retinoic acid (ATRA) and arsenic trioxide (ATO) has transformed APL into a highly curable disease with long-term cure rates exceeding 90% in the current era [ 2 , 3 ]. Despite these remarkable therapeutic advances, early death (ED) remains the primary barrier to long-term survival persisting as the major cause of treatment failure [ 4 ].
ED is generally defined as death from any cause within the first 30 days of diagnosis. While clinical trials report ED rates of 2–6%, real-world data reveal significantly higher rates reaching up to 37% [ 5 , 6 , 7 , 8 ]. A substantial proportion of these fatalities occur within the first week of diagnosis, termed “very early death” (VED); however, data regarding the incidence and specific risk factors for VED remain limited [ 9 , 10 , 11 ]. While intracranial hemorrhage, disseminated intravascular coagulation (DIC), infections and differentiation syndrome (DS) are identified as the leading causes of ED, standard risk stratification models are primarily designed to predict relapse rather than early mortality [ 6 , 12 , 13 ].
Recently, novel risk scores developed by Österroos et al. and Cai et al. have demonstrated superior performance over the Sanz score in predicting ED and have been validated in independent cohorts [ 6 , 14 ]. While these models incorporate specific baseline variables, numerous studies have broadly identified advanced age, high white blood cell (WBC) count, thrombocytopenia, poor Eastern Cooperative Oncology Group (ECOG) performance status, and elevated lactate dehydrogenase (LDH) and creatinine levels as key predictors of early mortality [ 1 , 9 , 15 ]. However, these risk stratification systems rely exclusively on static baseline parameters obtained at admission. Given the rapid physiological changes induced by induction therapy, static markers may fail to capture the dynamic evolution of endothelial injury, coagulopathy, DS and infectious complications during the critical first month of treatment.
The endothelial activation and stress index (EASIX), calculated using creatinine, LDH and platelet counts, serves as a validated surrogate marker for endothelial activation and stress in both benign and malignant conditions. High baseline EASIX scores have been correlated with poor survival in multiple myeloma and Hodgkin and non-Hodgkin lymphoma [ 16 , 17 , 18 , 19 , 20 , 21 , 22 ]. Particularly in allogeneic hematopoietic stem cell transplantation (allo-HSCT), where endothelial damage is prominent, the dynamic monitoring of EASIX scores has been associated with inferior survival and increased microvascular complications [ 23 ]. This prognostic utility also extends to immune-mediated thrombotic thrombocytopenic purpura (iTTP), a disorder characterized by intravascular microthrombosis. In these patients, elevated EASIX scores at diagnosis and during follow-up are strongly linked to poorer treatment responses and increased relapse rates [ 24 ].
Despite the evolution of treatment modalities in APL, ED remains a critical clinical challenge. While scoring systems exist to determine ED risk at diagnosis, there is currently no dynamic scoring system capable of identifying which patients experience increased mortality risk and which patients approach a safer range during treatment. In this study, recognizing endothelial dysfunction as a key driver of APL-associated mortality, we aimed to assess the prognostic value of baseline EASIX scores and their dynamic changes during early induction for ED and long-term survival.
2. Materials and Methods
2.1. Study Design and Patients
This multicenter, retrospective cohort study analyzed data from adult patients consecutively diagnosed with APL at two tertiary referral centers in Turkey (Bursa Uludag University and Tekirdag Namik Kemal University) between January 2008 and May 2025. The study population comprised adult patients (aged ≥ 18 years) with de novo APL, diagnosed according to the World Health Organization (WHO) classification criteria [ 25 ]. The diagnosis was genetically verified by detecting the PML::RARA fusion transcript via a reverse transcriptase polymerase chain reaction (RT-PCR) or by the t(15;17) translocation using fluorescence in situ hybridization (FISH). Exclusion criteria encompassed cases with unconfirmed PML::RARA or t(15;17) status, secondary APL, concomitant hematological disorders, other active malignancies necessitating treatment and pregnancy. Figure 1 illustrates the patient selection flowchart. For the final cohort of 131 patients, comprehensive data regarding demographic information, clinicopathological characteristics, laboratory parameters, molecular profiles, risk stratification scores, therapeutic responses and survival outcomes were retrospectively extracted from medical records.
2.2. Treatment Regimens and Supportive Care
Due to national reimbursement policies restricting the use of ATO in the frontline setting, all patients received the standard AIDA protocol. Oral ATRA, 45 mg/m 2 /day, was initiated immediately upon clinical suspicion and continued until complete remission. Concurrent chemotherapy consisted of idarubicin (12 mg/m 2 ) administered intravenously on days 2, 4, 6 and 8. Aggressive supportive care was instituted to manage coagulopathy. Transfusion thresholds were set to maintain platelet counts ≥ 30 × 10 9 /L, fibrinogen levels ≥ 150 mg/dL (via fresh frozen plasma or fibrinogen concentrate), and hemoglobin levels ≥ 8 g/dL. For prophylaxis against DS, intravenous dexamethasone (10 mg every 12 h) was administered for the first 5–10 days in high-risk patients (WBC ≥ 10 × 10 9 /L) lacking active infection. Upon the development of DS signs or symptoms, ATRA was temporarily withheld, and high-dose dexamethasone was continued until symptom resolution. Patients achieving complete remission proceeded to three cycles of anthracycline–ATRA-based consolidation. Maintenance therapy was continued for two years, comprising intermittent ATRA (45 mg/m 2 /day for 15 days every 3 months), daily 6-mercaptopurine (50 mg/m 2 ) and weekly methotrexate (15 mg/m 2 ) [ 26 , 27 , 28 ].
