A scoring system to predict the risk of death during induction with anthracycline plus cytarabine-based chemotherapy in patients with de novo acute myeloid leukemia§

Authors


  • David Valcárcel and Pau Montesinos conceived the study, analyzed, and interpreted the data; David Valcárcel and Pau Montesinos wrote the paper; David Valcárcel and Pau Montesinos performed the statistical analyses; Isabel Sánchez-Ortega, Salut Brunet, Jordi Esteve, David Martínez-Cuadrón, José M. Ribera, Mar Tormo, Javier Bueno, Rafael Duarte, Andrés Llorente, P. Torres, Ramón Guardia, Miguel A Sanz, and Jorge Sierra included data of patients treated in their institutions, reviewed the manuscript, and contributed to the final draft.

  • §

    We thank the physicians and centers participating in the study: J. Esteve, Clinical Hospital (Hospital Clínic), Barcelona, Spain; S. Brunet, Santa Creu i Sant Pau Hospital (Hospital de la Santa Creu i Sant Pau) , Barcelona, Spain; J. Berlanga, Catalan Institute of Oncology (Institut Català d' Oncologia), Hospitalet de Llobregat, Spain; J. M. Ribera, Germans Trias i Pujol Hospital (Hospital Germans Trias i Pujol), Badalona, Spain; M. Tormo, University Clinical Hospital (Hospital Clínic Univeristari), Valencia, Spain; J. Bueno, Vall d' Hebrón Hospital (Hospital de la Vall d' Hebrón), Barcelona, Spain; A. Llorente, Joan XXIII Hospital (Hospital Joan XXIII), Tarragona, Spain; M. Queipo de Llano, Clinical Hospital of Málaga (Hospital Clínico de Málaga), Spain; J. Besalduch, Son Dureta Hospital (Hospital Son Dureta), Palma de Mallorca, Spain; C. Pedro, del Mar Hospital (Hospital del Mar), Barcelona, Spain; J. M. Sánchez Villegas, Arnau de Vilanova Hospital (Hospital Arnau de Vilanova), Lleida, Spain; J. M. Moraleda, Virgen de la Arrixaca Hospital (Hospital Virgen de la Arrixaca), Murcia, Spain; J. M. Martí, Mutua Hospital (Hospital Mutua), Tarrasa, Spain; Ll. Font, Verge de la Cinta Hospital (Hospital Verge de la Cinta), Tortosa, Spain; J. Bargay, Son Llatzer Hospital (Hospital Son Llatzer), Palma de Mallorca, Spain; P. Vivancos, Teknon Clinic (Clínica Teknon), Barcelona, Spain; D. Hernández, La Paz Hospital (Hospital La Paz), Madrid, Spain; and Guillermo Martín, Jesús Martínez, Ignacio Lorenzo, and Javier Palau, University Hospital La Fe (Hospital Universitario La Fe), Valencia, Spain.

Abstract

BACKGROUND:

A prognostic index to predict induction death in adult patients receiving induction chemotherapy for de novo acute myeloid leukemia (AML) was developed.

METHODS:

The authors analyzed 570 patients (aged 16-70 years) included in 2 multicenter trials of the CETLAM Group to develop a scoring system (study cohort). The scoring system was tested in 209 patients from an external single institution (validation cohort). Induction regimens consisted of anthracycline and cytarabine combination with or without etoposide. Induction death was defined as death in the first 42 days without evidence of leukemic resistance.

RESULTS:

The cumulative incidence of induction death was 11% in the study cohort and 18% in the validation cohort. Median age was 48 years in the study cohort and 56 years in the validation cohort (P < .001). Multivariate analysis in the study cohort showed the following adverse risk factors for induction death: leukocyte count >100 × 109/L, serum creatinine >1.2 mg/dL, and age ≥50 years. According to these factors, the authors developed a predictive score: low risk (no risk factors), intermediate risk (1 factor), and high risk (2 or 3 factors). The cumulative incidence of induction death in the 3 respective groups was 5%, 13%, and 26% (P < .001). The scoring system was applied in the validation cohort, resulting in cumulative incidence rates of induction death of 6%, 19%, and 32%, for the low-risk, intermediate-risk, and high-risk categories, respectively (P < .001).

