Journal of Critical Care (2012) 27, 739.e1–739.e6

Prognostic factors in critically ill patients with hematologicmalignancies admitted to the intensive care unitChang Dong Yeo MDa, Jin Woo Kim MDa, Seok Chan Kim MDa, Young Kyoon Kim MDa,Kwan Hyoung Kim MDa, Hee Je Kim MDb, Seok Leeb, Chin Kook Rhee MDa,⁎

aDivision of Pulmonology and Critical Care Medicine, Department of Internal Medicine, School of Medicine,The Catholic University of Korea, Seoul 137-701, South KoreabDivision of Hematology, Department of Internal Medicine, School of Medicine, The Catholic University of Korea, Seoul,South Korea

0h

Keywords:Prognosis;Hematologic malignancy;Intensive care unit

AbstractObjective: Despite an improvement in the prognosis of patients with hematologic malignancies, themortality of such patients transferred to the intensive care unit (ICU) is high. This study determined thepredictors of mortality in a cohort of critically ill patients with hematologic malignancies admitted tothe ICU.Methods: We studied 227 critically ill patients with hematologic malignancies who were admitted tothe ICU between April 2009 and December 2011. A cohort of consecutive patients with hematologicmalignancies was reviewed retrospectively to identify clinically useful prognostic factors.Results: The ICU mortality rate was 84.1%, and the in-hospital mortality rate was 89.9%. The ICUmortality was significantly higher in patients with acute leukemia than in those with other malignancies.A significant difference between survivors and nonsurvivors was found in neutropenia and its recoveryduring the ICU stay, presence of cardiac dysfunction, the need for an invasive mechanical ventilator,use of inotropic/vasopressor agents, platelet count, aspartate transaminase level, pH, and AcutePhysiology And Chronic Health Evaluation II score. In the multivariate analysis, acute leukemia, needfor invasive mechanical ventilator, use of inotropic/vasopressor agents, and Acute Physiology AndChronic Health Evaluation II scores were independently associated with a worse outcome in patientswith hematologic malignancies admitted to the ICU.Conclusion: Higher mortality in patients with hematologic malignancies admitted to the ICU isassociated with more severe illness, as reflected by higher organ failure scores or respiratory orhemodynamic instability. Mortality is higher in patients with acute leukemia as compared with otherhematologic malignancies.Crown Copyright © 2012 Published by Elsevier Inc. All rights reserved.

⁎ Corresponding author. Tel.: +82 2 2258 6369; fax: +82 2 599 3589.E-mail address: chinkook@catholic.ac.kr (C.K. Rhee).

883-9441/$ – see front matter. Crown Copyright © 2012 Published by Elsevier Inc. All rights reserved.ttp://dx.doi.org/10.1016/j.jcrc.2012.07.014

739.e2 C.D. Yeo et al.

1. Introduction

In recent decades, the prognosis of patients with hema-tologic malignancies has improved substantially, particular-ly because of new, intensive chemotherapeutic regimens,hematopoietic stem cell transplantation (HSCT), and bettersupportive measures [1,2]. Unfortunately, more intensecancer treatment has led to increasing numbers of therapy-associated complications, which may be life threatening andoften necessitate transfer to an intensive care unit (ICU).Although favorable ICU survival rates have been reportedin critically ill patients with cancer in recent decades [3,4],they remain very low in patients with hematologicmalignancies, in whom ICU mortality rates are 30% to82% and in-hospital and long-term mortality rates are evenhigher [5-15].

As a result, the decision to transfer a severely ill patient tothe ICU or continue the ICU care presents an ethical dilemmato hematologists and intensive care specialists. Preferably,these decisions should be based on the prognoses tailored toindividual patients. Although several studies have identifiedprognostic factors aiming to maximize survival and limitunnecessary suffering and costs, few of those studies werepublished in the last 5 years [5,13,14,16]. Moreover, thesurvival in mechanically ventilated patients after HSCT hasshown an improvement with advances in the treatment andoutcome of patients admitted to the ICU [17]. Therefore, it isnecessary to re-evaluate the prognostic factors in these patients.

This study identified prognostic factors associated withmortality in a cohort of critically ill patients withhematologic malignancies admitted to the ICU.