2.3. Definitions and Risk Stratification
Baseline risk stratification utilized the Sanz [ 29 ] and International Society of Thrombosis and Haemostasis (ISTH) scores [ 30 ], while the Österroos (incorporating age, WBC, platelets) [ 6 ] and Cai models (integrating age, WBC, platelets, LDH) [ 14 ] were applied specifically for ED risk prediction. ED was defined as mortality from any cause within the first 30 days of diagnosis, whereas VED referred to mortality occurring in less than 7 days (<7 days) of diagnosis. DS was diagnosed according to the criteria established by Frankel et al. [ 31 ], based on the presence of characteristic signs, including unexplained fever, weight gain (>5 kg), respiratory distress, pulmonary infiltrates, pleural/pericardial effusion, hypotension, and acute renal failure.
The EASIX score was computed using the standard formula: [LDH (U/L) × Creatinine (mg/dL)]/Platelet Count (10 9 /L) [ 32 ]. To address the skewed distribution of the raw EASIX score, a binary logarithm (Log 2 ) transformation was applied [ 33 ]. We evaluated the baseline score (EASIX D0), the score on day 7 (EASIX D7) and the dynamic change (ΔEASIX), calculated as the day 7 score minus the baseline (D0) score. Day 7 was specifically selected as the temporal landmark for dynamic assessment for two primary reasons. Clinically, VED in APL is standardly defined as mortality within the first 7 days; utilizing day 7 allows for a robust landmark analysis to evaluate subsequent early mortality. Therapeutically, because the AIDA protocol involves anthracycline administration on days 2, 4, 6 and 8, day 7 effectively captures the peak cumulative endothelial damage near the completion of the active cytotoxic phase.
Overall survival (OS) was measured from diagnosis to death from any cause or last follow-up. Event-free survival (EFS) was calculated from diagnosis to the first event (relapse, death or last follow-up).
2.4. Statistical Analysis
Statistical analyses were performed using IBM SPSS Statistics version 29.0 (IBM Corp., Armonk, NY, USA) and R software (version 4.5.2; R Foundation for Statistical Computing, Vienna, Austria) within the RStudio (version 2026.01.0+392) environment. Data normality was assessed via the Shapiro–Wilk test. Continuous variables were presented as the mean ± standard deviation or median (min-max), while categorical variables were summarized as frequencies and percentages. Group comparisons utilized the Mann–Whitney U, one-way ANOVA or Kruskal–Wallis tests for continuous variables and the Chi-square or Fisher–Freeman–Halton tests for categorical data, depending on distribution and cell counts.
Prognostic cut-offs for EASIX scores (D0 for VED; D7 and ΔEASIX for mortality between days 7 and 30) were determined using receiver operating characteristic (ROC) curve analysis and the Youden index. To mitigate immortal time bias, a 7-day landmark analysis was applied for dynamic markers (D7/ΔEASIX), excluding VED (<7 days).
In this landmark cohort, survival probabilities (OS/EFS) were estimated via the landmark Kaplan–Meier method and compared using the Log-rank test. Variables yielding a p -value < 0.20 in univariate analysis were entered into the multivariate binary logistic regression (for 7–30 day mortality) and Cox proportional hazards regression (for OS/EFS) models. These multivariate models were constructed using the Backward stepwise (Likelihood Ratio) method to identify independent prognostic factors. Advanced visualizations (Sankey diagrams, violin plots) and model performance metrics (Net Reclassification Improvement [NRI]) were generated using specific R (version 4.5.2) packages (survival, survminer, ggalluvial, ggplot2, nricens). All tests were two-sided with a significance level of p < 0.05.
3. Results
3.1. Baseline Characteristics and Patient Profile
The baseline demographic and clinicopathological characteristics of the study cohort are summarized in Table 1 . The median age of the cohort was 46 years, with a female predominance (63.4%). The majority of patients (80.2%) had a good performance status (ECOG 0–1). The classical variant was the dominant subtype (91.6%). The frequency of Fms-like tyrosine kinase 3-internal tandem duplication mutation ( FLT3-ITD ) and FLT3-tyrosine kinase domain (TKD) mutations was 15.3% and 5.3%, respectively. Risk stratification analysis indicated that 22.9% of the cohort was classified as high-risk according to the Sanz score. When the Österroos model was applied, 37.4% were categorized as high-risk and 10.7% as ultra-high-risk.
3.2. Distinct Clinicobiological Profiles of Early Mortality
Early mortality occurred in 33 patients (25.2%) within the first 30 days of induction. An analysis of the temporal distribution of mortality distinguished two specific subgroups: 8 patients (6.1%) experienced VED (<7 days), while the remaining 25 deaths (19.1%) occurred during the subsequent early period (7–30 days). The primary causes of VED were intracranial hemorrhage (n = 4), DIC-related multi-organ failure (n = 3) and sepsis (n = 1). During the subsequent early period, mortalities were attributed to sepsis (n = 10), DIC-related multi-organ failure (n = 5), fatal bleeding (n = 4), differentiation syndrome (n = 3), and thrombosis (n = 3). In contrast, late mortalities were exclusively driven by primary disease complications secondary to clinical relapse.
A detailed comparative analysis of clinical characteristics across these groups is presented in Table 2 . Regarding demographic factors, patients experiencing mortality between days 7 and 30 were significantly older (median 59 years) compared to both the VED group (38.5 years) and survivors (41.5 years) ( p < 0.001). Despite this age disparity, both early mortality cohorts (VED and 7–30 days) shared a phenotype of aggressive disease burden and poor functional status compared to long-term survivors. Patients in both mortality groups exhibited significantly poorer performance status (ECOG ≥ 2) and markedly elevated median levels of leukocytes, LDH and serum creatinine ( p < 0.05). Similarly, the prevalence of FLT3 mutations was significantly higher in both early death groups compared to survivors ( p = 0.003). Compared to survivors, established risk stratification models consistently assigned a significantly higher proportion of patients in the early death groups to their respective highest-risk tiers (specifically, the high-risk category for the Sanz model and the high- and ultra-high-risk categories for the Österroos and Cai models). Regarding treatment-related complications, although the incidence of DS was numerically higher in the VED (25%) and 7–30-day mortality groups (16%) compared to survivors (10.2%), this difference did not reach statistical significance ( p = 0.248).
A distinct temporal pattern was observed in endothelial stress markers. Baseline EASIX (D0) scores were significantly elevated in the VED group compared to the other cohorts ( p = 0.015), with no significant difference observed between the 7–30-day mortality group and survivors at admission. Reflecting this, as illustrated in Figure 2 , ROC analysis identified an EASIX D0 cut-off of 3.58 (AUC: 0.734; sensitivity: 87.5%; specificity: 61.0%; p = 0.007) as a specific predictor for fatal events occurring in the very early period (<7 days). In contrast, dynamic endothelial assessment was specifically analyzed for its ability to predict mortality occurring between days 7 and 30. In this landmark window, EASIX D7 and ΔEASIX scores significantly discriminated the mortality group from survivors ( p < 0.001 and p = 0.020, respectively). Consequently, as demonstrated in Figure 2 , an EASIX D7 cut-off of 2.46 (AUC: 0.754; sensitivity: 80.0%; specificity: 50.0%; p < 0.001) was identified as a significant predictor.
Furthermore, a worsening endothelial status—defined as an increase in the EASIX score (ΔEASIX)—was strongly predictive of mortality in the subsequent early period (7–30 days). As shown in Figure 3 , non-survivors exhibited progressive deterioration, whereas survivors typically displayed a stable or improving course. Statistically, the optimal cut-off for this increase was determined as >0.35 (AUC: 0.651; sensitivity: 56.0%; specificity: 79.6%; p = 0.034); patients exceeding this threshold were significantly more likely to experience early mortality, while those maintaining values below this limit were predominantly survivors.
3.3. Dynamic Risk Stratification and Predictors of Early Mortality
The prognostic impact of dynamic risk changes is visualized in a multi-stage Sankey diagram, as shown in Figure 4 . Initial assessment revealed a discordance between static risk scores and biological endothelial stress. Among patients stratified as low-risk by the Österroos model, 22% already had high baseline EASIX scores (EASIX D0 > 3.58), whereas 36.5% of the high/ultra-high-risk strata had low initial EASIX scores. Crucially, the baseline Österroos model successfully identified the patients at the highest immediate risk: all eight VEDs occurred exclusively in the high- or ultra-high-risk groups. Longitudinal tracking exposed critical risk shifts that determined patient fate. While 48.5% (33/68) of the Österroos low-risk cohort deteriorated to a high EASIX status by day 7, notable plasticity was observed in the Österroos high-risk group. Specifically, 34.7% (17/49) of these patients successfully downstaged to a low-EASIX status. This biological recovery, termed the “Rescue Effect,” was associated with a marked reduction in mortality from 56.8% to 27.8%. Consequently, the integration of day 7 dynamic status provided a statistically significant improvement in risk reclassification for non-survivors (Net Reclassification Improvement, NRI+ = 0.500; p = 0.015).
As detailed in Table 3 , multivariate logistic regression analysis was performed on the landmark cohort to identify independent predictors of early mortality. In this model, worsening endothelial function (ΔEASIX > 0.35) emerged as the strongest independent predictor, increasing the risk by 12.4-fold (OR: 12.41; 95% CI: 1.96–78.46; p = 0.007). The baseline Österroos risk score also retained independent significance (OR: 8.58; p = 0.002), confirming the complementary value of static and dynamic assessments.
3.4. Long-Term Survival Outcomes
The long-term prognostic impact of dynamic endothelial assessment was evaluated using a landmark analysis to mitigate immortal time bias. Kaplan–Meier survival curves, stratified by the kinetic trajectory of ∆EASIX (cut-off: >0.35), are illustrated in Figure 5 and Figure 6 . Independent prognostic factors for these outcomes were identified via multivariate Cox proportional hazards regression models, as detailed in Table 4 and Table 5 . With a median follow-up of 35.7 months, landmark analysis confirmed that a worsening endothelial status (high ∆EASIX) was associated with significantly inferior long-term outcomes. The low-∆EASIX cohort maintained robust EFS (1-year: 83.6%; 3-year: 77.8%), whereas the high-∆EASIX group exhibited markedly lower rates (1-year: 59.5%; 3-year: 56.3%; p = 0.041). A parallel disparity was evident in OS, where the low-∆EASIX group achieved superior 1- and 3-year rates (84.8% and 78.0%) compared to the high-∆EASIX cohort (59.5% and 59.5%; p = 0.026). Furthermore, consistent with these dynamic kinetic findings, elevated static day 7 EASIX scores were also predictive of inferior long-term survival, as detailed in Supplementary Figures S1 and S2 .
In the multivariate analysis, a worsening endothelial status (ΔEASIX > 0.35) was an independent predictor of both poor EFS (HR: 5.70; 95% CI: 1.72–18.82; p = 0.004) and poor OS (HR: 3.69; 95% CI: 1.19–11.40; p = 0.023). Similarly, the baseline Österroos risk score (>low) maintained its independent significance for both EFS (HR: 8.58; p = 0.002) and OS (HR: 7.29; p = 0.032). Notably, consistent with the early mortality analyses, traditional static risk factors such as FLT3 mutation status and Sanz score lost their independent prognostic value when adjusted for dynamic endothelial markers in the long-term models.