CONCLUSIONS:

By using this validated and simple scoring system, the risk of induction death in patients with AML can be predicted accurately. The score may be helpful to design risk-adapted induction strategies. Cancer 2011;. © 2011 American Cancer Society.

INTRODUCTION

Induction chemotherapy of acute myeloid leukemia (AML), the first step to cure this disease, is associated to substantial mortality because of infections, hemorrhages, and other toxic effects of chemotherapy or the leukemia itself. Although supportive measures to prevent life-threatening events (eg, transfusions and antimicrobial agents) have improved in the last years, up to 10% to 20% of patients die during the induction phase.1-4 Because the identification of patients at high risk of induction death may allow the individualization of the chemotherapy regimens and supportive measures, the development of a scoring system to predict the risk of induction death is of interest.

Several risk factors have been associated with higher rates of induction death, such as older age,1 high leukocyte count,2, 3 and bad performance status (PS),1, 4 but only few studies have been designed to build scoring systems to predict accurately the risk of induction death based on pretreatment patient and disease characteristics.5-7 Although demonstrating its prognostic value to predict induction death in patients with AML, these scoring systems often include a high number of variables and may not be easy to use in daily clinical practice. Furthermore, these studies have been performed in series including mostly elderly patients with AML,6, 7 who are often considered unfit for intensive regimens. In addition, as far as we know, none of these scoring systems to predict induction death has been validated in an external cohort of AML patients.

The objective of this study was to identify the risk factors for induction death in a large cohort of 570 consecutive patients treated between 1999 and 2006 in 20 Spanish centers of the CETLAM group (Grupo Cooperativo para el Estudio y Tratamiento de las Leucemias Agudas y Mielodisplasias). By using these risk factors, we aim to build a scoring system to predict induction death. The external validity of the scoring system was tested in an independent set of patients from a single institution.

MATERIALS AND METHODS

Patients and Treatment

Study cohort

From June 1999 to February 2006, 578 patients diagnosed with de novo AML (except acute promyelocytic leukemia [APL]) from 20 hospitals were enrolled in 2 consecutive multicenter Spanish CETLAM trials (LMA-99 [N = 324] and LMA-03 [N = 244]). Age limits for inclusion were 16 and 60 years in LMA-99 and 18 and 70 years in LMA-03. The LMA-99 induction chemotherapy consisted of idarubicin 12 mg/m2 intravenously (IV) on days 1, 3, and 5; etoposide 100 mg/m2 IV on days 1, 2, and 3; and cytarabine 500 mg/m2 twice a day by IV infusion over 2 to 3 hours on days 1, 3, 5, and 7. Priming with granulocyte colony-stimulating factor (G-CSF) at daily dose of 150 μg/m2 subcutaneously was added to the same regimen in the LMA-03 protocol, beginning 24 hours before treatment and lasting until the last day of chemotherapy. G-CSF was not administered or discontinued if the white blood cell (WBC) count exceeded 30 × 109/L and was started or resumed when the value decreased below that level.

Validation cohort

This cohort included 209 patients diagnosed with de novo AML (not APL) consecutively treated with induction chemotherapy from January 1996 to March 2008 at a single independent institution (Hospital Universitario La Fe, Valencia, Spain). Induction chemotherapy regimen consisted of idarubicin 12 mg/m2 IV on days 1 to 3 and cytarabine 200 mg/m2 by continuous 24-hour IV infusion on days 1 to 7, in 137 patients (66%). In 47 patients (22%), chemotherapy comprised mitoxantrone 10 mg/m2 IV on days 1 to 3 and cytarabine 150 mg/m2 by continuous 24-hours IV infusion on days 1 to 7. The remaining 25 patients (12%) received idarubicin 12 mg/m2 IV or daunorubicin 60 mg/m2 on days 1 to 3 combined with cytarabine 100 mg/m2 by continuous 24-hour IV infusion on days 1 to 7, plus etoposide 100 mg/m2 IV on days 1 to 3. Priming with G-CSF was not administered.

