Association of CDSS score and 60-day mortality in Chinese patients with non-APL acute myeloid leukemia: a retrospective cohort study

The latest CDSS was established in 2017 by a consensus of Chinese experts on the diagnosis and treatment of DIC [14]. According to this consensus, DIC is diagnosed when a patient with hematologic malignancies has abnormal laboratory findings [14].This consensus assigns quantitative weights to individual laboratory and clinical parameters for DIC diagnosis. The CDSS criteria includes following five rules. (1) A primary disease that causes DIC, hematologic malignancy e.g.is present, 2 points. (2) Prolongation of PT and APTT: PT < 3S and APTT < 10 S,0 point; PT ≥ 3S and APTT ≥ 10 S,1 point; PT > 6S, 2 points. (3)D-dimer level: < 5 μg/mL, 0 point; ≥ 5 μg/mL, 2 points; ≥ 9 μg/mL, 3 points. (4) Fibrinogen level: ≥ 1.0 g/L,0 point; < 1.0 μg/mL, 1point. (5) Platelet counts: > 50 × 10⁹/L,0 point; Platelet counts < 50 × 10⁹/L,1 point; Diagnosis of DIC is CDSS score ≥ 6 points for AML Patients (Table S1). All, except PLT count, were measured using the blood coagulation analyzer Sysmex CS-5100 System TM (Siemens Healthcare Diagnostics, Erlangen, Germany) with the corresponding diagnostic reagents kit (Siemens Healthcare Diagnostics Products, Germany, Marburg, Germany). and the reference interval for PT was 9–13 s, APTT 20–40 s, Fib 2–4 g/L, and DD 0–0.55 mg/L, in accordance with the national standard. Commercially available control materials for internal quality control (IQC) (Siemens Healthcare Diagnostics Products, Marburg, Germany) were used to test blood samples daily. PLT counts were obtained from routine blood samples, measured using a Sysmex XE-2100 or XN-1000 s. All five items, including the CDSS score, were included in external quality assessment (EQA) activities organized by the National Center for Clinical Laboratories (NCCL) every year, and criteria for feedback reports were fulfilled during this study. All data (except PLT) from this study, including CDSS and other items/parameters, were the first test results of patients after admission for AML. If the CDSS score of the patients was ≥ 6, they were diagnosed with overt DIC, and if the CDSS score was 5 points with PLT ≥ 50 × 10⁹/L for the first time, another two PLT count were required for monitoring; if the PLT decreased by 50% within 48 h, the CDSS score would be calculated repeatedly.

The primary endpoint and outcome of interest were death within 60 days of admission. The data collectors for clinical information at the first diagnosis were blinded to the survival data.

Data, including survival status, were collected from the electronic medical record system or follow-up telephone calls. Baseline examinations included FAB subtype, genomic risk category, bleeding and thrombosis, pulmonary infection, blood and bone marrow (BM) parameters. The biomarkers included white blood cell (WBC) count, PLT count, prothrombin time (PT), activated partial thromboplastin time (APTT), DD, Fg, creatine kinase isoenzyme-MB (CK-MB), myoglobin (Myo), antithrombin (AT), albumin (Alb), creatinine (Crea), glucose (Glu), serum ferritin (SF), and BM blast. All laboratory data were focused on measurements performed within the first 48 h of admission to reduce the probability that serum biomarker levels were affected by anti-leukemia and transfusion therapy. These parameters comprise routine testing, which physicians commonly use to evaluate a patient’s physical condition. The chemotherapy was completed within 60 days.

This study aimed to observe the impact of CDSS score on 60-day mortality in patients with AML. The patients were divided into two groups based on their CDSS scores. A descriptive analysis was applied to all patients. Continuous data were expressed as mean and standard deviation (SD) or median and interquartile range (quartile 1–quartile 3 [interquartile range (IQR)], as appropriate. Categorical variables were expressed as proportions (%). Variables were compared using the chi-square test (categorical variables), one-way ANOVA (normal distribution), and Kruskal–Wallis (skewed distribution) tests [19].

Multivariate Cox regression analysis was used to assess the independent association between CDSS score and 60-day mortality. In the analysis, confounding factor is an important issue, we performed some different statistical models to verify the results’ stability. In the final model, we adjusted the factors basing the following two rules. (1) For univariate analysis, we adjusted for variables, of which the p values were less than 0.1(Table S2). (2) For multivariate analysis, variables were chosen on the basis of previous findings and clinical constraints. Dummy variables were used to indicate the missing covariance values, if the missing data variables were greater than 10%. Considering the strong heterogeneity of WBC in leukemia, WBC were used as a dichotomous variable to adjust in multivariate Cox regression analysis.

Survival curves were plotted by Kaplan–Meier and log-rank analyses. Subgroup analyses were stratified by relevant effect covariates. Analyses were stratified according to the results of the univariate analysis (P-value < 0.1) and basis of previous findings and clinical constraints, including sex, age, WBC, Glu, ALB, chemotherapy and bleeding, to examine the effect of these factors on the above associations. The likelihood ratio test was used to assess effect modification according to prespecified subgroups using interaction terms between subgroup indicators and CDSS scores. Interactions across subgroups were tested using the likelihood ratio test. All analyses were performed using R 3.3.2 (http://​www.​R-project.​org, The R Foundation) and Free Statistics (version 1.5). Differences with a two-sided P < 0.05 were considered statistically significant [22].

Of the 706 newly diagnosed patients with AML during the study period, 570 ones were eligible for analysis (Fig. 1). The descriptive characteristics of the eligible study population with a CDSS < 6 and ≥ 6 are reported in Table 1. Patient age was 51.3 (SD 20.4) years (range, 1–94 years), 306 (53.7%) were male, and 137 (24.0%) suffered from 60-day mortality. In addition, 342 (60.1%) patients received combined chemotherapy, and 92 (16.2%) patients single chemotherapy drugs. The combined chemotherapy regimens were a combination of cytarabine and anthracycline “7 + 3” or a combination of cytarabine and the other, 7 patients were treated with the combination of cytarabine and venetoclax. Single chemotherapy meant patients only used decitabine or hydroxyurea [23]. The median CDSS score level was 3.4 ± 1.4 points (range, 0–9 points). 59 (10.4%) cases had CDSS scores ≥ 6 points. The 60-day mortality of patients with CDSS score ≥ 6 was as high as 47.5%, significantly higher than those < 6 at 21.3% (P < 0.001).

Additionally, among CDSS laboratory integral items, the PLT lower than 50 × 10⁹/L had the highest abnormal rate with 61.1% (348 cases), followed by the D-dimer (≧5.0 mg/L) 24.2% (138 cases). There were 8 cases (1.4%) with prolongation of APTT (≧10 S) and 51 cases (8.9%) that of PT prolongation (≧3 S). Among 59 patients with overt DIC, the abnormality rate of DD was as high as 96.6% (57 cases, including DD ≥ 5 mg/L in 11 cases and ≥ 9 mg/L in 46 cases), followed by PLT < 50 × 10⁹/L in 93.2% (55 cases), prolongation of PT (≥ 3S), 40.7% (24 cases), Fg(< 1.0 g/L) 18.6%(11 cases), prolongation of APTT (≥ 10S) 4.1% (4 cases).

We evaluated the cohort with ISTH creteria, and found there were 89 (15.6%) patients who had elevated ISTH scores (≥ 5) diagnostic of DIC, with early mortality 42.7%, among those patients there were 58 ones with CDSS score ≥ 6, 28 patients suffered from early death. The morbidity of DIC was 10.4% according to creteria of CDSS, which was lower than that of ISTH one, but the mortality was 47.5%, higher than the former.

In the extended multivariate Cox models (Table 2), we observed that the hazard ratios (HRs) of the CDSS score (increase per 1 point) were consistently significant in all three models (HRs range 1.26–1.39). Patients with a CDSS score ≥ 6 (overt DIC) had a 189% higher risk of 60-day mortality than patients with a CDSS score < 6 (non-overt DIC) (HR = 2.89, 95% confidence interval (CI) 1.91–4.38). After adjusted for all covariates, results showed a 198% higher 60-day mortality rate in patients with a CDSS score ≥ 6 (HR = 2.98, 95% CI 1.24–7.19, model III, Table 2) than in those with that < 6. The covariates were confirmed after a univariate analysis of risk factors with a P-value ≤ 0.027 (Table S2). Dummy variables were used for serum ferritin due to its missing data was14%. In this observation population, WBC ranged from 0.29 × 10⁹/L to 540.52 × 10⁹/L, with a median of 19.7 × 10⁹/L. We divided WBC into two groups with 20 × 10⁹/L, then used binary WBC as one of the adjusted variables of multivariate Cox regression analysis. We also analyzed the data using the International Society on Thrombosis and Hemostasis (ISTH) criteria and obtained similar results (Table S3). The Kaplan–Meier curve showed that mortality was higher by day 60 in patients with a CDSS score ≥ 6 (log-rank test: P < 0.0001, Fig. 2).

Analyses and interactive analyses were stratified according to confounders to detect whether the association between CDSS score and 60-day mortality of AML was present in different subgroups. The confounders that were confirmed after a univariate analysis of risk factors (Table S2) and previous findings and clinical constraints, included age, sex, WBC, Glu, ALB, chemotherapy and bleeding (Fig. 3). No significant interaction was observed in any subgroups (P-value for interaction > 0.05).

There were 137 patients suffered from 60-day mortality, the main causes were organ failure, hemorrhage or Thrombosis, infection, particular pulmonary infection (Table 3).