4. Discussion
Current risk stratification models for early death and relapse in APL rely exclusively on static baseline parameters [ 6 , 14 , 29 ], failing to capture the intense biological volatility of the induction phase. Given the rapid and profound physiological shifts characteristic of this distinct AML subtype, such static assessments inevitably lead to significant risk misclassification. Consequently, despite remarkable therapeutic advances, early death remains a persistent clinical challenge [ 34 ], necessitating markers that reflect real-time physiological evolution. To the best of our knowledge, this is the first study to evaluate the dynamic evolution of EASIX score during induction therapy and its prognostic impact in patients with APL. Our findings demonstrate that APL is not a static disease; rather, it follows a rapidly shifting clinicobiological trajectory determined by the host’s endothelial response to the leukemic insult and induction therapy. We identified that while baseline scores are valuable for predicting immediate fatalities (<7 days), the “kinetic trajectory” of endothelial stress is the decisive factor for subsequent outcomes. Specifically, a worsening endothelial status—defined as an increase in ΔEASIX (>0.35)—emerged as a robust, independent predictor across the entire prognostic spectrum, significantly correlating with early mortality (between days 7 and 30), inferior EFS and poor OS. Conversely, we described a “Rescue Effect,” in which high-risk patients who achieved endothelial stabilization by day 7 demonstrated significantly improved survival, highlighting the critical importance of dynamic risk reclassification.
Despite the curative potential of current regimens, ED persists as the primary barrier to survival. While clinical trials report ED rates as low as 3–6%, real-world registries consistently document significantly higher rates ranging from 17% to 29% [ 34 , 35 ]. Our cohort confirmed this real-world disparity with an ED rate of 25.2% and VED rate of 6.1%. Beyond the relapse-oriented Sanz score, contemporary risk models developed by Österroos and Cai have specifically targeted early mortality by incorporating key variables such as age, leukocytosis, thrombocytopenia, and elevated LDH [ 6 , 14 ]. Regarding immediate mortality, Infante et al. reported a VED rate of 4.8% and identified serum creatinine as a significant independent predictor of death within the first 7 days [ 10 ]. Additionally, other independent studies have consistently highlighted poor ECOG performance status as a critical determinant of early death [ 9 ]. Crucially, in the modern era where late relapses are infrequent, long-term survival is mathematically determined by the ability to survive the induction phase [ 1 , 36 ]. Consequently, clinical markers that successfully predict early mortality—such as poor ECOG status and Österroos score—also emerged as significant predictors for EFS and OS in our analysis. Serving inherently as a composite marker of vascular stress, EASIX demonstrates a robust capacity to predict both early fatal events and long-term survival. In this context, the discordance between Österroos and baseline EASIX scores in our Sankey diagram further highlights their distinct pathophysiological targets. While Österroos reflects leukemic burden and demographic risk, EASIX specifically quantifies host endothelial vulnerability. Thus, a patient may be high-risk according to Österroos (e.g., hyperleukocytosis) but have a low EASIX if systemic endothelial dysfunction has not ensued or vice versa. Clinically, these scores are highly complementary, a relationship robustly corroborated by our multivariate models, where both independently retained prognostic significance. Integrating dynamic EASIX allows clinicians to identify hidden high-risk patients experiencing profound endothelial dysfunction despite a seemingly favorable traditional profile.
APL is defined by a “double-hit” mechanism involving concomitant primary hyperfibrinolysis and DIC, standing in sharp contrast to sepsis-associated coagulopathy, which is typically characterized by fibrinolytic shutdown [ 37 , 38 ]. This coagulopathy is driven not only by the overexpression of Annexin A2 and tissue-type plasminogen activator (t-PA), which generate excessive plasmin, but also by the release of tissue factor (TF) and cancer procoagulant (CP) from leukemic promyelocytes, leading to profound thrombin generation [ 38 , 39 , 40 , 41 , 42 ]. Furthermore, the release of inflammatory cytokines induces the expression of adhesion molecules on the vascular endothelium, promoting leukemic cell sequestration and microvascular injury [ 42 , 43 ]. Since this biological storm is fluid, static snapshots often fail to capture the escalating risk. Indeed, Infante et al. demonstrated that worsening ISTH DIC score over time was a stronger predictor of mortality than baseline values, validating the concept that the evolution of coagulopathy is the critical determinant of survival [ 11 ]. Importantly, while APL-associated early mortality is clinically heterogeneous in its direct causes ranging from catastrophic hemorrhage and severe differentiation syndrome to fulminant sepsis, the underlying pathophysiology consistently converges on profound systemic endothelial dysfunction and microvascular injury. EASIX has previously been validated as a predictor of endothelial-driven complications, including sinusoidal obstruction syndrome (VOD), graft-versus-host disease (GVHD), and chimeric antigen receptor T-cell (CAR-T) toxicity [ 20 , 32 , 33 ]. As a result, dynamic EASIX serves as a robust predictor of all-cause early mortality, capturing this shared terminal pathway of systemic vascular collapse rather than being limited exclusively to hemorrhagic events. Within this unique pathophysiological framework, the individual components of EASIX directly mirror the severity of APL-induced vascular collapse: elevated LDH reflects profound tissue ischemia and leukemic cell turnover; thrombocytopenia represents the exhaustive consumptive coagulopathy; and rising creatinine signifies acute microvascular hypoperfusion and capillary leak. In this manner, their combined kinetic trajectory precisely tracks the clinical deterioration leading to mortality.