Supportive measures including anti-infectious prophylaxis and blood product transfusions were determined by local protocols. Management of febrile episodes was not standardized and varied according to the practice of each institution. Written informed consent before chemotherapy was obtained according to institutional guidelines, and the research ethics board approved the protocols at each center according to the Declaration of Helsinki.

Definitions and Study Endpoints

AML diagnosis was made according to French-American-British (FAB)8, 9 classification in the LMA-99 trial and based on the World Health Organization (WHO) criteria10 in the LMA-03 trial. In the validation cohort, the FAB classification was used before 2001 and the WHO from 2002. FAB M0 and M7 AML subtypes were immunologically confirmed.9, 10 Results of cytogenetic studies were categorized according to the Medical Research Council classification.11 Patients with genetically diagnosed APL, as well as those with secondary AML (ie, with documented antecedents of preleukemic disease or exposure to leukemogenic agents) were excluded from the study.

Remission induction response was assessed according to the recently revised criteria by Cheson et al.12 A morphologic complete remission (CR) designation requires <5% blasts in a bone marrow (BM) aspirate sample, an absolute neutrophil count >1 × 109/L, and platelets >100 × 109/L. Partial remission (PR) was defined as the persistence of ≥5% blasts in BM but with a reduction of at least 50% compared with diagnosis. If possible, those patients received a second induction course identical to the first course. Otherwise, patients not included in previous response categories were considered as resistant to induction chemotherapy.

Induction death was defined as death from any cause within the first 42 days from start of chemotherapy, unless PR or resistance was observed (ie, patients dying in the first 42 days in whom chemotherapy resistance was documented, as well as those receiving a second course of induction chemotherapy because of PR, were not considered as having induction death). Causes of induction death were determined after review of patient discharge summaries, in-hospital reports, or autopsy results if available. Causes of induction death include the following categories:

  • Infection, when death is because of a clinical, radiological, or microbiologically documented infection.

  • Hemorrhage, when a major bleeding occurs in a vital organ (central nervous system, lungs). Gastrointestinal hemorrhage requires massive melena or hematemesis accompanied by fall in blood pressure.

  • Other causes, that is, any other cause not classified as infection or hemorrhage.

Statistical Analysis

The chi-square test, with Yates correction if necessary, Mann-Whitney U test, and t test were used to analyze differences in the distribution between patient subsets of categorical, and continuous nonparametric and parametric variables, respectively. A univariate analysis was performed in the study cohort to determine the possible relationship between patient and disease characteristics and induction death. The probability of induction death was estimated by the cumulative incidence method (for marginal probability),13, 14 and was calculated from the date of start of chemotherapy. Patients with resistance before death in the first 42 days, as well as those starting a second course of chemotherapy, were considered as having a competing risk event for induction death at the time of resistance or start of the second course, respectively. Age, serum levels of lactate dehydrogenase (LDH) and creatinine, WBC counts in peripheral blood (PB), percentage of blasts in BM, and time between diagnosis and treatment were analyzed as continuous variables in a Cox univariate analysis. If any of these parameters was identified as statistically significant, the best cutoff value was investigated. Other categorical parameters included in the univariate analysis were sex, FAB subtype, treatment protocol, cytogenetics, presence of FLT3 internal tandem duplication, and MLL molecular status. Characteristics selected for inclusion in the multivariate analysis, using a Cox stepwise logistic regression, were those for which there was at least a trend for association in univariate analysis (P < .1). The variables remaining significant (P < .05) in the multivariate analysis were used to construct a scoring system to classify the patients in groups according to their risk of induction death. The hazard ratio (HR) of the logistic regression model was used to assign the corresponding score weight of every risk factor. To validate the predictive model, patients of the validation cohort were classified in risk categories according to the scoring system. The probability of induction death was calculated in each risk category using the cumulative incidence method. Computations were performed using SPSS version 15 (SPSS, Chicago, Ill), with the exception of cumulative incidence analyses, which were carried out with NCSS 2004 (Number Cruncher Statistical System; NCSS, Kaysville, Utah).

RESULTS

Patients

Study cohort

Five hundred seventy-eight patients were consecutively enrolled in the multicenter LMA-99 and LMA-03 trials; of them, 6 patients were excluded because they died before chemotherapy, and 2 because they were lost to follow-up. The main characteristics of the 570 patients included in the study cohort are shown in Table 1. Briefly, the median age was 48 years (range, 16-70), 57% were male, and median WBC was 14.6 × 109/L (range, 0.7-410 × 109/L). Seventy-three (13%) patients had a WBC count >100 × 109/L. Median serum creatinine value was 0.9 mg/dL (range, 0.1-3.9 mg/dL); 61 (11%) patients had a creatinine value >1.2 mg/dL, which was considered the upper limit of normal value. Overall, 411 (71.9%) patients achieved CR, after 1 (n = 338; 59.2%) or 2 (n = 73; 12.8%) induction courses; 96 (16.8%) patients were resistant, and 63 (11.4) patients died.

Table 1. Patient Characteristics
CharacteristicStudy Cohort, n=570Validation Cohort, n=209P
Median [range]No. (%)Median [range]No. (%)
  1. Abbreviations: FAB, French-American-British; LDH, lactate dehydrogenase; MRC, Medical Research Council; WBC, white blood cells.

Male sex 322 (56.5)114 [54.6] .63
Age, y48.6 [16-70] 56.5 [16-70]  
 ≥50 275 (48.2) 130 (62.2)<.001
FAB classification     
 M0 32 (5.6) 10 (4.8).003
 M1 115 (20.2) 57 (27.3) 
 M2 127 (22.3) 57 (27.3) 
 M4 128 (20.7) 33 (15.8) 
 M5 114 (20) 29 (13.9) 
 M6 20 (3.5) 15 (7.2) 
 M7 4 (0.7) 2 (0.9) 
 Unknown/not done 40 (7) 4 (1.9) 
Cytogenetic risk, MRC     
 Favorable 75 (13.2) 20 (9.6).55
 Intermediate 330 (57.9) 126 (60.3) 
 Adverse 92 (16.1) 33 (15.8) 
 Unknown/not done 74 (12.8) 30 (14.4) 
LDH, IU/L, n=467678 [70-12,354] 662 [65-25,005]  
 >2000 64 (13.7) 20 (9.6).56
Creatinine, mg/dL, n=5050.9 [0.11-3.9] 0.9 [0.25-2.4]  
 >1.2 55 (9.6) 39 (18.7).005
WBC × 109/L14.6 [0.7-410] 10.0 [0.6-385]  
 >100 73 (12.8) 22 (10.5).43
Bone marrow blasts, %68 [11-100] 72 [10-100] .89

Validation cohort

This cohort included 209 patients (45% female) with a median age of 56 years (range, 16-70 years). Median WBC at diagnosis was 10 × 109/L (range, 0.6-385 × 109/L). Twenty-two (10%) patients had a WBC count >100 × 109/L. The median creatinine value was 0.9 mg/dL (range, 0.25-2.4 mg/dL); 39 patients (18.7%) had a creatinine value >1.2 mg/dL (Table 1). Overall, 128 patients achieved CR (61%), after 1 (n = 113; 54%) or 2 (n = 15; 7%) induction courses; 44 patients (21%) were considered resistant, and 37 patients (18%) died during induction (of them, 35 died in the first 42 days after the start of chemotherapy and were considered as induction death).

The main differences between patients in the study and validation cohorts were the higher frequency of patients older than 50 years, levels of creatinine >1.2 mg/dL, and lower frequency of FAB monocytic subtypes in the validation cohort (P < .001, P = .005, and P = .003, respectively) (Table 1).

Causes of Induction Death

In the study cohort, 60 patients (9.5%) died in the first 42 days after induction therapy at a median of 20 (range, 1-38) days. The causes of death were infection (n = 35), hemorrhage (n = 7), and other causes (n = 18; ie, multiorgan failure [n = 8], respiratory failure [n = 5], renal failure [n = 2], cardiac failure [n = 1], and indeterminate [n = 2]). Three patients died in CR after the first 42 days; 2 died of myocardial infarct after 45 and 99 days, and in the third the cause of death was tuberculosis on day 66.

In the validation cohort, 35 patients (18%) died in the first 42 days after induction therapy, at a median of 18 days (range, 1-39 days). The causes of death were infection (n = 20), hemorrhage (n = 6), and other causes (n = 9; ie, multiorgan failure [n = 4], respiratory failure [n = 2], renal failure [n = 2], and cardiac failure [n = 1]).

Prognostic Factors for Induction Death in the Study Cohort

In univariate analysis, the factors associated with an increased risk of induction death were age ≥50 years, WBC >100 × 109/L, serum creatinine >1.2 mg/dL, serum LDH >2000 UI/L, BM blasts >70%, and male sex (Table 2). In the multivariate analysis, the variables that retained statistical significance were age ≥50 years (HR, 2.1; 95% confidence interval [CI], 1.2-3.9; P = .009), serum creatinine >1.2 mg/dL (HR, 2.1; 95% CI, 1.2-3.8; P = .025), and WBC >100 × 109/L (HR, 2.3; 95% CI, 1.2-4.4; P = .015) (Table 2).

Table 2. Univariate and Multivariate Analyses of Induction Death in the Study Cohort
Adverse CategoryUnivariate Analysis, PMultivariate Analysis
HR (95% CI)P
  1. Abbreviations: BM, bone marrow; CI, confidence interval; HR, hazard ratio; LDH, lactate dehydrogenase; NS, not significant.

Age ≥50 years.0122.1 (1.2-3.6).005
Leukocytosis >100 × 109/L.0041.9 (1-3.6).05
Creatinine >1.2 mg/dL<.0012 (1.1-3.9).03
LDH >2000 UI/L.02NS
Blasts in BM >70%.085NS
Male sex.053NS

Development of the Scoring System in the Study Cohort

The scoring system consisted of 3 variables (age, serum creatinine, and WBC), each with the same weight (1 point for each if the value was in the bad risk category). Five hundred fifty-four patients (97%) had all the scoring index variables available. Three risk groups were established: low risk (0 points, ie, no risk factors), intermediate risk (1 point, ie, 1 risk factor), and high risk (2 or 3 points, ie, 2 or 3 risk factors). The distribution of patients according to the scoring system was as follows: low risk 43%, intermediate risk 46%, and high risk 11% (Table 3). The cumulative incidence rates of induction death among patients in the low-risk, intermediate-risk, and high-risk categories were 5%, 13%, and 26%, respectively (P < .001) (Fig. 1). The relative risk of induction death in the intermediate-risk and high-risk groups compared with the low-risk category was 2.5 (95% CI, 1.3-4.8; P = .004) and 5.4 (95% CI, 2.6-11.4; P < .001), respectively (Table 3).

Figure 1.

Cumulative incidence of induction death is shown according to scoring system in the study cohort.

Table 3. Scoring System in the Study Cohort
Number of Risk FactorsaNumber of Patients (%)Risk ScoreNumber of Patients (%)Risk of Induction Death
HR [95% CI]P
  • Abbreviations: CI, confidence interval; HR, hazard ratio.

  • a

    Risk factors were: creatinine level >1.2 mg/dL; white blood cell count >100 × 109/L; age ≥50 years.