2. Patients and methods

We retrospectively reviewed 227 consecutive patientswith hematologic malignancies who were admitted to thehematology ICU of Seoul St Mary's Hospital (Seoul, Korea)between April 2009 and December 2011. In this hospital,more than 250 HSCTs are performed annually. Thehematology ICU is a 5-bed unit and is equipped forneutropenia precautions and has laminar flow with high-frequency particulate air filtration. Intensive care unitmanagement consists of hemodynamic monitoring andstandard supportive care, with invasive and noninvasivemechanical ventilation using lung protective strategies andrenal replacement therapy. Transfer to the ICU and do-not-resuscitate code are determined by hematologists collabo-rating with intensive care specialists. Do-not-resuscitatedecisions were made when the diseases were far advancedand there was little possibility of survival despite ofaggressive treatment. Patients were identified from a medicalrecord review. The Institutional Review Board of Seoul StMary's Hospital approved the study and waived therequirement for informed consent from the patients studied.

The following variables on the day of the ICU admissionwere analyzed: age, sex, hematologic diagnosis and diseasestatus, prior hematologic treatment, treatment-related com-plications, reason for ICU admission, need for an invasivemechanical ventilator (IMV), use of inotropic/vasopressoragents, laboratory values, ICU mortality, and in-hospitalmortality. Based on these data, we calculated the AcutePhysiology And Chronic Health Evaluation (APACHE) IIscore for each patient on the day of ICU admission.

Disease status was based on the available chart informa-tion and the most recent bone marrow biopsy data. Completeremission was defined as less than 5% of blast cells inmarrow aspirates in patients with leukemia, the disappear-ance of peripheral and deep lymphadenopathy and othermalignant foci in patients with lymphoma, and thedisappearance of monoclonal immunoglobulin in bloodand urine and less than 5% plasma cells in bone marrowaspirates in patients with myeloma. Bronchiolitis obliteranssyndrome was diagnosed clinically according to the NationalInstitutes of Health–sponsored consensus developmentproject for clinical trials on chronic graft-versus-host disease[19]. Neutropeniawas defined as an absolute neutrophil countless than 1000 cells/mm3. Neutropenia recovery during ICUwas defined if the patient was in the ICU for the 7-day periodcentered on the day the neutrophil count rose above 1000cells/mm3. For patients with more than 1 hospital admissionduring the study period, only the first admission was includedin the analysis to ensure the independence of observations.Inotropic/vasopressor agents included dobutamine, norepi-nephrine, or epinephrine at any dose or dopamine at a dosagegreater than 5 μg/kg per minute.

2.1. Statistical analysis

All results are reported as the means ± SEM or fre-quencies (%). We evaluated differences between survivorsand nonsurvivors using the t test for continuous variablesand the χ2 test or Fisher exact test for categorical variables.We used logistic regression analysis to examine therelationship between multiple variables and ICU survival.Goodness of fit was computed to assess the relevance of thelogistic regression model using the Hosmer-Lemeshow test[18]. Odds ratios and their 95% confidence intervals werecomputed. All tests were 2 sided, and P b .05 was consideredstatistically significant. All statistical analyses were per-formed using the SPSS software (ver. 15.0.0 for Windows;SPSS, Inc, Chicago, IL).

3. Results

3.1. Demographic characteristics

Table 1 summarizes the patient data. Of the 227 patientswith hematologic malignancies admitted to our ICU between

Table 1 Characteristics of 227 patients with hematologicmalignancies

n (%) or mean ± SEM

Age 51.2 ± 1.0Male sex 140 (61.7%)Type of hematologic malignancyAML 117 (51.5%)Acute lymphoblastic leukemia 34 (15.0%)Chronic myelogenous leukemia 12 (5.3%)Multiple myeloma 31 (13.7%)Myelodysplastic syndrome 15 (6.6%)Malignant lymphoma 18 (7.9%)Relapsed malignancy 38 (16.7%)Status of hematologic malignancyActive status 178 (78.4%)Complete remission 49 (21.6%)Prior treatmentChemotherapy in last 30 d 131 (57.7%)Previous HSCT 74 (32.6%)Autologous HSCT in last 30 d 6 (2.6%)Allogenic HSCT in last 30 d 11 (4.8%)Overall APACHE II score 19.4 ± 0.5ICU length of stay (d) 10.8 ± 0.8Need for an invasive mechanicalventilation