Induction therapy further complicates this vascular landscape. While ATRA is essential for differentiation, it triggers the release of inflammatory cytokines and induces neutrophil extracellular traps (NETosis), which can aggravate endothelial injury and increase the risk of differentiation syndrome in the initial phase [ 42 , 44 ]. Conversely, recent evidence suggests that ATO may exert an endothelial-protective effect, downregulating adhesion molecules and reducing vascular permeability compared to chemotherapy-based regimens [ 45 , 46 ]. In our study, which used the AIDA protocol without upfront ATO, the dynamic elevation of EASIX likely captured the cumulative endothelial insult from leukemic hyperfibrinolysis, chemotherapy-induced toxicity and ATRA-mediated NETosis. This suggests that dynamic endothelial monitoring is particularly vital in settings where chemotherapy remains the backbone of induction. However, given that the ATRA plus ATO combination is now the standard of care for low-to-intermediate-risk APL, a critical question is whether the predictive value of ΔEASIX remains consistent in chemotherapy-free settings. While frontline ATO-based regimens have dramatically improved long-term survival and reduced relapse rates, they have not completely abrogated early mortality, which still occurs in approximately 7–10% of patients, even in recent clinical trials [ 35 , 36 ]. Because these early fatal events remain fundamentally driven by the inherent leukemic coagulopathy, severe differentiation syndrome and DIC-related multi-organ failure, we hypothesize that the dynamic prognostic utility of ΔEASIX will remain highly relevant in ATO-treated patients. Nevertheless, given the disparate endothelial effects of these agents, the optimal prognostic thresholds of the ΔEASIX score may differ from those established in chemotherapy-treated cohorts.
Our study has several limitations. First, the retrospective, two-center design and modest sample size precluded the evaluation of intermediate time points (e.g., days 3 and 14) for a more detailed kinetic assessment of the EASIX score. Second, while the day 7 landmark analysis was methodologically necessary to mitigate immortal time bias, it inherently excluded very early deaths, introducing an unavoidable selection bias. Third, since platelet count is a component of the EASIX formula, daily platelet transfusions can potentially affect the score. Although patients received standardized supportive care, our retrospective design prevented adjusting for exact transfusion volumes. Fourth, although the absolute number of early deaths is consistent with comparable real-world APL cohorts, this limited number of events naturally restricts the statistical power of our multivariate models. While stringent stepwise selection methods were utilized to minimize overfitting, the independent prognostic factors identified herein require prospective external validation. Finally, since our cohort was restricted to patients treated with the AIDA protocol, future studies are warranted to confirm the prognostic utility of dynamic EASIX monitoring in chemotherapy-free, ATO-based regimens.
5. Conclusions
In conclusion, our study is the first to demonstrate that dynamic endothelial stress evolution is a critical prognostic factor in APL. We identified that worsening endothelial status (ΔEASIX > 0.35) is a robust, independent predictor of early mortality and significantly stratifies patients for long-term survival. By capturing the physiological volatility of induction that static biomarkers miss, dynamic EASIX monitoring offers a practical tool for real-time risk refinement. Clinically, a high ΔEASIX trajectory at day 7 serves as a “red flag” requiring immediate preemptive action. For these high-risk patients, clinicians should prioritize stricter transfusion thresholds to counteract profound coagulopathy, maintain rigorous surveillance for severe infections like sepsis and optimize ongoing prophylactic measures for differentiation syndrome. Conversely, patients with stable endothelial function by day 7 may safely proceed with standard care. Furthermore, given the impact of ATRA-induced NETosis on endothelial damage, this high-risk cohort is an ideal target population for evaluating emerging therapies aimed at preventing NETosis. Ultimately, further large-scale, prospective, multicenter studies are required to validate these findings and establish the clinical utility of this dynamic model across different therapeutic protocols.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Supplementary Materials
The following supporting information can be down-loaded at https://www.mdpi.com/article/10.3390/cancers18050843/s1 , Figure S1: Landmark Kaplan-Meier analysis of EFS stratified by day 7 EASIX status; Figure S2: Landmark Kaplan-Meier analysis of OS stratified by day 7 EASIX status.
Author Contributions
F.Ç.H. designed this study, analyzed the results, wrote and revised the final manuscript and submitted the article. H.Ö., S.E.G.B., N.G.A.T., O.S., İ.K., E.A., Ş.Ç., T.G.K., A.M.Y., İ.E.P., V.G. and T.E. collected the data and analyzed the results. M.B., S.A., Y.B., V.Ö. and F.Ö. provided supervision, interpreted the results and revised the final manuscript. All authors have read and agreed to the published version of the manuscript.
Institutional Review Board Statement
Ethical approval was obtained from the Clinical Research Ethics Committee of the Bursa Uludag University Faculty of Medicine (Approval Date: 30 April 2025; Decision No: 2025/640-9/3). All study procedures adhered to the tenets of the World Medical Association Declaration of Helsinki.
Informed Consent Statement
Informed consent for this retrospective study was not required. Patient consent was waived due to the retrospective nature of the study and the use of anonymized clinical data.
Data Availability Statement
The raw data supporting the conclusions of this article will be made available by the authors on request.
Conflicts of Interest
The authors declare no conflicts of interest.
References
1. Chean D. Kemp H. Fodil S. Darmon M. Azoulay E. Dupont T. Early Mortality in Patients with Acute Promyelocytic Leukemia: A Systematic Review and Meta-Analysis Crit. Care 2025 29 490 10.1186/s13054-025-05743-7 41239429 PMC12619144
2. Platzbecker U. Avvisati G. Cicconi L. Thiede C. Paoloni F. Vignetti M. Ferrara F. Divona M. Albano F. Efficace F. Improved Outcomes with Retinoic Acid and Arsenic Trioxide Compared with Retinoic Acid and Chemotherapy in Non-High-Risk Acute Promyelocytic Leukemia: Final Results of the Randomized Italian-German APL0406 Trial J. Clin. Oncol. 2017 35 605 612 10.1200/JCO.2016.67.1982 27400939
3. Kantarjian H. Kadia T. DiNardo C. Daver N. Borthakur G. Jabbour E. Garcia-Manero G. Konopleva M. Ravandi F. Acute Myeloid Leukemia: Current Progress and Future Directions Blood Cancer J. 2021 11 41 10.1038/s41408-021-00425-3 33619261 PMC7900255
4. Stahl M. Tallman M.S. Acute Promyelocytic Leukemia (APL): Remaining Challenges towards a Cure for All Leuk. Lymphoma 2019 60 3107 3115 10.1080/10428194.2019.1613540 31842650 PMC7479633
5. Lo-Coco F. Avvisati G. Vignetti M. Thiede C. Orlando S.M. Iacobelli S. Ferrara F. Fazi P. Cicconi L. Di Bona E. Retinoic Acid and Arsenic Trioxide for Acute Promyelocytic Leukemia N. Engl. J. Med. 2013 369 111 121 10.1056/NEJMoa1300874 23841729
6. Österroos A. Maia T. Eriksson A. Jädersten M. Lazarevic V. Wennström L. Antunovic P. Cammenga J. Deneberg S. Lorenz F. A Risk Score Based on Real-World Data to Predict Early Death in Acute Promyelocytic Leukemia Haematologica 2022 107 1528 1537 10.3324/haematol.2021.280093 35081688 PMC9244824
7. Baysal M. Gürsoy V. Hunutlu F.C. Erkan B. Demirci U. Bas V. Gulsaran S.K. Pinar I.E. Ersal T. Kirkizlar T.A. The Evaluation of Risk Factors Leading to Early Deaths in Patients with Acute Promyelocytic Leukemia: A Retrospective Study Ann. Hematol. 2022 101 1049 1057 10.1007/s00277-022-04798-8 35190843
8. Jillella A.P. Lee S.J. Altman J.K. Luger S.M. Tallman M.S. Foran J.M. Bradshaw D. Law L.Y. Bryan L.J. Abou Zahr A. Academic Community Partnership in Acute Promyelocytic Leukemia and Early Mortality: The ECOG-ACRIN EA9131 Trial JAMA Oncol. 2025 11 400 407 10.1001/jamaoncol.2024.7033 40014329 PMC11869096
9. Lehmann S. Ravn A. Carlsson L. Antunovic P. Deneberg S. Möllgrd L. Derolf Å.R. Stockelberg D. Tidefelt U. Wahlin A. Continuing High Early Death Rate in Acute Promyelocytic Leukemia: A Population-Based Report from the Swedish Adult Acute Leukemia Registry Leukemia 2011 25 1128 1134 10.1038/leu.2011.78 21502956
10. Infante J. Esteves G. Raposo J. de Lacerda J.F. Predictors of Very Early Death in Acute Promyelocytic Leukemia: A Retrospective Real-World Cohort Study Ann. Hematol. 2023 102 3031 3037 10.1007/s00277-023-05422-z 37650885 PMC10567916
11. Infante J.B. Esteves G.V. Raposo J. de Lacerda J.F. Disseminated Intravascular Coagulation Score Evolution in 48 h Predicts Early Death in Acute Promyelocytic Leukemia Patients Eur. J. Haematol. 2024 112 840 844 10.1111/ejh.14177 38305491
12. De La Serna J. Montesinos P. Vellenga E. Rayón C. Parody R. León A. Esteve J. Bergua J.M. Milone G. Debén G. Causes and Prognostic Factors of Remission Induction Failure in Patients with Acute Promyelocytic Leukemia Treated with All-Trans Retinoic Acid and Idarubicin Blood 2008 111 3395 3402 10.1182/blood-2007-07-100669 18195095
13. Lehmann S. Deneberg S. Antunovic P. Rangert-Derolf Å. Garelius H. Lazarevic V. Myhr-Eriksson K. Möllgård L. Uggla B. Wahlin A. Early Death Rates Remain High in High-Risk APL: Update from the Swedish Acute Leukemia Registry 1997–2013 Leukemia 2017 31 1457 1459 10.1038/leu.2017.71 28232742
14. Cai P. Wu Q. Wang Y. Yang X. Zhang X. Chen S. An Effective Early Death Scoring System for Predicting Early Death Risk in de Novo Acute Promyelocytic Leukemia Leuk. Lymphoma 2020 61 1989 1995 10.1080/10428194.2020.1742910 32228267
15. Di Bona E. Avvisati G. Castaman G. Vegna M.L. De Sanctis V. Rodeghiero F. Mandelli F. Early Haemorrhagic Morbidity and Mortality during Remission Induction with or without All-Trans Retinoic Acid in Acute Promyelocytic Leukaemia Br. J. Haematol. 2000 108 689 695 10.1046/j.1365-2141.2000.01936.x 10792270
16. Yu J.W. Yang C.J. Yang J.F. Fang J.J. Relationship of Endothelial Activation and Stress Index on Concurrent Acute Kidney Injury in Patients with Heart Failure during Intensive Care Unit: A Competing Risk Model Analysis BMC Cardiovasc. Disord. 2025 26 22 10.1186/s12872-025-05403-2 41351032 PMC12797834
17. Kamaci Sener D. Haki C. Koc Yamanyar G. Kandemir F.N. Kamisli S. Sarac K. Association of Endothelial Activation and Stress Index with Prognosis in Posterior Circulation Infarcts Treated with Recanalization Therapy Diagnostics 2025 15 3234 10.3390/diagnostics15243234 41464233 PMC12732078
18. Thanhakun R. Wudhikarn K. Bunworasate U. Rattanathammethee T. Norasetthada L. Kanya P. Chaloemwong J. Wongkhantee S. Phiphitaporn P. Chansung K. Endothelial Activation and Stress Index as a Prognostic Factor of Diffuse Large B-Cell Lymphoma: The Report from the Nationwide Multi-Center Thai Lymphoma Study Group Ann. Hematol. 2023 102 3533 3541 10.1007/s00277-023-05437-6 37718327
19. Li M. Chen H. Gao L. Li F. Zhang Q. Zhou X. Yang Z. Gao W. Zhao H. Ma Y. Endothelial Activation and Stress Index (EASIX) as a Predictive Biomarker for Survival in Patients with T-Cell Lymphoma Hematology 2025 30 2519974 10.1080/16078454.2025.2519974 40551427
20. Li Y. Liu J. Luo L. Zhou L. Wu Z. Sang W. Jiang H. Wang J. Wang X. Li P. Bone Marrow Blasts- and Modified EASIX-Guided Risk Stratification for Coagulopathy and Outcomes after CAR-T Therapy in Relapsed/Refractory B-ALL Thromb. Res. 2025 256 109527 10.1016/j.thromres.2025.109527 41218266
21. Gursoy V. Gul O.O. Kose E. Sadri S. Clinical Practice of the Endothelial Activation and Stress Index and Its Modified Versions in Multiple Myeloma and the Effects on Prognosis Ther. Adv. Hematol. 2026 17 20406207251387409 10.1177/20406207251387409 41509514 PMC12775338
22. Hunutlu F.Ç. Öztop H. Gürsoy V. Ersal T. Elgün E. Yavuz Ş. İldemir Ekizoğlu S. Ekizoğlu A.A. Özkocaman V. Özkalemkaş F. Modified Endothelial Activation and Stress Index: A New Predictor for Survival Outcomes in Classical Hodgkin Lymphoma Treated with Doxorubicin-Bleomycin-Vinblastine-Dacarbazine-Based Therapy Diagnostics 2025 15 185 10.3390/diagnostics15020185 39857069 PMC11764177
23. Nawas M.T. Sanchez-Escamilla M. Devlin S.M. Maloy M.A. Ruiz J.D. Sauter C.S. Giralt S.A. Perales M.A. Scordo M. Dynamic EASIX Scores Closely Predict Nonrelapse Mortality after Allogeneic Hematopoietic Cell Transplantation Blood Adv. 2022 6 5898 5907 10.1182/bloodadvances.2022007381 35977079 PMC9661383
24. Escribano-Serrat S. Salas M.Q. Pascual-Izquierdo C. Fernández Villalobos M. Gómez-Seguí I. Fidalgo T. Fernández Docampo M. Martínez Nieto J. Lamas Ó. Lozano M. EASIX, a New Tool to Predict Response and Refractoriness in Immune-Mediated Thrombotic Thrombocytopenic Purpura Front. Immunol. 2025 16 1700907 10.3389/fimmu.2025.1700907 41333464 PMC12665919
25. Sanz M.A. Fenaux P. Tallman M.S. Estey E.H. Löwenberg B. Naoe T. Lengfelder E. Döhner H. Burnett A.K. Chen S.J. Management of Acute Promyelocytic Leukemia: Updated Recommendations from an Expert Panel of the European LeukemiaNet Blood 2019 133 1630 1643 10.1182/blood-2019-01-894980 30803991 PMC6509567
26. Fenaux P. Chastang C. Chevret S. Sanz M. Dombret H. Archimbaud E. Fey M. Rayon C. Huguet F. Sotto J.J. A randomized comparison of all transretinoic acid (ATRA) followed by chemotherapy and ATRA plus chemotherapy and the role of maintenance therapy in newly diagnosed acute promyelocytic leukemia Blood 1999 94 1192 1200 10.1182/blood.V94.4.1192 10438706
27. Sanz M.A. Martín G. Rayón C. Esteve J. González M. Díaz-Mediavilla J. Bolufer P. Barragán E. Terol M.J. González J.D. A modified AIDA protocol with anthracycline-based consolidation results in high antileukemic efficacy and reduced toxicity in newly diagnosed PML/RARalpha-positive acute promyelocytic leukemia Blood 1999 94 3015 3021 10.1182/blood.V94.9.3015 10556184
28. Lo-Coco F. Avvisati G. Vignetti M. Breccia M. Gallo E. Rambaldi A. Paoloni F. Fioritoni G. Ferrara F. Specchia G. Front-Line Treatment of Acute Promyelocytic Leukemia with AIDA Induction Followed by Risk-Adapted Consolidation for Adults Younger than 61 Years: Results of the AIDA-2000 Trial of the GIMEMA Group Blood 2010 116 3171 3179 10.1182/blood-2010-03-276196 20644121
29. Sanz M.A. Lo Coco F. Martín G. Avvisati G. Rayón C. Barbui T. Díaz-Mediavilla J. Fioritoni G. González J.D. Liso V. Definition of relapse risk and role of nonanthracycline drugs for consolidation in patients with acute promyelocytic leukemia: A joint study of the PETHEMA and GIMEMA cooperative groups Blood 2000 96 1247 1253 10.1182/blood.V96.4.1247 10942364
30. Levi M. Toh C.H. Thachil J. Watson H.G. Guidelines for the Diagnosis and Management of Disseminated Intravascular Coagulation Br. J. Haematol. 2009 145 24 33 10.1111/j.1365-2141.2009.07600.x 19222477
31. Sentero D. Hosenpud J. Retinoic Acid Syndrome in Acute Promyelocytic Leukemia Wis. Med. J. 1997 96 35 38 10.7326/0003-4819-117-4-292 9128433
32. Luft T. Benner A. Jodele S. Dandoy C.E. Storb R. Gooley T. Sandmaier B.M. Becker N. Radujkovic A. Dreger P. EASIX in Patients with Acute Graft-versus-Host Disease: A Retrospective Cohort Analysis Lancet Haematol. 2017 4 e414 e423 10.1016/S2352-3026(17)30108-4 28733186
33. Luft T. Benner A. Terzer T. Jodele S. Dandoy C.E. Storb R. Kordelas L. Beelen D. Gooley T. Sandmaier B.M. EASIX and Mortality after Allogeneic Stem Cell Transplantation Bone Marrow Transplant. 2020 55 553 561 10.1038/s41409-019-0703-1 31558788 PMC8082940
34. Li G. Wu J. Li R. Pan Y. Ma W. Xu J. Nan M. Hou L. Improvement of Early Death in Acute Promyelocytic Leukemia: A Population-Based Analysis Clin. Lymphoma Myeloma Leuk. 2023 23 e78 e84 10.1016/j.clml.2022.11.005 36567214
35. Voso M.T. Guarnera L. Lehmann S. Döhner K. Döhner H. Platzbecker U. Russell N. Dillon R. Thomas I. Ossenkoppele G. Acute Promyelocytic Leukemia: Long-Term Outcomes from the HARMONY Project Blood 2025 145 234 243 10.1182/blood.2024026186 39504485
36. Platzbecker U. Adès L. Montesinos P. Ammatuna E. Fenaux P. Baldus C. Bernardi M. Berthon C. Bocchia M. Bonmati C. Arsenic Trioxide and All-Trans Retinoic Acid Combination Therapy for the Treatment of High-Risk Acute Promyelocytic Leukemia: Results from the APOLLO Trial J. Clin. Oncol. 2025 43 3160 3169 10.1200/JCO-25-00535 40825164
37. Mitrovic M. Suvajdzic N. Bogdanovic A. Kurtovic N.K. Sretenovic A. Elezovic I. Tomin D. International Society of Thrombosis and Hemostasis Scoring System for Disseminated Intravascular Coagulation ≥ 6: A New Predictor of Hemorrhagic Early Death in Acute Promyelocytic Leukemia Med. Oncol. 2013 30 478 10.1007/s12032-013-0478-y 23371042
38. Mantha S. Tallman M.S. Devlin S.M. Soff G.A. Predictive Factors of Fatal Bleeding in Acute Promyelocytic Leukemia Thromb. Res. 2018 164 S98 S102 10.1016/j.thromres.2018.01.038 29703492
39. Mantha S. Tallman M.S. Soff G.A. What’s New in the Pathogenesis of the Coagulopathy in Acute Promyelocytic Leukemia? Curr. Opin. Hematol. 2016 23 121 126 10.1097/MOH.0000000000000221 26760586
40. Falanga A. Consonni R. Marchetti M. Locatelli G. Garattini E. Gambacorti Passerini C. Gordon S.G. Barbui T. Cancer Procoagulant and Tissue Factor Are Differently Modulated All- Trans-Retinoic Acid in Acute Promyelocytic Leukemia Cells Blood 1998 92 143 151 10.1182/blood.V92.1.143.413k18_143_151 9639510
41. Hisada Y. Gangaraju R. Hemostatic Abnormalities in Acute Promyelocytic Leukemia: Clinical Implications and Mechanisms Curr. Opin. Hematol. 2025 32 239 244 10.1097/MOH.0000000000000880 40626510 PMC12240473
42. Hambley B.C. Tomuleasa C. Ghiaur G. Coagulopathy in Acute Promyelocytic Leukemia: Can We Go Beyond Supportive Care? Front. Med. 2021 8 722614 10.3389/fmed.2021.722614 PMC8415964 34485349
43. Ma G. Liu F. Lv L. Gao Y. Su Y. Increased Promyelocytic-Derived Microparticles: A Novel Potential Factor for Coagulopathy in Acute Promyelocytic Leukemia Ann. Hematol. 2013 92 645 652 10.1007/s00277-013-1676-6 23344645
44. Wang Y. Wang C. Zuo N. Yang H. Fang S. Shi J. Extracellular Traps Increase Burden of Bleeding by Damaging Endothelial Cell in Acute Promyelocytic Leukaemia Front. Immunol. 2022 13 841445 10.3389/fimmu.2022.841445 35479063 PMC9035902
45. Yang M.H. Chang K.J. Li B. Chen W.S. Arsenic Trioxide Suppresses Tumor Growth through Antiangiogenesis via Notch Signaling Blockade in Small-Cell Lung Cancer Biomed Res. Int. 2019 2019 4647252 10.1155/2019/4647252 31093499 PMC6481139
46. Hrgovic I. Zöller E. Doll M. Hailemariam-Jahn T. Jakob T. Kaufmann R. Meissner M. Kleemann J. Arsenic Trioxide Decreases Lymphangiogenesis by Inducing Apoptotic Pathways and Inhibition of Important Endothelial Cell Receptors Curr. Issues Mol. Biol. 2024 46 67 80 10.3390/cimb46010006 38275666 PMC10813910