0239 (43)Low239 (43)1 
1257 (46)Intermediate257 (46)2.5 [1.3-4.8].004
244 (8)High58 (11)5.4 [2.6-11.4]<.001
314 (3)

There were no significant differences either in the cause of death or in the median days to death between the 3 groups (data not shown).

Validation of the Scoring System

All patients of the validation cohort had the 3 variables of the scoring system available. Seventy-one patients (34%) were classified as low-risk, 97 (46%) as intermediate-risk, and 41 (20%) as high-risk patients (Table 4). The cumulative incidence rates of induction death were 6%, 19%, and 32% in the low-risk, intermediate-risk, and high-risk groups, respectively (P < .001) (Fig. 2). The relative risk of induction death in the intermediate-risk and high-risk groups compared with the low-risk category was 3.6 (95% CI, 1.2-10.6; P = .01) and 6.8 (95% CI, 2.4-19.2, P < .001), respectively (Table 4).

Figure 2.

Cumulative incidence of induction death is shown according to scoring system in the validation cohort.

Table 4. Scoring System in the Validation Cohort
Number of Risk FactorsaNumber of Patients (%)Risk ScoreNumber of Patients (%)Risk of Induction Death
HR [95% CI]P
  • Abbreviations: CI, confidence interval; HR, hazard ratio.

  • a

    Risk factors were: creatinine level >1.2 mg/dL; white blood cell count >100 × 109/L; age ≥50 years.

071 (34)Low71 (34)11
197 (46)Intermediate97 (46)3.6 [1.2-10.6].01
229 (14)High41 (20)6.8 [2.4-19.2]<.001
312 (6)

DISCUSSION

This study shows that induction death in patients with AML can be predicted accurately using a simple and novel scoring system based on 3 universally available patient and disease baseline characteristics (age, leukocytes in PB, and serum creatinine levels). The score was raised in a large series of patients with de novo AML homogenously treated with intensive regimens containing idarubicin, cytarabine, and etoposide. The score was validated in an independent set of patients from a single institution that received induction chemotherapy with anthracycline and cytarabine (with or without the addition of etoposide). This validated scoring system may allow selecting patients with high risk for induction death to adapt chemotherapy intensity and supportive care measures.

Many attempts have been made to establish the patient and disease characteristics associated with a worse outcome of induction chemotherapy in patients with AML,1-5 but only a limited number of studies have developed scoring systems to predict induction death.5-7 Our study differs from the previous studies in some aspects: 1) it was performed in the context of a multicenter trial, in contrast to those of Estey et al, Giles et al, and Kantarjian et al, which were performed at a single institution (The University of Texas MD Anderson Cancer Center [MDACC]); 2) the series of Gilles et al and Kantarjian et al, were focused on elderly patients with AML and high-risk myelodysplastic syndrome (MDS), in contrast to ours, in which only adult patients younger than 71 years old were included, whereas MDS and secondary AML were excluded; and 3) induction chemotherapy regimens were relatively similar in all patients, in contrast to those administered at MDACC (in many patients regimens included investigational schedules with topotecan or fludarabine). We should highlight that our scoring system was raised in patients treated with anthracycline plus cytarabine regimens and therefore is only applicable in this setting. Conversely, the MDACC studies analyzed the predictive value of scoring systems using different definitions for induction death than in our study (death in the first 42 days in ours vs in the first 28 or 56 days in the MDACC studies). The main difference, and the most innovative finding of our study, compared with previous attempts, is that our scoring system was specifically developed and externally validated in AML patients.

Our scoring system was based on 3 patient and disease characteristics that are routinely available and that have been previously reported as prognostic factors for CR and/or induction death: age,1 WBC counts in PB,2-4 and serum creatinine levels. The high WBC counts may increase the risk of coagulation disorders, pulmonary and central nervous system leukostasis,3 and renal failure, increasing the mortality hazard.15 Increased levels of serum creatinine have been consistently associated with the risk of tumor lysis syndrome, which in turn is related to higher rates of induction death.16 Conversely to WBC and creatinine levels, age is a prognostic factor correlated not only with an increased risk of induction death, but also with increased rates of leukemic resistance to induction chemotherapy and with shorter remission duration. In contrast to our findings, Greenwood et al4 reported that age had no influence on induction death risk when WBC and PS (measured by the Eastern Cooperative Oncology Group scale) were included in the analysis. This discordance may be explained, at least in part, by the close relation between age and poor PS.1 We acknowledge that our scoring system could probably be improved by the inclusion of PS and other relevant parameters (eg, comorbidities, presenting platelet count, coagulopathy, or uricemia). Unfortunately, these variables were not available in the vast majority of patients. The usefulness and applicability of the present scoring system is limited, especially in elderly AML patients, because of the noninclusion of comorbidities or PS, and further studies are needed to show the real impact of age versus patient health status on induction death. It is possible, however, that the simplicity of our scoring system is a consequence of the relatively limited number of variables that have been included in univariate analysis.

Concerning the risk stratification for induction death, our scoring system classified AML patients in 3 groups, identifying a group with 5-fold higher risk of induction death compared with the lowest risk group. Of note, the study and the validation cohorts differed in some baseline characteristics, the most important being higher median age and creatinine levels in the validation cohort. Consequently, the overall induction death rate was higher in the validation cohort compared with the study cohort. However, when the scoring index was applied, the induction death rates were similar in both the study cohort and the validation cohort according to the risk category. In comparison with previous predictive models, our scoring system is relatively simple, involving 3 variables that are widely and rapidly available in the routine clinical care. In fact, the predictive models of Estey et al5 and Kantarjian et al7 consisted of equations with 8 and 5 variables, respectively. The study of Giles et al6 applied to elderly patients with AML the comorbidity index score previously developed for hematopoietic cell transplantation recipients by Sorror et al,17 based on 15 complex variables.

Our scoring system to predict the risk of induction death in adult patients with de novo AML might have implications for patient care and health organization. For example, high-risk patients might be referred to centers with more expertise to receive rigorous supportive care (an interesting approach in the setting of cooperative groups). Also, elderly high-risk patients could receive modified induction regimens with less intensive chemotherapy. However, our scoring system is not optimal for the risk-adapted implementation of specific supportive measures (ie, antibiotic therapy or transfusions), as it does not assess for the risk of specific life-threatening complications (ie, infections or hemorrhages). In addition, the score could be helpful in selecting obvious candidates for intensive chemotherapy, mainly those with low and intermediate risk of induction death. However, patient selection depends also on some variables predicting induction death that were not included in our analysis, including comorbidities, PS, and others. In addition, leukemic characteristics affecting the chances of achieving a CR, such as cytogenetics and FLT3 internal tandem duplication mutations, as well as other circumstances (eg, patient, family, and physician preferences), will influence patient selection.

In conclusion, our validated scoring system allows easy risk stratification for induction death in the clinical setting. This stratification may be helpful in designing risk-adapted induction strategies in adult patients younger than 71 years with de novo AML.

FUNDING SOURCES

This study was supported by grants G03/008, PI051162, PI051433, PI052312, PI080672 and EC07/90065 from the Health Research Fund - Carlos III Health Institute, Ministry of Health (Fondo de Investigaciones Sanitarias - Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo), grants RD06/0020/0031 and RD06/0020/0101 from the Thematic Network of Cooperative Research on Cancer (Red Temática de Investigación Cooperativa en Cáncer) from Mutua Madrileńa Foundation (Fundación Mutua Madrileńa) and from Cellex Foundation (Fundación Cellex) and by grant 2006/0127 from the Research Foundation of University Hospital La Fe-Bancaja Aids (Fundación para la Investigación Hospital Universitario La Fe-Ayudas Bancaja).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary