Outcomes of cancer and non-cancer patients with acute kidney injury and need of renal replacement therapy admitted to general intensive care units

Transcription

1 NDT Advance Access published July 28, 2010 Nephrol Dial Transplant (2010) 1 of 7 doi: /ndt/gfq441 Original Article Outcomes of cancer and non-cancer patients with acute kidney injury and need of renal replacement therapy admitted to general intensive care units Elizabeth Maccariello 1,2,3, Carla Valente 1,2, Lina Nogueira 1,2, Helio Bonomo Jr. 1,2, Marcia Ismael 1,2, Jose Eduardo Machado 1,2, Fernanda Baldotto 1,2, Marise Godinho 1,2, Eduardo Rocha 1,2,4 and Marcio Soares 1,5 1 NepHro Consultoria em Doenças Renais, Rio de Janeiro, Brazil, 2 Rede D Or de Hospitais, Rio de Janeiro, Brazil, 3 Department of Nephrology, Hospital Universitário Antônio Pedro, Universidade Federal Fluminense, Rio de Janeiro, Brazil, 4 Department of Nephrology, Faculdade de Medicina, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil and 5 Intensive Care Unit, Instituto Nacional de Câncer, Rio de Janeiro, Brazil Correspondence and offprint requests to: Elizabeth Maccariello; emaccariello@yahoo.com.br Abstract Background. Studies on cancer patients with acute kidney injury (AKI) are restricted to specialized intensive care units (ICUs). The aim of this study was to compare the characteristics and outcomes of cancer and non-cancer patients requiring renal replacement therapy (RRT) for AKI in general ICUs. Methods. A prospective cohort study was conducted in 14 ICUs from three tertiary care hospitals. A total of 773 (non-cancer 85%; cancer 15%) consecutive patients were included over a 44-month period. Logistic regression was used to identify factors associated with hospital mortality. Results. Continuous RRT was used in 79% patients. The main contributing factors for AKI were sepsis (72%) and ischaemia/shock (66%); AKI was multifactorial in 87% of cancer and in 71% non-cancer patients. Hospital mortality rates were higher in cancer (78%) than in non-cancer patients (68%) (P = 0.042). However, in multivariate analyses, older age, medical admission, poor chronic health status, comorbidities, ICU days until the RRT start and number of associated organ dysfunctions were associated with hospital mortality. The diagnosis of cancer was not independently associated with mortality [odds ratio = 1.54 (95% confidence interval, ), P = 0.115]. Mortality in cancer patients was mostly dependent on the number of associated organ dysfunctions. Of note, 85% cancer patients recovered renal function at hospital discharge. Conclusions. In general ICUs, one in six patients requiring RRT has cancer. Despite a relatively higher mortality, the presence of cancer was not independently associated with mortality in the present cohort. Keywords: acute kidney injury; cancer; intensive care unit; outcome; renal replacement therapy Introduction Up to half of cancer patients experience acute kidney injury (AKI) and the majority of them require renal replacement therapy (RRT) while in the intensive care unit (ICU) [1 7]. AKI in these patients occurs either as a consequence of the cancer itself (urinary tract obstruction, acute tumour lysis syndrome), anticancer treatments (drug-induced nephropathy, major surgical procedures) or diverse-associated severe clinical conditions (sepsis, hypovolaemia) [1,8]. Although survival seems to be improving over the last decade, the development of AKI in critically ill cancer patients remains associated with high mortality rates [2,3,7, 9 13]. In addition, the diagnosis of a malignancy per se seems to be no longer independently associated with a higher risk of death [2,3]. However, the few studies on this specific subgroup of ICU cancer patients were performed in specialized ICUs, thereby imposing restrictions to the generalization of results to general ICUs [2,3,7,9 11, 14]. Therefore, we performed the present multicentre study with three major aims: (i) to evaluate and compare the characteristics and outcomes of cancer and non-cancer ICU patients with AKI requiring RRT, (ii) to determine the impact of cancer diagnosis on hospital mortality in critically ill AKI patients and (iii) to