Devlin, Sean M.

Abstract Artificial intelligence (AI)–enabled interpretation of electrocardiogram (ECG) images (AI-ECGs) can identify patterns predictive of future adverse cardiac events. We hypothesized that such an approach would provide prognostic information for the risk of cardiac complications and mortality in patients undergoing hematopoietic cell transplantation (HCT). We retrospectively subjected ECGs obtained before HCT to an externally trained, deep-learning model designed to predict the risk of atrial fibrillation (AF). Included were 1377 patients (849 autologous [auto] HCT and 528 allogeneic [allo] HCT recipients). The median follow-up was 2.9 years. The 3-year cumulative incidence of AF was 9% (95% confidence interval [CI], 7-12) in patients who underwent auto HCT and 13% (10%-16%) in patients who underwent allo HCT. In the entire cohort, pre-HCT AI-ECG estimate of AF risk correlated highly with the development of clinical AF (hazard ratio [HR], 7.37; 95% CI, 3.53-15.4; P < .001), inferior survival (HR, 2.4; 95% CI, 1.3-4.5; P = .004), and greater risk of nonrelapse mortality (NRM; HR, 95% CI, 3.36; 1.39-8.13; P = .007), without increased risk of relapse. Association with mortality was only noted in allo HCT recipients, where the risk of NRM was greater. The use of cyclophosphamide after transplantation resulted in greater 90-day incidence of AF (13% vs 5%; P = .01) compared to calcineurin inhibitor–based graft-versus-host disease prophylaxis, corresponding to temporal changes in AI-ECG AF prediction after HCT. In summary, AI-ECG can inform risk of posttransplant cardiac outcomes and survival in HCT patients and represents a novel strategy for personalized risk assessment.

Abstract Background Acute promyelocytic leukemia (APL) is commonly complicated by complex coagulopathy. The early use of all-trans retinoic acid (ATRA) appears to have reduced induction mortality; however, with modern regimens the risk of early hemorrhagic death remains at about 5-10% in clinical trials and up to 30% in population-based studies. Previously identified as markers of bleeding risk in this setting are: white blood cell count (WBC), peripheral blood blast count, platelet count, fibrinogen, prothrombin time (PT), activated partial thromboplastin time (aPTT) and creatinine. Nonetheless, uncertainty remains as to which ones are independent predictors. Materials and Methods The database for study was from APL patients enrolled on three large clinical trials that used ATRA for induction: APML3 (single arm of ATRA + idarubicin ± prednisone), APML4 (single arm of ATRA + idarubicin + arsenic trioxide + prednisone) and ECOG 2491 (intergroup I0129, consisting of daunorubicin + cytarabine vs ATRA). Known determinants of bleeding were assessed at baseline and considered as potential predictors of hemorrhagic death in the first 30 days of induction. In univariate analysis, using Wilcoxon Rank Sum and Fisher’s exact test, we looked at the association of the parameters mentioned above with hemorrhagic death within 30 days of induction. Results A total of 619 patients were enrolled: ECOG 2491 (intergroup I0129) - 394, APML3 - 101, and APML4 -124. Among these, thirty (4.8%) deaths from bleeding during the first 30 days (early deaths, ED) following induction were recorded. Baseline characteristics of the patients are shown in Table 1 and Table 2. Initial coagulation parameters (PT, aPTT and fibrinogen) were not available for the ECOG trial. Parameters with significant differences between the ED cases and the remaining of the cohort included WBC (median: 7.4 vs 2.4 /Kmcl respectively, p<0.001), peripheral blood blast count (median: 1.0 vs 0.2 /Kmcl respectively, p=0.049) and platelet count (median: 20 vs 33 /Kmcl respectively, p=0.002). Conclusions In this large cohort of approximately 600 APL patients enrolled on 3 large clinical trials, higher WBC, peripheral blast count and lower platelet count emerged as univariate predictors of early hemorrhagic death. Surprisingly, age, PT, aPTT, fibrinogen and creatinine clearance did not differ between the patients who died from bleeding and the remaining of the cohort. Multivariate analysis will clarify the relative strength of the significant predictors in assessing ED, and will be available, along with coagulation parameters for ECOG, at the time of presentation. These data hopefully will allow us to build a clinical prediction rule to identify patients at the highest risk of early hemorrhagic death. Such a stratification tool could prove useful in future trials testing novel approaches to treat the coagulopathy associated with APL. : Abstract 948. Table 1 Baseline Characteristics of Patients, Continuous Variables Variable Total Death from Bleeding within 30 days No Yes N Median (IQR) N Median (IQR) N Median (IQR) p-value Age (years) 619 40.0 (26.0 - 54.0) 589 40.0 (26.0 - 54.0) 30 47.0 (26.0 - 63.0) 0.15 White Blood Cell Count (K/mcL) 611 2.5 (1.2 - 8.2) 582 2.4 (1.1 - 7.6) 29 7.4 (3.1 - 20.7) <0.001 Peripheral blood blast count (K/mcL) 581 0.2 (0.0 - 2.2) 552 0.2 (0.0 - 1.9) 29 1.0 (0.1 - 5.7) 0.049 Hemoglobin (g/dL) 615 9.3 (8.0 - 10.5) 585 9.3 (7.9 - 10.5) 30 9.4 (8.2 - 10.7) 0.75 Platelets (K/mcL) 615 31.0 (17.0 - 56.0) 585 33.0 (17.0 - 57.0) 30 20.0 (13.0 - 34.0) 0.002 Prothrombin time (s) 184 16.0 (13.3 - 18.2) 179 16.0 (13.3 - 18.1) 5 16.7 (16.5 - 144.3) 0.15 Activated partial thromboplastin time (s) 225 29.0 (26.0 - 32.0) 216 29.0 (26.0 - 32.0) 9 31.2 (29.0 - 38.0) 0.07 Fibrinogen (mg/dL) 217 1.8 (1.3 - 2.6) 208 1.8 (1.3 - 2.6) 9 2.3 (2.0 - 2.5) 0.36 Creatinine Clearance (mg/dL)* 225 389 (261.2 - 511.6) 216 392 (264.9 - 512.4) 9 262 (219.2 - 432.0) 0.15 *calculated using the Jelliffe formula : Table 2 Baseline Characteristics of Patients, Categorical Variables. Variable Total Death from Bleeding within 30 days No Yes N % N % N % p-value ECOG Performance Status 579 93.54 552 93.72 27 90.00 0.50 Good(0-2) Poor(3-4) 36 5.82 33 5.60 3 10.00 Missing 4 0.65 4 0.68 0 0 FAB Classification 248 40.06 239 40.58 9 30.00 >0.95 M3 M3v 67 10.82 65 11.04 2 6.67 Missing 304 49.11 285 48.39 19 63.33 Disclosures No relevant conflicts of interest to declare.

Key Points High WBC is an independent predictor of early HD in APL.