109 (48.0%)

Use of inotropic agent 124 (54.6%)

739.e3Prognostic factors of the hematologic malignancy in the ICU

April 2009 and December 2011, 140 (61.7%) were male, andthe mean age was 51.18 ± 1.0 years. The most commonhematologic diseases were acute myelogenous leukemia(AML) (51.5%), acute lymphoblastic leukemia (15.0%), andmultiple myeloma (13.7%). Thirty-eight patients (16.7%)had relapsed, and 49 patients (21.6%) were in completeremission with prior treatment. Of the 227 patients, 131(57.7%) had received systemic chemotherapy in the last 30days, and 74 (32.6%) had undergone a prior HSCT. In theprevious 30 days, 6 patients (2.6%) underwent autologousHSCT, and 11 patients (4.8%) underwent allogeneicmyeloablative HSCT. Overall APACHE II score was 19.4 ±0.5. Of the 227 patients, 109 (48.0%) required IMV, and 124patients (54.6%) received inotropic/vasopressor agents.

3.2. Factors affecting the ICU survival

The ICU mortality was 84.1%, and Table 2 shows theresults of the univariate analyses comparing survivors andnonsurvivors. The mortality was significantly higher inpatients with acute leukemia than in those with othermalignancies. Neutropenia was more common in nonsurvi-vors than in survivors. In addition, neutropenia recoveryduring ICU was more common in nonsurvivors than insurvivors. Of the indications for ICU admission, cardiacdysfunction was more common in survivors. Nonsurvivorswere more likely to need IMV and to require inotropicagents. The APACHE II scores were significantly higher innonsurvivors than survivors. Of the laboratory variables, the

platelet count was significantly higher, and aspartate amino-transferase (AST) level and pH were lower in survivors. Nosignificant difference was found between survivors andnonsurvivors in the ICU in age, sex, underlying diseases, ordisease status. However, when the 10 variables found to besignificant in the univariate analyses were introduced into alogistic regression model, 4 were independently associatedwith ICU mortality: acute leukemia, need for IMV, use ofinotropic/vasopressor agents, and APACHE II were asso-ciated with a marked increase in the mortality of patientswith hematologic malignancies admitted in the ICU(Table 3).

3.3. Factors affecting the hospital survival

The hospital mortality rate was 89.9%, and 3-, 6-, and 12-month mortalities were 93.2%, 96.2%, and 99.1%, respec-tively. Among the factors affecting the hospital mortality,acute leukemia, neutropenia and its duration more than 21days, presence of cardiac dysfunction, the need for IMV, useof inotropic agents, AST, alanine aminotransferase, totalbilirubin, platelet count, and APACHE II were significantlydifferent between survivors and nonsurvivors. When thesevariables were introduced into a logistic regression model,total bilirubin and APACHE II were independently associ-ated with the hospital mortality.

4. Discussion

In this cohort study, we found that acute leukemia, IMV,use of inotropic agents, and APACHE II score were inde-pendently associated with a worse outcome. These results areconsistentwith previous studies,which suggested that IMV is astrong predictor of mortality in patients with hematologicmalignancies [5-7,11,12,19]. Other factors thought to be asso-ciated with high mortality when admitted to the ICU includeold age, the presence of hemodynamic instability, neutropenia,number of failing organs, and the presence of graft-versus-host disease after allogeneic HSCT [5,7-12,14-16].

Our cohort showed relatively high mortality comparedwith previous studies [5-12,15], although age and overallAPACHE II score were not significantly different. Theseresults could be partially explained by the fact that theproportion of difficult-to-treat acute leukemia in our serieswas relatively high because lots of patients with poor prog-nostic factors (old age [20-22], relapsed [23], previouslyfailed to treat, or no candidate for HSCT donors [24]) weretransferred to our institution. As a result, our institutionperformed many HSCTs using multinational unrelated orhaplotype-mismatched donors [25,26], and we included a lotof patients with old age or relapsed leukemia in this study.Thus, we postulate that relatively high mortality in theseresults might be attributable to severity of hematologicmalignancies and difficult to treat cases.

Table 2 Comparison between survivors and nonsurvivors in ICU

Survivors (n = 36) Nonsurvivors (n = 191) Odds ratio 95% CI P

Age 51.69 ± 2.68 51.08 ± 1.07 .824Male sex 25 (69.4%) 115 (60.2%) 0.66 0.28-1.52 .296Acute leukemia 16 (44.4%) 135 (70.7%) 3.01 0.72-0.95 .002Active status 30 (83.3%) 148 (77.5%) 0.69 0.24-1.88 .434Relapsed malignancy 8 (22.2%) 30 (15.7%) 0.65 0.25-1.73 .337Treatment-related complicationsNeutropenia on ICU admission 7 (19.4%) 85 (44.5%) 3.32 1.31-8.79 .005Neutropenia duration N21 d 2 (5.6%) 33 (17.3%) 3.55 0.78-22.50 .074Neutropenia recovery during ICU stay 2 (5.6%) 37 (19.4%) 4.08 0.90-25.77 .044Bronchiolitis obliterans syndrome 1 (2.8%) 6 (3.1%) 1.14 0.13-25.80 .908Indication for ICU admissionRespiratory failure 14 (38.9%) 106 (55.5%) 1.14 0.13-25.80 .908Septic shock/severe sepsis 18 (50.0%) 114 (59.7%) 1.48 0.68-3.21 .280Cardiac dysfunction 4 (11.1%) 3 (1.6%) 0.13 0.02-0.72 .002Neurologic disorder 2 (5.6%) 14 (7.3%) 1.35 0.27-8.99 .703Need for mechanical ventilator 5 (13.9%) 104 (54.4%) 7.41 2.60-22.75 .001Use of inotropic agent 8 (22.2%) 116 (60.7%) 5.41 2.21-13.69 .001Laboratory valuesAST (IU/L) 91.42 ± 52.55 255.78 ± 51.51 .027ALT (IU/L) 77.78 ± 34.10 178.54 ± 40.14 .58Creatinine (mg/dL) 2.01 ± 0.31 1.58 ± 0.87 .529Bilirubin (mg/dL) 1.29 ± 0.21 3.42 ± 0.38 .180Platelet count (×109/L) 75.47 ± 10.31 44.84 ± 4.02 .001pH 7.26 ± 0.17 7.37 ± 0.01 .008APACHE II score 13.0 ± 1.1 20.6 ± 0.5 .001

CI indicates confidence interval; ALT, alanine aminotransferase.

739.e4 C.D. Yeo et al.

In some previous studies [5-9,11,12,16,19], the type ofhematologic malignancy does not predict mortality. On theother hand, 2 reports [27,28] suggested an associationbetween underlying AML and high mortality. However,these reports examined less than 100 patients with acuteleukemia, and the data were collected more than 10 years ago.Our data showed that acute leukemia was independentlyassociated with a worse prognosis. According to epidemio-logical data for the most recent 25-year period, the incidenceof AML has been increasing with age, and the survival rate isvery low in elderly patients, although the overall 5-year

Table 3 Multivariate analysis of patient characteristics associated wi

Odds

Acute leukemia 3.40Neutropenia on ICU admission 1.13Neutropenia recovery during ICU stay 1.90Cardiac dysfunction 0.87Need for an invasive mechanical ventilator 4.75Use of inotropic agents 3.83AST (IU/L) 1.01Platelet count (×109/L) 1.00pH 2.09APACHE II score 1.12

Goodness of fit (Hosmer-Lemeshow) χ2 P = .961.

survival has been increasing [29]. The poor prognosis inpatients with AML may be partially explained by populationaging despite advances in management of acute leukemia[30,31]. A second explanation for our finding is the aggre-ssiveness of AML compared with other malignancies, such aschronic lymphocytic leukemia [32,33].

Intensive chemotherapy and subsequent HSCT com-monly cause neutropenia, and its recovery in hematologicmalignancies might be prolonged compared with that in solidtumors due to malignant infiltration of the bone marrow [16].Moreover, patients with hematologic malignancies have

th ICU mortality

ratio 95% CI P

1.25-9.26 .0170.34-3.79 .8410.32-11.40 .4830.14-5.26 .8801.40-16.12 .0121.40-10.50 .0090.99-1.00 .3220.99-1.01 .2020.14-1.62 .2360.83-0.95 .001

739.e5Prognostic factors of the hematologic malignancy in the ICU

higher mortality, more immunosuppression, and morevulnerability to sepsis when neutropenia occurs [34]. As aresult, patients with acute respiratory distress syndrome whohave recovered from neutropenia have showed poor out-comes, and neutropenia is suggested to be associated withhospital mortality [12,16,35,36]. However, several studiessuggest that neutropenia did not affect outcome and our datafailed to show that neutropenia was an independent predictorof a poor prognosis [5,6]. These conflicting findings warrantfurther larger studies to confirm this relationship.

Our study revealed that patients requiring IMV andinotropic/vasopressor had worse prognosis. Obviously, pa-tients with respiratory distress and hemodynamic instabilityhave high organ failure scores. Our data highlight the impor-tance of respiratory and hemodynamic status in predictingoutcome in patient with hematologic malignancies [5,7].

Although cardiac dysfunction was not an independentpredictor of outcome in this study, patients with cardiacdysfunction had a lower risk of death compared with othercauses of ICU admission, such as respiratory failure orsevere sepsis. Another study supports this finding; patientswith an established diagnosis of cardiac dysfunction hadbetter outcomes than patients with normal cardiac function[37]. This can be explained by the difference in mortalitybetween cardiogenic pulmonary edema and acute lung injuryof infectious etiology. However, only 6 patients in our serieshad cardiac dysfunction, the diagnosis of which includedcardiogenic pulmonary edema, arrhythmia, and cardiac tam-ponade. Further study of patients with hematologic malig-nancies who have cardiac dysfunction is needed.

Although the different pH between survivors and non-survivors may be interesting to report, we could not suggestthat the presence of respiratory alkalosis might affect thisdifference. We had investigated the parameters of arterialblood gas analysis; however, the levels of CO2 and HCO3

were not significantly different between survivors andnonsurvivors. These findings suggest that further studyregarding acid-base balance in critically ill patients withhematologic malignancies is needed.

The intent of the APACHE II score and similar scoringsystems is to predict the outcome of critically ill patients [38].For patients with malignancies transferred to the ICU, theAPACHE II system could be a good basis for evaluating theseverity of organ dysfunction [39]. Our data were consistentwith prior results, suggesting that the APACHE II is anindependent predictor of outcome and survival in critically illpatients with hematologic malignancies [7,14,36]. However,scoring systems such as APACHE II, the Simplified AcutePhysiology Score II, and Sequential Organ Failure Assess-ment do not perform well in this specific group of patients[38,40]; they predicted relatively high mortality rates insurvivors and low rates in nonsurvivors [13,38]. Therefore,the APACHE II score alone should not be used for individualdecision making [7]. Additional factors for predictingmortality in patients with hematologic malignancy in theICU also should be considered.

Although APACHE-II score gives a risk of in-hospitalmortality rather than ICU mortality, we used ICU mortalityas an end point. Patients who received intensive care andsurvived in ICU, then failed to get complete remission, andfinally died would be considered as dead, when we used in-hospital mortality as an end point. Intensive chemotherapy toget complete remission and followed HSCT frequently resultin life-threatening complications. Being survived withadequate intensive care in ICU could give a chance to becured to patients with hematologic malignancies. Therefore,we wanted to focus on factors, which could affect survival inICU by receiving intensive care. Moreover, many previousstudies also used ICU mortality as an end point [5-9].

Our study has several limitations. First, it was aretrospective observational study. However, using a singlecohort whose treatment was based on the same protocol, wecarefully evaluated all patients admitted to the ICU who wereenrolled in the study. The aim of this study was to identifyfactors predicting the ICU outcome, so the setting of ourstudy may not have differed greatly from that of a prospectiveobservational study. Second, our study was conducted in asingle institution, which limits the generalizability of ourfindings to other patient populations. However, Seoul StMary's Hospital is one of the largest HSCT centers in Asia.To our knowledge, our series included the greatest numberof patients enrolled per year in any study to date, so our datamay be representative of this disease group.

In conclusion, increased mortality in patients with hema-tologic malignancies admitted to the ICU is associated withmore severe illness, as reflected in higher organ failure scoresor respiratory or hemodynamic instability. Moreover,mortality was higher in patients with acute leukemia com-pared with other hematologic malignancies. Further studyregarding the prognostic factors in patients with hematologicmalignancies admitted to the ICU is needed.

References

[1] Brenner H. Long-term survival rates of cancer patients achieved bythe end of the 20th century: a period analysis. Lancet 2002;360(9340):1131-5.

[2] Brenner H, Gondos A, Arndt V. Recent major progress in long-termcancer patient survival disclosed by modeled period analysis. J ClinOncol 2007;25(22):3274-80.

[3] Sculier JP, Markiewicz E. Medical cancer patients and intensive care.Anticancer Res 1991;11(6):2171-4.

[4] Staudinger T, et al. Outcome and prognostic factors in critically illcancer patients admitted to the intensive care unit. Crit Care Med2000;28(5):1322-8.

[5] Geerse DA, et al. Prognosis of patients with haematologicalmalignancies admitted to the intensive care unit: Sequential OrganFailure Assessment (SOFA) trend is a powerful predictor of mortality.Eur J Intern Med 2011;22(1):57-61.

[6] Cherif H, et al. Predictors of short and long-term outcome in patientswith hematological disorders admitted to the intensive care unit for alife-threatening complication. Support Care Cancer 2007;15(12):1393-8.

739.e6 C.D. Yeo et al.

[7] Ferra C, et al. Outcome and prognostic factors in patients withhematologic malignancies admitted to the intensive care unit: a single-center experience. Int J Hematol 2007;85(3):195-202.

[8] Park MR, et al. Outcomes in critically ill patients with hematologicmalignancies who received renal replacement therapy for acute kidneyinjury in an intensive care unit. J Crit Care 2011;26(1):107.e1-6.

[9] Yang TM, et al. Outcome of hematopoietic stem cell recipients whowere mechanically ventilated and admitted to intensive care units. JFormos Med Assoc 2007;106(4):295-301.

[10] Lamia B, et al. Changes in severity and organ failure scores asprognostic factors in onco-hematological malignancy patients admit-ted to the ICU. Intensive Care Med 2006;32(10):1560-8.

[11] Kroschinsky F, et al. Outcome and prognostic features of intensivecare unit treatment in patients with hematological malignancies.Intensive Care Med 2002;28(9):1294-300.

[12] Benoit DD, et al. Outcome and early prognostic indicators inpatients with a hematologic malignancy admitted to the intensivecare unit for a life-threatening complication. Crit Care Med 2003;31(1):104-12.

[13] Nishida K, Palalay MP. Prognostic factors and utility of scoringsystems in patients with hematological malignancies admitted to theintensive care unit and required a mechanical ventilator. Hawaii Med J2008;67(10):264-9.

[14] Hampshire PA, et al. Admission factors associated with hospitalmortality in patients with haematological malignancy admitted to UKadult, general critical care units: a secondary analysis of the ICNARCCase Mix Programme Database. Crit Care 2009;13(4):R137.

[15] Bird GT, et al. Outcomes and prognostic factors in patients withhaematological malignancy admitted to a specialist cancer intensivecare unit: a 5 yr study. Br J Anaesth 2012;108(3):452-9.

[16] Rhee CK, et al. Risk factors for acute respiratory distress syndromeduring neutropenia recovery in patients with hematologic malignan-cies. Crit Care 2009;13(6):R173.

[17] Rubenfeld GD, Crawford SW. Withdrawing life support frommechanically ventilated recipients of bone marrow transplants: acase for evidence-based guidelines. Ann Intern Med 1996;125(8):625-33.

[18] Hosmer DW, Taber S, Lemeshow S. The importance of assessing thefit of logistic regression models: a case study. Am J Public Health1991;81(12):1630-5.

[19] Vandijck DM, et al. Impact of organ dysfunction on mortality in ICUpatients with hematologic malignancies. Intensive Care Med2010;36(10):1744-50.

[20] Eom KS, et al. Hematopoietic stem cell transplant following remissioninduction chemotherapy including gemtuzumab ozogamicin is afeasible and effective treatment option in elderly patients with acutemyeloid leukemia. Leuk Lymphoma 2011;52(12):2321-8.

[21] Lee SE, et al. Favorable outcomes of intravenous busulfan,fludarabine, and 400 cGy total body irradiation-based reduced-intensity conditioning allogeneic stem cell transplantation for acutemyelogenous leukemia with old age and/or co-morbidities. Int JHematol 2010;92(2):342-50.

[22] Eom KS, et al. Gemtuzumab ozogamicin in combination withattenuated doses of standard induction chemotherapy can successfullyinduce complete remission without increasing toxicity in patients with

acute myeloid leukemia aged 55 or older. Eur J Haematol 2007;79(5):398-404.

[23] Eom KS, et al. FLANG salvage chemotherapy is an effective regimenthat offers a safe bridge to transplantation for patients with relapsed orrefractory acute myeloid leukemia. Med Oncol 2011;28(Suppl. 1):S462-70.

[24] Kim HJ, et al. Successful prevention of acute graft-versus-host diseaseusing low-dose antithymocyte globulin after mismatched, unrelated,hematopoietic stem cell transplantation for acute myelogenousleukemia. Biol Blood Marrow Transplant 2009;15(6):704-17.

[25] Kim HJ, et al. Hematopoietic stem cell transplatation with usingmultinational unrelated donors for acute myeloneous leukemia.Korean J Hematol 2007;42(2):98-105.

[26] Kim HJ, et al. Haplotype mismatch transplantation using high-doseCD34+ cells with stratified new conditioning regimens in patients withacute myeloid leukemia. Korean J Hematol 2003;38(4):221-7.

[27] Bahammam AS, et al. Outcome of patients with hematologicalmalignancies admitted to the intensive care unit with life-threateningcomplications. Saudi Med J 2005;26(2):246-50.

[28] Depuydt PO, et al. Outcome in noninvasively and invasively ventilatedhematologic patients with acute respiratory failure. Chest 2004;126(4):1299-306.

[29] Maynadie M, et al. Twenty-five years of epidemiological recording onmyeloid malignancies: data from the specialized registry of hemato-logic malignancies of Cote d'Or (Burgundy, France). Haematologica2011;96(1):55-61.

[30] Ribera JM. Advances in acute lymphoblastic leukemia in adults. CurrOpin Oncol 2011;23(6):692-9.

[31] Estey EH. Acute myeloid leukemia: 2012 update on diagnosis, riskstratification, and management. Am J Hematol 2012;87(1):89-99.

[32] Peters SG, Meadows 3rd JA, Gracey DR. Outcome of respiratoryfailure in hematologic malignancy. Chest 1988;94(1):99-102.

[33] Massion PB, et al. Prognosis of hematologic malignancies does notpredict intensive care unit mortality. Crit Care Med 2002;30(10):2260-70.

[34] Taccone FS, et al. Characteristics and outcomes of cancer patients inEuropean ICUs. Crit Care 2009;13(1):R15.

[35] Azoulay E, et al. Deterioration of previous acute lung injury duringneutropenia recovery. Crit Care Med 2002;30(4):781-6.

[36] Owczuk R, et al. Patients with haematological malignancies requiringinvasive mechanical ventilation: differences between survivors andnon-survivors in intensive care unit. Support Care Cancer 2005;13(5):332-8.

[37] Villa F, et al. Etiology of respiratory failure is related to mortality incritically ill patients affected by a hematological malignancy: aretrospective study. Minerva Anestesiol 2010;76(1):7-12.

[38] Sculier JP, et al. Scoring systems in cancer patients admitted for anacute complication in a medical intensive care unit. Crit Care Med2000;28(8):2786-92.

[39] Groeger JS, et al. Multicenter outcome study of cancer patientsadmitted to the intensive care unit: a probability of mortality model.J Clin Oncol 1998;16(2):761-70.

[40] Lemeshow S, Klar J, Teres D. Outcome prediction for individualintensive care patients: useful, misused, or abused? Intensive Care Med1995;21(9):770-6.