compare outcome predictors between the two groups of patients. Materials and methods Design and setting This was a prospective observational study conducted from December 2004 to July 2008 at 14 ICUs of three tertiary care hospitals in Rio de Janeiro, Brazil. The ICUs included four medical surgical ICUs, three The Author Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 2 E. Maccariello et al. Table 1. Main patient characteristics and outcomes a All patients (n = 773) Cancer patients (n = 118, 15%) Non-cancer patients (n = 655, 85%) P-value b Age (years) 70.5 ± ± ± Male gender 418 (54%) 66 (56%) 352 (54%) Hospital stay before ICU admission (days) 0 (0 1) 0 (0 4) 0 (0 0) <0.001 Medical admission 630 (82%) 75 (65%) 553 (84%) <0.001 Body mass index (kg/m 2 ) 26.3 ± ± ± Poor chronic health status (Knaus C or D) 307 (40%) 38 (32%) 269 (41%) Chronic renal injury 300 (39%) 26 (22%) 274 (42%) <0.001 Charlson index (points) 3 (1 5) 5 (2 7) 3 (1 5) <0.001 Charlson index (except cancer-related points) 3 (1 4) 2 (0 3) 3 (1 5) <0.001 Charlson index 1 point (except cancer-related points) 614 (79%) 83 (70%) 531 (81%) Type of cancer Locoregional solid tumour 59 (8%) 59 (50%) NA Metastatic solid tumour 27 (4%) 27 (23%) Haematological malignancy 32 (4%) 32 (27%) SAPS II (points) 46.5 ± ± ± SAPS II (except cancer-related points) 45.8 ± ± ± SOFA at start of RRT (points) 8 (5 10) 9 (7 11) 8 (5 10) SOFA at start of RRT (except renal points) 6 (3 8) 7 (5 9) 6 (3 8) <0.001 Associated organ dysfunctions total (n) 2 (1 2) 2 (1 3) 2 (1 2) <0.001 Individual organ dysfunctions Respiratory 432 (56%) 71 (60%) 361 (55%) Cardiovascular 524 (68%) 89 (75%) 435 (66%) Haematological 174 (23%) 47 (40%) 127 (19%) <0.001 Hepatic 108 (14%) 36 (31%) 72 (11%) <0.001 Neurological 70 (9%) 3 (3%) 67 (10%) Sepsis 560 (72%) 92 (78%) 468 (72%) Mechanical ventilation at start of RRT 587 (76%) 96 (81%) 491 (75%) Vasopressors at start of RRT 532 (69%) 91 (77%) 441 (67%) Outcome data ICU LOS (days) 19 (9 38) 17 (6 35) 19 (9 39) Hospital LOS (days) 28 (13 54) 24 (12 52) 28 (13 54) End-of-life decisions 168 (22%) 32 (28%) 136 (21%) ICU mortality 499 (65%) 83 (70%) 416 (64%) Hospital mortality 538 (70%) 92 (78%) 446 (68%) ICU = intensive care unit; SAPS = Simplified Acute Physiology Score; SOFA = Sequential Organ Failure Assessment; NA = not applicable; RRT = renal replacement therapy; LOS = length of stay. a Results expressed as mean ± standard deviation, median (interquartile range), n (%). b Comparisons between cancer and non-cancer patients. coronary/cardiac surgery units, two neurological, one respiratory care, two medical and two surgical ICUs. The ethics committees at each site approved the study and the need for informed consent was waived. The study was absolutely observational and all decisions related to patients care [including end-of-life (EOL) decisions] were at discretion of the medical team responsible for the patients. EOL decisions (to withhold or withdraw life-sustaining therapies) were generally taken in patients who did not recover from the acute illness despite full ICU care. Selection of participants, data collection and definitions Patients were retrieved from a prospective cohort of patients who received RRT during the ICU stay at the participating hospitals during the study period. Both patients with AKI and acute on chronic kidney injury were evaluated. Patients with chronic kidney injury had a previous glomerular filtration rate <60 ml/min/per 1.73 m 2 for at least 3 months [15]. Patients with non-renal indications for RRT, end-stage renal disease requiring chronic dialysis and those with ICU stay <24 h and readmissions were not considered. Patients with cancer must have had a pathologically proven diagnosis of malignancy. For the purpose of the present study, cancer patients were classified into three major groups: haematological malignancies, solid tumour with locoregional involvement and solid tumour with distant metastasis. Demographic, clinical and laboratory data were prospectively collected including hospital location before ICU admission, main diagnosis for ICU admission, comorbidities, previous chronic health status (Knaus scale: A: no limitation of activity, B: moderate limitation, C: severe limitation and D: bedridden or institutionalized) [16], pre-morbid renal function and the Simplified Acute Physiology Score (SAPS) II at ICU admission [17], and contributing factors for AKI were also recorded. The measurement of comorbidities was performed using the Charlson comorbidity index [18]. The Sequential Organ Failure Assessment (SOFA) score [19], the need for mechanical ventilation (MV) and vasopressors for >24 h on the first day of RRT were recorded and considered in the analysis. Individual organ failures were defined as a SOFA score 2 points in each domain [20]. Multiple organ failure was defined as the presence of 2 associated organ failures. Patients were classified based on the reason for ICU admission in medical, scheduled surgical and emergency surgical. Infection was defined as the presence of a pathogenic microorganism in a sterile milieu (such as blood or cerebrospinal fluid) and/or clinically suspected infection that justified the administration of antibiotics [20]. Sepsis was diagnosed according to the current definitions [21]. Vital status at hospital discharge was the outcome of interest. AKI was classified using the RIFLE [R (risk); I (injury); F (failure); L (loss) E (end-stage renal disease)] criteria [22] at the time of initiation of RRT [15]. Urine output <400 ml/day was considered as oliguria. Decisions to start, change the method and cease RRTwere taken together by the nephrologist and the intensivist responsible for the patient on an individual basis. The decision to start RRT was based on the presence of one or more of the following characteristics: metabolic acidosis, urea above 100 mg/ dl, hyperkalaemia above 6.0 meq/l, evidence of Volume overload, severe hyperphosphataemia or dysnatraemia, oligoanuria and uraemic signs or symptoms [15]. Prescribed RRT modes were daily conventional dialysis, daily extended dialysis and continuous RRT (CRRT) taking into consideration the patient s haemodynamic status, being CRRT employed in patients receiving vasoactive drugs and in those with potential for haemo-

3 Outcomes of cancer and non-cancer patients with AKI 3 Table 2. Clinical and laboratory data related to kidney function a All patients (n = 773) Cancer patients (n = 118, 15%) Non-cancer patients (n = 655, 85%) P-value b ICU days until the start of dialysis 3 (1 10) 2 (1 8) 3 (1 10) Start of dialysis on the 1st day of ICU 302 (39%) 41 (35%) 261 (40%) Initial method of dialysis Conventional 131 (17%) 10 (8%) 121 (18%) Extended 30 (4%) 3 (3%) 27 (4%) Continuous 612 (79%) 105 (89%) 507 (77%) Indications for RRT Azotaemia 439 (57%) 59 (50%) 380 (58%) Volume overload 428 (55%) 68 (58%) 360 (55%) Metabolic acidosis 390 (51%) 77 (65%) 313 (48%) Oliguria 209 (27%) 34 (29%) 175 (27%) Dysnatraemia 82 (11%) 8 (7%) 74 (11%) Hyperkalaemia 53 (7%) 10 (8%) 43 (7%) RIFLE class before the start of RRT Risk 188 (24%) 29 (25%) 159 (24%) Injury 157 (20%) 28 (24%) 129 (20%) Failure 428 (55%) 61 (52%) 36 (56%) Contributing factors for AKI Sepsis 560 (72%) 92 (78%) 468 (72%) Ischaemia/shock 508 (66%) 87 (74%) 421 (64%) Radiocontrast/nephrotoxins 182 (24%) 38 (32%) 144 (22%) Haemolysis/rhabdomyolysis 43 (6%) 3 (3%) 40 (6%) Urinary tract obstruction (cancer related) 21 (3%) 9 (7%) 12 (2%) Multiple myeloma 7 (1%) 7 (6%) 0 <0.001 More than one contributing factor for AKI 565 (73%) 103 (87%) 462 (71%) <0.001 Laboratory data at the start of RRT Creatinine (mg/dl) 1.9 ( ) 1.7 ( ) 2.0 ( ) Urea (mg/dl) 116 (66 169) 100 (55 152) 118 (69 173) Sodium ± ± ± Potassium 4.5 ± ± ± Phosphorus 4.7 ± ± ± ph 7.31 ( ) 7.28 ( ) 7.32 ( ) HCO 3 concentration (meq/l) 18.0 ( ) 16.0 ( ) 18.5 ( ) <0.001 Lactate (mmol/l) 1.8 ( ) 2.1 ( ) 1.7 ( ) ICU = intensive care unit; AKI = acute kidney injury; RRT = renal replacement therapy. a Results expressed as mean ± standard deviation, median (interquartile range), n (%). b Comparisons between cancer and non-cancer patients. dynamic instability [15]. FAD 100 (Braun, Rio de Janeiro, Brazil) and Prisma or AK series (Gambro ) machines, and polisulfone (F series, Fresenius ) and AN69S (Gambro ) membranes were used in the RRT procedures. Customized dialysis and replacement solutions were prescribed on a daily basis aiming to optimize patient s metabolic control and were produced at the hospital pharmacy as appropriate. Data processing and statistical analysis Data were entered in a computer database by a single data manager. Data consistency was assessed by a single author (M.S.) through a rechecking procedure of a 10% random sample of patients. All entered data were also checked for outlying and implausible values. Continuous variables were presented as mean ± standard deviation or median (25 75% interquartile range). Univariate and multivariate logistic regression analyses were used to identify factors associated with hospital mortality [23]. Linearity between each continuous variable and the dependent variable was demonstrated using locally weighted scatterplot smoothing [23]. In case of nonlinearity, the variable was stratified according to the inflection points and clinical significance. For categorical variables with multiple levels, the reference level was attributed to the one with the lowest probability of the dependent variable. yielding P-values <0.2 by univariate analysis and those considered clinically relevant were entered in the multivariate analysis to estimate the independent association of each covariate with the dependent variable. SAPS II was not entered in the multivariate analyses because it encompasses other covariates such as age, variables used to define organ failures, severe comorbidities and underlying malignancies [17]. Results were summarized as odds ratios (OR) and respective 95% confidence intervals (CI). Possible interactions were tested. The area under the receiver operating characteristic curve (AROC) was used to assess the models discrimination; an AROC of 1.0 denotes perfect, while a value close to 0.50 indicates no apparent accuracy [24]. The Hosmer Lemeshow goodness-of-fit test was used to evaluate agreement between the observed and expected results across all strata of probabilities of the outcome of interest (calibration) [23]. With this test, P-values >0.05 indicate a good fit for the model. Two-tailed P-values <0.05 were considered statistically significant. Statistical analyses were carried out with SPSS for Windows, version 13.0 (SPSS Inc., Chicago, IL). Results Characterization of the study population A total of 773 patients with a mean age of 70.5 ± 16.0 years were included over the study period, and out of them, 118 (15%) had a proven diagnosis of a malignancy. From these patients, 86 (73%) had solid tumours and 32 (27%) had haematological malignancies. The most frequent type of malignancies were lower gastrointestinal (n = 25, 21%), urogenital (n = 18, 15%), liver and billiary tract (n = 15, 13%), upper gastrointestinal (n = 11, 9%), leukaemias (n = 10, 8%), multiple myeloma (n = 7, 6%), non- Hodgkin s lymphoma (n = 6, 6%), lung (n =6,5%)and

4 4 E. Maccariello et al. Table 3. Univariate analysis of factors associated with hospital mortality in all patients (n = 773) a Survivors (n = 235, 30%) Non-survivors (n = 538, 70%) Odds ratios (95% CI) P-value Age (years) 63.4 ± ± ( ) <0.001 Male gender 137 (58%) 281 (52%) 0.78 ( ) Hospital stay before ICU admission (days) b 0(0 1) 0 (0 1) 1.08 ( ) Medical admission 189 (80%) 441 (82%) 1.11 ( ) Poor chronic health status (Knaus C or D) 44 (19%) 263 (49%) 4.15 ( ) <0.001 Body mass index (kg/m 2 ) 26.7 ± ± ( ) Charlson index (except cancer-related points) 2 (0 4) 3 (1 4) 1.11 ( ) Charlson index 1 point (except cancer-related points) 164 (70%) 450 (84%) 2.21 ( ) <0.001 SAPS II (except cancer-related points) 39.7 ± ± ( ) <0.001 SOFA score (except renal points) c 4(1 7) 7 (4 8) 1.25 ( ) <0.001 Individual organ dysfunctions c Respiratory 109 (46%) 323 (60%) 1.73 ( ) Cardiovascular 113 (48%) 411 (76%) 3.49 ( ) <0.001 Haematological 32 (17%) 142 (26%) 2.28 ( ) <0.001 Hepatic 18 (8%) 90 (17%) 2.42 ( ) Neurological 15 (6%) 55 (10%) 1.67 ( ) Associated organ dysfunctions total (n) c 1(0 2) 2 (1 3) 1.88 ( ) <0.001 Cancer No 209 (89%) 443 (83%) Yes 26 (11%) 92 (17%) 1.66 ( ) Type of cancer None 209 (89%) 446 (82%) Locoregional solid tumour 16 (7%) 43 (8%) 1.26 ( ) Metastatic solid tumour 7 (3%) 20 (3%) 1.34 ( ) Haematological malignancy 3 (1%) 29 (4%) 4.50 ( ) Mechanical ventilation c 138 (59%) 449 (84%) 3.55 ( ) <0.001 Vasopressors c 117 (50%) 415 (77%) 3.40 ( ) <0.001 ICU days until the start of dialysis (days) c 1(0 5) 5 (1 13) 1.54 ( ) <0.001 Sepsis 136 (58%) 424 (79%) 2.71 ( ) <0.001 Start of dialysis on the 1st day of ICU 127 (54%) 175 (33%) 0.41 ( ) <0.001 Chronic renal injury 105 (44%) 194 (36%) 0.70 ( ) Oliguria 50 (21%) 159 (29%) 1.55 ( ) RIFLE class before the start of RRT Risk 59 (25%) 130 (24%) Injury 42 (18%) 116 (21%) 1.25 ( ) Failure 134 (57%) 298 (55%) 1.00 ( ) More than one contributing factor for AKI 142 (60%) 423 (79%) 2.41 ( ) <0.001 Lactate (mmol/l) c 1.6 ( ) 2.1 ( ) 1.13 ( ) SD = standard deviation; IQR = interquartile range; CI = confidence interval; ICU = intensive care unit; AKI = acute kidney injury; SAPS = Simplified Acute Physiology Score; SOFA = Sequential Organ Failure Assessment; RRT = renal replacement therapy. a Results expressed as mean ± SD, median (IQR), n (%). b Ln days was used to estimate the OR. c At the start of renal replacement therapy. others (n = 20, 17%). The main patient characteristics are depicted in Table 1. Patients were usually admitted to the ICU because of a medical complication. Six hundred and fourteen (79%) patients had comorbidities other than cancer and the most frequent were systemic arterial hypertension (71%), diabetes (32%), coronary artery disease (28%), chronic heart failure (22%), chronic pulmonary disease (14%) and cirrhosis (6%). Excluding cancer-related points, cancer and non-cancer patients were similar in terms of severity of illness evaluated by the SAPS II score. However, the severity of organ dysfunctions was higher in patients with cancer. Clinical and laboratory data related to kidney function are shown in Table 2. RRT was implemented at 3 (1 10) days of ICU stay and most of the patients were classified as failure (55%) according to RIFLE immediately before the start of RRT. CRRT was the preferential initial modality of RRT (79%) and was more frequently used in cancer patients (89 vs 77%, P = 0.017). Azotaemia, evidence of Volume overload and metabolic acidosis were the main indications to start RRT in both groups. However, metabolic acidosis was more frequent in patients with cancer (65 vs 48%, P = 0.016). The main contributing factors for AKI were sepsis (72%), ischaemia/shock (66%) and contrast/nephrotoxins (24%). AKI was usually multifactorial (>1 contributing factor for AKI) (73%), particularly in patients with cancer (87 vs 71%, P < 0.001). All urinary tract obstructions observed in cancer patients (n = 7) were related to the malignancy. No patient experienced acute tumour lysis syndrome. Outcome analysis The main patients outcome data were generally similar in cancer and non-cancer patients, except for a higher hos-

5 Outcomes of cancer and non-cancer patients with AKI 5 Table 4. Multivariate analysis of predictors of hospital mortality in all patients (n = 773) Coefficient Odds ratio (95% CI) P-value Age (years) ( ) <0.001 Poor chronic health status (Knaus C or D) No 1.00 Yes ( ) <0.001 Charlson index (except cancer-related points) ( ) Associated organ ( ) <0.001 dysfunctions total (n) a ICU days until the start of ( ) <0.001 dialysis (Ln days) Oliguria No 1.00 Yes ( ) Constant Area under receiver operating characteristic curve = 0.82 (95% CI, ); Hosmer Lemeshow goodness-of-fit (χ 2 = 3.916; P = 0.865). ICU = intensive care unit; CI = confidence interval. a At the start of renal replacement therapy. pital mortality observed in patients with cancer (Table 1). Of note, no differences among the frequencies of EOL decisions were observed. Twenty-six (22%) cancer patients and 209 (32%) non-cancer patients were discharged alive from the hospital. Out of them, 4 (15%) and 66 (32%) were still on RRT, and 22 (85%) and 143 (68%) were no longer requiring the continuing of RRT at hospital discharge (P = 0.140), respectively. The results of univariate analyses of factors associated with increased hospital mortality are depicted in Table 3. Studying all 773 patients, non-survivors were older than survivors (73.6 ± 13.8 vs 63.4 ± 18.2 years, P < 0.001). As expected, non-survivors had higher SAPS II and SOFA (except renal domain) score points and required MV and vasopressors more frequently. Older age, gender, poor chronic health status, hospital days prior to ICU admission, presence of comorbidities, the number of associated organ dysfunctions, the diagnosis of cancer, need for MV, ICU days until the start of RRT, previous chronic kidney injury, oliguria, RIFLE classification, sepsis, multifactorial AKI and lactate levels were entered in a multivariate logistic regression analysis having the hospital mortality as the dependent variable. The final model is presented in Table 4 and had both good discrimination and calibration. Adjusting for other covariates, the diagnosis of cancer was not independently associated with increased hospital mortality [OR = 1.54 (95% CI, ), P = 0.115]. Finally, differences between outcome predictors in noncancer and cancer patients were explored and two models were fitted (Table 5). Univariate analyses are reported in Tables 1 and 2 of the online supplementary material. Hospital mortality in non-cancer patients was dependent on older age, poor chronic health status, presence of comorbidities other than cancer, number of failing organs, need for MV and time to start of RRT in the ICU. In patients with cancer, adjusting for the type of ICU admission and the time until the start of RRT, mortality was mostly dependent on the number of associated organ dysfunctions (Table 5). Discussion In this large prospective multicentre study, patients with cancer accounted for 15% of all patients with AKI requiring RRT in the participating ICUs. The few studies comparing characteristics and outcomes between critically ill cancer and non-cancer patients presenting with this severe complication were conducted in specialized ICUs in referral cancer centres [2,3]. In those studies, most of the patients had haematological malignancies [2,3]. The present study was carried out in general ICUs and, as expected, patients more frequently had solid tumours. Such findings are in accordance with those reported in two recent multicentre studies on patients with cancer requiring ICU admission [20,25]. AKI in patients with cancer was more frequently multifactorial and developed usually in the context of multiple organ failure than non-cancer patients. Moreover, CRRT was more frequently used in patients with cancer as these patients presented with higher severity of organ dysfunctions and required more often using vasopressors than non-cancer patients. Despite relatively higher hospital mortality, ICU mortality rates and duration of hospitalization were comparable between cancer and non-cancer patients. However, mortality rates reported for both non-cancer and cancer patients are in agreement to current literature [2,3,7,11,15,25 29]. More importantly, adjusting for other prognostic factors, the diagnosis of cancer per se was not independently associated with a worse outcome. The higher mortality rate in patients with cancer can be attributed to a more pronounced severity of associated organ dysfunctions upon the start of RRT as expressed by higher SOFA scores. Of note, >85% of surviving cancer patients were not dependent on RRT at hospital discharge. Corroborating the findings of the present study, in the study by Soares et al., renal function recovered in 82% of patients at 6 months of follow-up [7]. We performed multivariate analyses to identify independent predictors of hospital mortality and three models were fitted. Adjusting for other covariates, including the presence of a malignancy, outcomes in RRT patients were mostly dependent on well-known predictors of mortality such as older age, poor chronic health status, presence of comorbidities, number of failing organs, time to start of RRT in the ICU and oliguria [15,25,27,28]. Our results also reinforce that the main prognostic factor in critically ill patients with cancer is the severity of organ dysfunctions [3,7,10,13,25]. Nevertheless, many limitations must be taken into consideration in our study. First, data were retrieved from a large prospective database of three tertiary care hospitals. Although the type of cancer was known in all patients and the performance status could be deduced from the chronic health status, additional data to better

6 6 E. Maccariello et al. Table 5. Multivariate analyses of predictors of hospital mortality in cancer and non-cancer patients (n = 773) Non-cancer patients (n = 655) Cancer patients (n = 118) Coefficient Odds ratio (95% CI) P-value Coefficient Odds ratio (95% CI) P-value Age (years) ( ) <0.001 Medical admission No 1.00 Yes ( ) <0.001 Poor chronic health status (Knaus C or D) No 1.00 Yes ( ) <0.001 Charlson index (except cancer-related points) ( ) Oliguria a No 1.00 Yes ( ) Mechanical ventilation a No 1.00 Yes ( ) ICU days until the start of dialysis (Ln days) ( ) < ( ) Associated organ dysfunctions total (n) a ( ) < ( ) Constant Model for non-cancer patients: area under receiver operating characteristic curve = 0.84 (95% CI, ); Hosmer Lemeshow goodness-of-fit (χ 2 = 5.636; P = 0.688). Model for cancer patients: area under receiver operating characteristic curve = 0.75 (95% CI, ); Hosmer Lemeshow goodness-of-fit (χ 2 = 8.325; P = 0.402). a At the start of renal replacement therapy. characterize the underlying malignancy (e.g. presence of neutropaenia, disease stage and status, cancer-related complications) and anticancer treatments were unavailable. Additionally, we were not able to analyse subgroups of patients with cancer, and hence, the present study is underpowered to evaluate patients with haematological malignancies. However, we believe that our results may be more representative of the general practice in non-specialized hospitals and therefore more suitable to generalization. In large multicentre studies in general ICUs, patients with cancer accounted for 10 to 21% of all admissions, and solid tumours were much more frequent than haematological malignancies [20,25,30]. Second, we cannot rule out possible selection biases concerning regional specificities related to standards of care, including criteria used to start dialysis, implement EOL decisions and ICU admission/discharge policies. Third, bone marrow transplant (BMT) patients were not evaluated as this treatment is not regularly performed in nonspecialized centres. In addition, we consider that BMT (in particular, allogeneic BMT) patients should be studied separately as they have peculiarities that differentiate them in terms of outcomes with mortality rates that remain disproportionably high when RRT is required because of AKI [2,31]. Finally, endpoints in outcome studies should not be restricted to survival. Although we have evaluated the recovering of renal function at hospital discharge, it is necessary to evaluate and compare survival rates, renal function and aspects related to quality of life in medium- and long-term follow-up [32]. Particularly in patients with cancer, it is also needed to study the impact of AKI on the continuation of anticancer treatments [32]. Conclusion In summary, one in six patients requiring RRT in general ICUs has cancer. Despite a relatively higher mortality, the presence of cancer was not independently associated with increased mortality in the present cohort. However, additional studies are needed to compare the outcomes of cancer patients admitted to specialized and general ICUs. Supplementary data Supplementary data is available online at oxfordjournals.org. Conflict of interest statement. None declared. References 1. Darmon M, Ciroldi M, Thiery G et al. Clinical review: specific aspects of acute renal failure in cancer patients. Crit Care 2006; 10: Benoit DD, Hoste EA, Depuydt PO et al. Outcome in critically ill medical patients treated with renal replacement therapy for acute renal failure: comparison between patients with and those without haematological malignancies. Nephrol Dial Transplant 2005; 20: Darmon M, Thiery G, Ciroldi M et al. Should dialysis be offered to cancer patients with acute kidney injury? Intensive Care Med 2007; 33: Benoit DD, Vandewoude KH, Decruyenaere JM et al. Outcome and early prognostic indicators in patients with a hematologic malignancy admitted to the intensive care unit for a life-threatening complication. Crit Care Med 2003; 31:

7 Outcomes of cancer and non-cancer patients with AKI 7 5. Peigne V, Rusinová K, Karlin L et al. Continued survival gains in recent years among critically ill myeloma patients. Intensive Care Med 2009; 35: Maschmeyer G, Bertschat FL, Moesta KT et al. Outcome analysis of 189 consecutive cancer patients referred to the intensive care unit as emergencies during a 2-year period. Eur J Cancer 2003; 39: Soares M, Salluh JI, Carvalho MS et al. Prognosis of critically ill patients with cancer and acute renal dysfunction. J Clin Oncol 2006; 24: Lameire N, Van Biesen W, Vanholder R. Acute renal problems in the critically ill cancer patient. Curr Opin Crit Care 2008; 14: Létourneau I, Dorval M, Bélanger R et al. Acute renal failure in bone marrow transplant patients admitted to the intensive care unit. Nephron 2002; 90: Berghmans T, Meert AP, Markiewicz E et al. Continuous venovenous haemofiltration in cancer patients with renal failure: a single-centre experience. Support Care Cancer 2004; 12: Vieira JM Jr, Castro I, Curvello-Neto A et al. Effect of acute kidney injury on weaning from mechanical ventilation in critically ill patients. Crit Care Med 2007; 35: Pène F, Percheron S, Lemiale V et al. Temporal changes in management and outcome of septic shock in patients with malignancies in the intensive care unit. Crit Care Med 2008; 36: Lecuyer L, Chevret S, Guidet B et al. Case volume and mortality in haematological patients with acute respiratory failure. Eur Respir J 2008; 32: Lanore JJ, Brunet F, Pochard F et al. Hemodialysis for acute renal failure in patients with hematologic malignancies. Crit Care Med 1991; 19: Maccariello E, Soares M, Valente C et al. RIFLE classification in patients with acute kidney injury in need of renal replacement therapy. Intensive Care Med 2007; 33: Knaus WA, Wagner DP, Draper EA et al. APACHE - acute physiology and chronic health evaluation: a physiologically based classification system. Crit Care Med 1981; 9: Le Gall J-R, Lemeshow S, Saulnier F. A new simplified acute physiology score (SAPS II) based on a European/North American multicenter study. JAMA 1993; 270: Charlson ME, Pompei P, Ales KL et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40: Vincent JL, Moreno R, Takala J et al. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996; 22: Taccone FS, Artigas AA, Sprung CL et al. Characteristics and outcomes of cancer patients in European ICUs. Crit Care 2009; 13: R Levy MM, Fink MP, Marshall JC et al SCCM/ESICM/ACCP/ ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003; 31: Bellomo R, Ronco C, Kellum JA et al. Acute Dialysis Quality Initiative workgroup. Acute renal failure - definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004; 8: R204 R Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: Wiley Interscience, Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143: Soares M, Caruso P, Silva E et al. Characteristics and outcomes of patients with cancer requiring admission to intensive care units: a prospective multicenter study. Crit Care Med 2010; 38: Metnitz PG, Moreno RP, Almeida E et al. SAPS 3 From evaluation of the patient to evaluation of the intensive care unit. Part 1: Objectives, methods and cohort description. Intensive Care Med 2005; 31: Metnitz PG, Krenn CG, Steltzer H et al. Effect of acute renal failure requiring renal replacement therapy on outcome in critically ill patients. Crit Care Med 2002; 30: Uchino S, Kellum JA, Bellomo R et al. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA 2005; 294: Santos WJ, Zanetta DM, Pires AC et al. Patients with ischaemic, mixed and nephrotoxic acute tubular necrosis in the intensive care unit a homogeneous population? Crit Care 2006; 10: R de Mendonca A, Vincent JL, Suter PM et al. Acute renal failure in the ICU: risk factors and outcome evaluated by the SOFA score. Intensive Care Med 2000; 26: Hahn T, Rondeau C, Shaukat A et al. Acute renal failure requiring dialysis after allogeneic blood and marrow transplantation identifies very poor prognosis patients. Bone Marrow Transplant 2003; 32: Soares M, Azoulay E. Critical care management of lung cancer patients to prolong life without prolonging dying. Intensive Care Med 2009; 35: Received for publication: ; Accepted in revised form: