J Neurooncol DOI 10.1007/s11060-015-2015-7
CLINICAL STUDY
Neuro-oncological patients admitted in intensive-care unit: predictive factors and functional outcome E. Tabouret1,2 • C. Boucard1 • R. Devillier3 • M. Barrie1 • S. Boussen3 D. Autran1 • O. Chinot1,2 • N. Bruder4
•
Received: 12 September 2015 / Accepted: 22 November 2015 Ó Springer Science+Business Media New York 2015
Abstract The prognosis of oncology patients admitted to the intensive care unit (ICU) is considered poor. Our objective was to analyze the characteristics and predictive factors of death in the ICU and functional outcome following ICU treatment for neuro-oncology patients. A retrospective study was conducted on all patients with primary brain tumor admitted to our institutional ICU for medical indications. Predictive impact on the risk of death in the ICU was analyzed as well as the functional status was evaluated prior and following ICU discharge. Seventy-one patients were admitted to the ICU. ICU admission indications were refractory seizures (41 %) and septic shock (17 %). On admission, 16 % had multi-organ failure. Ventilation was necessary for 41 % and catecholamines for 13 %. Twentytwo percent of patients died in the ICU. By multivariate analysis, predictive factors associated with an increased risk of ICU death were: non-neurological cause of admission [p = 0.045; odds ratio (OR) 5.405], multiple organ failure (p = 0.021; OR 8.027), respiratory failure (p = 0.006; OR 9.615), and hemodynamic failure (p = 0.008; OR 10.111).
Electronic supplementary material The online version of this article (doi:10.1007/s11060-015-2015-7) contains supplementary material, which is available to authorized users. & E. Tabouret [email protected] 1
Department of Neuro-Oncology, AP-HM, Timone, 264, rue Saint Pierre, 13005 Marseille, France
2
Aix-Marseille Universite´, CRO2, UMR911, 13005 Marseille, France
3
Hematology Department, Institut Paoli Calmettes, 13009 Marseille, France
4
Department of Anesthesia and Intensive Care, Aix-Marseille Universite´; AP-HM, Timone, 13005 Marseille, France
In contrast, tumor type (p = 0.678) and disease control status (p = 0.380) were not associated with an increased risk of ICU death. Among the 35 evaluable patients, 77 % presented with a stable or improved Karnofsky performance status following ICU hospitalization compared with the ongoing status before discharge. In patients with primary brain tumor admitted to the ICU, predictive factors of death appear to be similar to those described in non-oncology patients. ICU hospitalization is generally not associated with a subsequent decrease in the functional status. Keywords Intensive care Neuro-oncology Glioma Primary cerebral lymphoma
Background The prognosis of oncology patients admitted to intensive care units (ICU) is traditionally considered poor, limiting patient presentations and admissions [1]. Socio-economic and ethical controversies may place additional constraints on the aggressive ICU management of oncology patients [2]. Thus, the identification of patients who may benefit from ICU management is a key step prior to admission. However, no multidisciplinary predictive scoring system is available for patient selection to date [3]. The ICU mortality rate in solid tumor patients has been reported as [50 % [4–7], while in conventional medicine, the ICU mortality rate is approximately 20 % [8, 9]. Over the past decade, both oncology outcomes and ICU patient management have improved, increasing the number of oncology patients admitted to the ICU and justifying current developments of disease-specific approaches to ICU management [10, 11]. The prognosis and quality of life (QoL) of patients admitted to neuro-oncology departments for primary brain
123
J Neurooncol
tumors have recently improved because of recent therapeutic advances in this field [12, 13]. However, these regimens remain principally based on toxic drugs for which severe complications may give rise to organ failure. These complications may differ from those of other oncology patients because of cerebral tumor involvement, which could potentially lead to specific complications, such as seizure and cerebral hemorrhage, a rare event in general oncological population. Moreover, the oncogenesis of primary brain tumors differs from that of other oncology patients by the absence of classical risk factors, such as tobacco and alcohol. Thus, neuro-oncology patients could be more frequently free of comorbidities with a preserved general status. Despite this neuro-oncologic specificity, no evaluation of the ICU management in this specific setting has been reported. ICU resources are limited in all countries, which may lead to inaccurate patient selection for the admission of neuro-oncology patients to the ICU because of a perceived lack of benefit from intensive treatment. Indeed, the known poor prognosis of these patients, the limited activity of treatments and the rarity of primary brain tumors lead to a lack of knowledge of the specificity and the optimal management of such patients by the intensivists who could be skeptical for aggressive approach. Thus, our objective was to evaluate the predictive factors of death in the ICU and the functional outcome following ICU admission in the specific field of neurooncology patients.
Methods Selection criteria We performed a monocentric, retrospective chart review of all patients with a diagnosis of primary cerebral brain tumor admitted to the ICU of our institution between January 2002 and December 2012. Patients were at least 18 years of age with histological proof of primary cerebral tumor [glioma, primary central nervous system lymphoma (PCNSL), ependymoma, medulloblastoma, and germinal tumor]. This protocol was approved by our institutional review board (AP HM). The study was conducted in accordance with the principles of the declaration of Helsinki. Only medical indications of admission to the ICU were considered. Patients admitted directly following brain surgery (ICU management needed by the surgery) were not included. Endpoints We searched our hospital database using the following criteria: diagnosis for admission to the ICU and either
123
patient origin from the neuro-oncology unit or discharge from the ICU to the neuro-oncology unit. Since all patients with primary brain tumor were referred to our neuro-oncological department, and considering that our institution was composed by only one ICU department including the neurological ICU, the cross of these two hospital databases was used in order to include all patients followed for primary brain tumor, admitted to ICU department. We then searched for predictive factors of death in the ICU or during the global hospital stay. Variables were assessed based on the severe sepsis definition [14]. Functional status was assed using the Karnofsky Performance Status (KPS), generally part of patient evaluation in our department and so integrated in the medical record. If missing, we retrospectively evaluated it based on medical record. The baseline KPS was the one evaluated at the last medical examination before developing the symptoms related to ICU admission and at 2 to 4 months after ICU discharge. Finally, we analyzed the overall patient outcome. Statistical analyses Predictive impact on the risk of death in the ICU or during the global hospitalization stay was analyzed using the Chi square (v2) test and the Mann–Whitney U-test. Multivariate analysis was performed using multivariate logistic regression. To search for an effect of time on mortality, the cohort was divided into two historical classes (before 2008 and since 2008) to compare patient characteristics or treatment modalities of each period using the v2 and the Mann–Whitney tests. Overall survival (OS) was defined by the time between the initial diagnosis to the time of death from any cause since admission to the ICU, censored at the date of last contact with living patients. Time-to-event endpoints were estimated using the method of Kaplan and Meier and compared using the log-rank test. Univariate and multivariate Cox proportional hazard regression models were used to estimate the hazard ratio. All analyses were conducted with a bilateral alpha type 1 error of 5 %.
Results Patients’ characteristics (Table 1; Online Table 1) From January 2002 to December 2012, among 2745 patients referred to our department, 71 patients with primary brain tumors were admitted to the ICU, representing less than 2.6 % of patients referred to our neuro-oncology department. Forty-nine percent of patients presented without comorbitites. Twenty-nine (45 %) patients presented with medical past history of seizure during their disease. The majority (68 %) had glioma as the primary
J Neurooncol Table 1 Patients’ characteristics at the time of ICU admission Data
N
Age (years)
56 (19–85)
%
Clinical status
%
Multiple organs failures
16
9
Respiratory disorders
32
17
Hepatic dysfunction
10
42
Hemodynamic instability
25 6 28
Tumor type Grade II glioma Anaplastic glioma Glioblastoma
6 12 30
Table 2 Clinical status at the time of ICU admission (numbers are noted in % or median range)
Primary cerebral lymphoma
18
25
Renal dysfunction
Other
5
7
Hematological anomalies
Newly diagnosed (before treatment)
9
14
Controlled (CR, PR, S)
27
42
Disease status
Uncontrolled (progressive disease) 28 Previous lines of chemotherapy Before ICU admission
44
Anemia Neutropenia Thrombopenia
11 14 14
Metabolic abnormality
17
Abnormal neurological examination
87
0
7
10
Glasgow coma score
13 (5–15)
1
34
52
SPAS II score
37 (0–94)
2
9
14
[2
16
24
Variables were defined based on the severe sepsis definition (Dellinger et al., Intensive Care Medicine 2013)
Use of bevacizumab before ICU admission
12
19
Use of GliadelÒ
3
4.5
Past medical history of seizure
29
45
Absent
35
49
Cardio-vascular only
16
23
Other
8
11
Multiplesa
12
17
Comorbidities
CR complete response; PR partial response; S stable disease a
Cardiovascular and other disease
tumor, while one quarter had PCNSL. Forty-two percent of patients presented with controlled brain tumors and 14 % of them were admitted to the ICU at the time of brain tumor diagnosis. Twenty-four percent of patients had previously received more than two lines of treatment. Finally, the median patient age was 56 years (range, 19–85), which was similar to the median age of the general population admitted to the ICU. Etiologies, clinical status at admission, and treatment in the ICU (Table 2) Most ICU admissions occurred for neurological disorders and were most frequently for refractory status epilepticus. Refractory status epilepticus was associated with more frequent seizure past medical history (p = 0.007). Septic shock or respiratory failure were the causes of ICU admission in 17 and 16 % of cases, respectively (Fig. 1). Median time before first symptom of failure to ICU admission was 2 days (range 0–31). Most patients had decreased consciousness at the time of admission to the
ICU (Glasgow coma score \15) (Table 2). Thirty-two percent of them had respiratory disorders and 25 % had hemodynamic instability. Blood count abnormalities were found in 28 % of the patients. Forty-one percent of patients (n = 27) required mechanical ventilation (tracheal intubation, n = 26 and non-invasive ventilation, n = 1). The median time of tracheal intubation was 2 days (range 1–50 days). Thirteen percent of patients (n = 8) received catecholamines: adrenaline (n = 2), noradrenaline (n = 3), or dobutamine (n = 3). Nine percent of patients (n = 6) received chemotherapy in the ICU. Median duration of ICU hospitalization was 3 days (range, 1–186) and median duration of conventional hospitalization after ICU was 4 (range, 1–18). Historical comparisons between patients admitted before and after the year 2008 showed no significant difference between patient characteristics, status at admission to the ICU, and treatment approaches. Predictive factors of death in the ICU or in the hospital (Table 3) Sixteen patients (22 %) died in the ICU, while six (11 %) of them died in the ward following ICU discharge. The presence of the following factors at the time of admission was significantly associated with an increased risk of death in the ICU: non-neurological etiology of admission, multiple organ failure, respiratory, hemodynamic, hepatic, or renal dysfunction. Regarding treatment approaches, mechanical ventilation and catecholamine infusion were significantly associated with a higher risk of death in the ICU. In contrast, an abnormal Glasgow coma score as well
123
J Neurooncol
Fig. 1 Causes of ICU admission
as tumor type, tumor control status, and the number of previous courses of chemotherapy were not associated with an increased risk of death in the ICU (Table 3). No patient receiving chemotherapy in the ICU died there, while only one of them died in the conventional unit. Finally, a longer duration between symptom onset and ICU admission tended to be associated with a higher risk of death in the ICU. Patients who died in the ICU were admitted with a median time of 4.5 days vs. 1 day for patients who did not die, and only 31 % of patients who died in the ICU were admitted within 24 h, vs. 40 % (p = 0.083). According to multivariate analysis adjusted for tumor type and disease control, etiology of admission [OR 5.4; 95 % confidence interval (CI) 1.041–28.051; p = 0.045], multiple organ failure [OR 8.0; 95 % CI 1.368–47.112; p = 0.021], respiratory failure [OR 9.6; 95 % CI 1.897–48.733; p = 0.006] and hemodynamic instability [OR 10.1; 95 % CI 1.827–55.952; p = 0.008] remained significantly associated with higher mortality in the ICU. Analyses of predictive markers of death during the hospital stay were the same as those for the ICU except for etiology, hyperbilirubinemia, and renal dysfunction (Online Table 2). A worse KPS was significantly associated with in-hospital death but not with death in the ICU. Functional status (Fig. 2) To analyze the potential functional impact of ICU admission, we compared the baseline patient KPS (before the onset of the first symptom related to ICU admission) with the KPS between 2 and 4 months after ICU discharge. Among the 49 patients alive after ICU discharge, KPS before and after ICU hospitalization could be assessed in 35 patients: 9 patients were admitted to the ICU at the time
123
of initial diagnosis and complete data were not available for 5 patients. Among these 35 evaluable patients, 27 (77 %) presented with a stable or improved KPS after ICU discharge. Survival analyses (Online Table 3) The 3-, 6-month-, 1-, and 2-year OS rate were 35, 24, 13, and 6 %, respectively. Factors that poorly impacted OS were as follows: advanced age, uncontrolled disease, number of previous line of chemotherapy prior to ICU admission, and KPS \ 70. Histological subtype tended to influence OS.
Discussion Despite the neuro-oncological specificity of patients, and in the context of recent improvements and increases in available therapies [12, 13, 15], to our knowledge, no evaluation of ICU management in this specific setting has been reported to date. In the present study, neuro-oncology patients admitted to the ICU represented only 2.6 % of patients referred to our neuro-oncology department. At present, the ICU mortality rate of our patients is 22 %, increasing to 33 % during hospitalization. This rate is close to the range reported for other ICU patients [8, 9]. In other cancer patients referred to an ICU, mortality is frequently higher, close to 50 % in the literature [4, 7]. This difference can be explained in part by the specificity of neuro-oncology patients admitted to the ICU. In our experience, half of patient admissions were because of neurological causes, mainly status epilepticus, which appeared to be a favorable factor associated with a shorter duration of stay in the ICU
J Neurooncol Table 3 Predictive factors of death in the ICU according to univariate analyses
Factors
Death in ICU risk p value
Relative risk
Qualitative analyses Etiologies (neurological versus other)
0.022
0.370
Multiple organ failure
\0.001
5.455
Respiratory disorder
\0.001
6.260
Hepatic dysfunction
0.003
4.920
Hyperbilirubinemia
0.016
5.250
\0.001
4.907
Renal dysfunction
0.023
4.575
Hematological abnormalities
0.054
Anemia
0.794
Neutropenia Thrombopenia
0.437 0.197
Diabetes mellitus
0.597
Hemodynamic instability
Metabolic abnormality
0.245
Impaired consciousness
0.205
Tracheal intubation
0.001
7.407
Catecholamine use
0.005
7.375
Chemotherapy in ICU
0.326
Histology
0.678
Glioblastoma versus other glioma
1.000
Disease control
0.380
Previous line(s) of chemotherapy (B1 versus [1)
0.409
Bevacizumab use
1.000
Past medical history (yes versus no)
0.949
Continuous variables (quantitative analyses) Age
0.978 a
Time from first symptom to IC admission Duration of IC hospitalization
\0.001
SAPS II score
a
0.083 0.873
Duration of ventilation
0.734
Karnofsky Performance Status before ICU
0.081
Which motivated ICU admission
Fig. 2 KPS evolution before and after ICU admission
(median: 3 days). These results differ from those reported in other studies, where neurological causes were very rare, with mortality rates often higher than 50 % [4, 5, 7, 16,
17]. Neurological failure is a marker of severity in these patients and is rarely the primary cause of ICU admission. In these reports, the major causes of ICU admissions were septic shock and respiratory failure. Moreover, in our study, only 14 % of admitted patients had newly diagnosed cancer. This rate grew to 71 % in some studies, including patients with other malignancies [18, 19]. Taken together, these distinct characteristics could explain, in part, the relatively low ICU mortality of our patients compared with that of other cancer patients reported in the literature. Predictive factors for ICU mortality in our series were close to those reported in other oncology ICU studies. We observed that respiratory, hemodynamic, hepatic, and renal dysfunction; therefore, multiple organ failure, were predictive of a higher mortality rate at the time of ICU
123
J Neurooncol
discharge. Again, as reported in other cohorts, tracheal intubation and catecholamine use were also predictive of a higher mortality rate [7, 11, 17, 18]. In addition, the length of time between the first symptom onset and ICU admission tended to predict ICU mortality in the present study. This factor was one of the most adverse reported by the ‘‘prospective multicentric study of the groupe de recherche respiratoire en reanimation onco-he´matologique’’ study [20]. This trial evaluated 1011 patients admitted to the ICU, mainly for acute respiratory failure and/or shock. A time to ICU admission of less than 24 h appeared to be associated with better hospital survival, while organ dysfunction was associated with higher hospital mortality. These results were in line with other literature reports in which earlier ICU admission appeared to be one of the most important prognostic factors for patients requiring ICU admission [5, 21]. Compared with results of the large ICON study of ICUs around the world, patients with primary tumors had comparable prognostic factors to those of general ICU patients [9]. In this study, we observe that almost all patients requiring respiratory assistance required tracheal intubation. This is contrary to results showing that non-invasive ventilation for cancer patients may improve the outcome [22]. However, tracheal intubation was related mainly to patients’ neurological status rather than severity of hypoxemia. The tumor type and disease control status did not influence the ICU mortality rate in our study, underlining the significant distinction between predictive factors of ICU mortality and disease-related prognostic factors of overall outcome. This is consistent with the results reported in the context of hematological malignancies. In contrast, other studies found that disease status was a principal determinant of ICU mortality in patients treated for solid tumors [4, 7, 18]. This finding underlines the need to assess ICU management in a disease-specific manner. The expected impairment of QoL is sometimes considered a limitation to the admission of oncology patients to the ICU. Since KPS was a functional status evaluation, which could reflect one part of the QoL, our results rather go against this statement, since the majority of patients (77 %) did not experience altered functional status following ICU admission. This suggests that ICU admission was not detrimental to surviving patients. This finding is in agreement with the results of a multicenter, prospective study evaluating health-related QoL after ICU admission. The authors reported that 80 % of survivors had no alteration in QoL 3 months after ICU admission [20]. Our study had limitations related to its design as it was a monocentric and retrospective study. A selection bias related to patient admission is possible. However, the final decision regarding ICU admission was always discussed between a senior intensivist and a senior neuro-oncologist.
123
The finding that the population of ICU neuro-oncology patients was comparable between two time periods suggests that our criteria did not change over time. Some patients were assessed as too severe by both senior physicians to benefit from ICU care, mostly due to uncontrolled brain tumor development. Another limitation might be that some patients were not included in the database. To avoid this bias, we performed multiple searches in our whole-patient population database with the help of the statistical department. All retrieved files were checked for brain tumor malignancy.
Conclusions The mortality rate of neuro-oncology patients admitted to the ICU was comparable to that of other populations of ICU patients, indicating that selected patients could benefit from ICU admission. The main cause of ICU admission was neurological worsening. Predictive factors of death appeared to be similar to those described for other malignancies, except for histological type and disease-control status, which did not impair patient outcome. Finally, ICU admission was not associated with a subsequent alteration of functional status. Acknowledgments This work was completed in the SIRIC of Marseille, grant INCa-DGOS-Inserm 6038. ARTCsud. Authors’ Contribution ET conceived the study, participated in its design, acquired and analysed data and drafted the manuscript. RD conceived the study, participated in its design, analyzed data and helped to draft the manuscript. SB, MB, CB, and DA participated in the acquisition of data. SB help to draft the manuscript. OC and NB participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript. Compliance with ethical standards Conflict of interest No potential conflict of interest relevant to this article was reported.
References 1. Kostakou E, Rovina N, Kyriakopoulou M, Koulouris NG, Koutsoukou A (2014) Critically ill cancer patient in intensive care unit: issues that arise. J Crit Care 29:817–822 2. Kulkarni AP, Divatia JV (2013) A prospective audit of costs of intensive care in cancer patients in India. Indian J Crit Care Med 17:292–297 3. Wigmore TJ, Farquhar-Smith P, Lawson A (2013) Intensive care for the cancer patient—unique clinical and ethical challenges and outcome prediction in the critically ill cancer patient. Best Pract Res Clin Anaesthesiol 27:527–543 4. Aygencel G, Turkoglu M, Turkoz Sucak G, Benekli M (2014) Prognostic factors in critically ill cancer patients admitted to the intensive care unit. J Crit Care 29:618–626
J Neurooncol 5. de Montmollin E, Tandjaoui-Lambiotte Y, Legrand M et al (2013) Outcomes in critically ill cancer patients with septic shock of pulmonary origin. Shock Augusta Ga 39:250–254 6. Lamia B, Hellot M-F, Girault C, Tamion F, Dachraoui F, Lenain P, Bonmarchand G (2006) Changes in severity and organ failure scores as prognostic factors in onco-hematological malignancy patients admitted to the ICU. Intensive Care Med 32:1560–1568 7. Azoulay E, Moreau D, Alberti C, Leleu G, Adrie C, Barboteu M, Cottu P, Levy V, Le Gall JR, Schlemmer B (2000) Predictors of short-term mortality in critically ill patients with solid malignancies. Intensive Care Med 26:1817–1823 8. Le Gall JR, Lemeshow S, Saulnier F (1993) A new simplified acute physiology score (SAPS II) based on a European/North American multicenter study. JAMA 270:2957–2963 9. Vincent J-L, Marshall JC, Namendys-Silva SA et al (2014) Assessment of the worldwide burden of critical illness: the intensive care over nations (ICON) audit. Lancet Respir Med 2:380–386 10. Thie´ry G, Azoulay E, Darmon M, Ciroldi M, De Miranda S, Le´vy V, Fieux F, Moreau D, Le Gall JR, Schlemmer B (2005) Outcome of cancer patients considered for intensive care unit admission: a hospital-wide prospective study. J Clin Oncol Off J Am Soc Clin Oncol 23:4406–4413 11. Ferra` C, Marcos P, Misis M et al (2007) Outcome and prognostic factors in patients with hematologic malignancies admitted to the intensive care unit: a single-center experience. Int J Hematol 85:195–202 12. Chinot OL, Wick W, Mason W et al (2014) Bevacizumab plus radiotherapy-temozolomide for newly diagnosed glioblastoma. N Engl J Med 370:709–722 13. Soussain C, Hoang-Xuan K, Taillandier L et al (2008) Intensive chemotherapy followed by hematopoietic stem-cell rescue for refractory and recurrent primary CNS and intraocular lymphoma: Socie´te´ Franc¸aise de Greffe de Moe¨lle Osseuse-The´rapie Cellulaire. J Clin Oncol Off J Am Soc Clin Oncol 26:2512–2518 14. Dellinger RP, Levy MM, Rhodes A et al (2013) Surviving sepsis campaign: international guidelines for management of severe sepsis and septic shock: 2012. Crit Care Med 41:580–637
15. Sampson JH, Heimberger AB, Archer GE et al (2010) Immunologic escape after prolonged progression-free survival with epidermal growth factor receptor variant III peptide vaccination in patients with newly diagnosed glioblastoma. J Clin Oncol Off J Am Soc Clin Oncol 28:4722–4729 16. Chang Y, Huh J-W, Hong S-B, Lee DH, Suh C, Kim S-W, Lim C-M, Koh Y (2014) Outcomes and prognostic factors of patients with lung cancer and pneumonia-induced respiratory failure in a medical intensive care unit: a single-center study. J Crit Care 29:414–419 17. Soares M, Silva UVA, Teles JMM, Silva E, Caruso P, Lobo SMA, Dal Pizzol F, Azevedo LP, de Carvalho FB, Salluh JIF (2010) Validation of four prognostic scores in patients with cancer admitted to Brazilian intensive care units: results from a prospective multicenter study. Intensive Care Med 36:1188–1195 18. Soares M, Toffart A-C, Timsit J-F et al (2014) Intensive care in patients with lung cancer: a multinational study. Ann Oncol Off J Eur Soc Med Oncol ESMO 25:1829–1835 19. Kim YJ, Kim M-J, Cho Y-J et al (2014) Who should be admitted to the intensive care unit? The outcome of intensive care unit admission in stage IIIB-IV lung cancer patients. Med Oncol Northwood Lond Engl 31:847 20. Azoulay E, Mokart D, Pe`ne F et al (2013) Outcomes of critically ill patients with hematologic malignancies: prospective multicenter data from France and Belgium—a groupe de recherche respiratoire en re´animation onco-he´matologique study. J Clin Oncol Off J Am Soc Clin Oncol 31:2810–2818 21. Larche´ J, Azoulay E, Fieux F, Mesnard L, Moreau D, Thiery G, Darmon M, Le Gall J-R, Schlemmer B (2003) Improved survival of critically ill cancer patients with septic shock. Intensive Care Med 29:1688–1695 22. Azoulay E, Alberti C, Bornstain C, Leleu G, Moreau D, Recher C, Chevret S, Le Gall JR, Brochard L, Schlemmer B (2001) Improved survival in cancer patients requiring mechanical ventilatory support: impact of noninvasive mechanical ventilatory support. Crit Care Med 29:519–525
123
CLINICAL STUDY
Neuro-oncological patients admitted in intensive-care unit: predictive factors and functional outcome E. Tabouret1,2 • C. Boucard1 • R. Devillier3 • M. Barrie1 • S. Boussen3 D. Autran1 • O. Chinot1,2 • N. Bruder4
•
Received: 12 September 2015 / Accepted: 22 November 2015 Ó Springer Science+Business Media New York 2015
Abstract The prognosis of oncology patients admitted to the intensive care unit (ICU) is considered poor. Our objective was to analyze the characteristics and predictive factors of death in the ICU and functional outcome following ICU treatment for neuro-oncology patients. A retrospective study was conducted on all patients with primary brain tumor admitted to our institutional ICU for medical indications. Predictive impact on the risk of death in the ICU was analyzed as well as the functional status was evaluated prior and following ICU discharge. Seventy-one patients were admitted to the ICU. ICU admission indications were refractory seizures (41 %) and septic shock (17 %). On admission, 16 % had multi-organ failure. Ventilation was necessary for 41 % and catecholamines for 13 %. Twentytwo percent of patients died in the ICU. By multivariate analysis, predictive factors associated with an increased risk of ICU death were: non-neurological cause of admission [p = 0.045; odds ratio (OR) 5.405], multiple organ failure (p = 0.021; OR 8.027), respiratory failure (p = 0.006; OR 9.615), and hemodynamic failure (p = 0.008; OR 10.111).
Electronic supplementary material The online version of this article (doi:10.1007/s11060-015-2015-7) contains supplementary material, which is available to authorized users. & E. Tabouret [email protected] 1
Department of Neuro-Oncology, AP-HM, Timone, 264, rue Saint Pierre, 13005 Marseille, France
2
Aix-Marseille Universite´, CRO2, UMR911, 13005 Marseille, France
3
Hematology Department, Institut Paoli Calmettes, 13009 Marseille, France
4
Department of Anesthesia and Intensive Care, Aix-Marseille Universite´; AP-HM, Timone, 13005 Marseille, France
In contrast, tumor type (p = 0.678) and disease control status (p = 0.380) were not associated with an increased risk of ICU death. Among the 35 evaluable patients, 77 % presented with a stable or improved Karnofsky performance status following ICU hospitalization compared with the ongoing status before discharge. In patients with primary brain tumor admitted to the ICU, predictive factors of death appear to be similar to those described in non-oncology patients. ICU hospitalization is generally not associated with a subsequent decrease in the functional status. Keywords Intensive care Neuro-oncology Glioma Primary cerebral lymphoma
Background The prognosis of oncology patients admitted to intensive care units (ICU) is traditionally considered poor, limiting patient presentations and admissions [1]. Socio-economic and ethical controversies may place additional constraints on the aggressive ICU management of oncology patients [2]. Thus, the identification of patients who may benefit from ICU management is a key step prior to admission. However, no multidisciplinary predictive scoring system is available for patient selection to date [3]. The ICU mortality rate in solid tumor patients has been reported as [50 % [4–7], while in conventional medicine, the ICU mortality rate is approximately 20 % [8, 9]. Over the past decade, both oncology outcomes and ICU patient management have improved, increasing the number of oncology patients admitted to the ICU and justifying current developments of disease-specific approaches to ICU management [10, 11]. The prognosis and quality of life (QoL) of patients admitted to neuro-oncology departments for primary brain
123
J Neurooncol
tumors have recently improved because of recent therapeutic advances in this field [12, 13]. However, these regimens remain principally based on toxic drugs for which severe complications may give rise to organ failure. These complications may differ from those of other oncology patients because of cerebral tumor involvement, which could potentially lead to specific complications, such as seizure and cerebral hemorrhage, a rare event in general oncological population. Moreover, the oncogenesis of primary brain tumors differs from that of other oncology patients by the absence of classical risk factors, such as tobacco and alcohol. Thus, neuro-oncology patients could be more frequently free of comorbidities with a preserved general status. Despite this neuro-oncologic specificity, no evaluation of the ICU management in this specific setting has been reported. ICU resources are limited in all countries, which may lead to inaccurate patient selection for the admission of neuro-oncology patients to the ICU because of a perceived lack of benefit from intensive treatment. Indeed, the known poor prognosis of these patients, the limited activity of treatments and the rarity of primary brain tumors lead to a lack of knowledge of the specificity and the optimal management of such patients by the intensivists who could be skeptical for aggressive approach. Thus, our objective was to evaluate the predictive factors of death in the ICU and the functional outcome following ICU admission in the specific field of neurooncology patients.
Methods Selection criteria We performed a monocentric, retrospective chart review of all patients with a diagnosis of primary cerebral brain tumor admitted to the ICU of our institution between January 2002 and December 2012. Patients were at least 18 years of age with histological proof of primary cerebral tumor [glioma, primary central nervous system lymphoma (PCNSL), ependymoma, medulloblastoma, and germinal tumor]. This protocol was approved by our institutional review board (AP HM). The study was conducted in accordance with the principles of the declaration of Helsinki. Only medical indications of admission to the ICU were considered. Patients admitted directly following brain surgery (ICU management needed by the surgery) were not included. Endpoints We searched our hospital database using the following criteria: diagnosis for admission to the ICU and either
123
patient origin from the neuro-oncology unit or discharge from the ICU to the neuro-oncology unit. Since all patients with primary brain tumor were referred to our neuro-oncological department, and considering that our institution was composed by only one ICU department including the neurological ICU, the cross of these two hospital databases was used in order to include all patients followed for primary brain tumor, admitted to ICU department. We then searched for predictive factors of death in the ICU or during the global hospital stay. Variables were assessed based on the severe sepsis definition [14]. Functional status was assed using the Karnofsky Performance Status (KPS), generally part of patient evaluation in our department and so integrated in the medical record. If missing, we retrospectively evaluated it based on medical record. The baseline KPS was the one evaluated at the last medical examination before developing the symptoms related to ICU admission and at 2 to 4 months after ICU discharge. Finally, we analyzed the overall patient outcome. Statistical analyses Predictive impact on the risk of death in the ICU or during the global hospitalization stay was analyzed using the Chi square (v2) test and the Mann–Whitney U-test. Multivariate analysis was performed using multivariate logistic regression. To search for an effect of time on mortality, the cohort was divided into two historical classes (before 2008 and since 2008) to compare patient characteristics or treatment modalities of each period using the v2 and the Mann–Whitney tests. Overall survival (OS) was defined by the time between the initial diagnosis to the time of death from any cause since admission to the ICU, censored at the date of last contact with living patients. Time-to-event endpoints were estimated using the method of Kaplan and Meier and compared using the log-rank test. Univariate and multivariate Cox proportional hazard regression models were used to estimate the hazard ratio. All analyses were conducted with a bilateral alpha type 1 error of 5 %.
Results Patients’ characteristics (Table 1; Online Table 1) From January 2002 to December 2012, among 2745 patients referred to our department, 71 patients with primary brain tumors were admitted to the ICU, representing less than 2.6 % of patients referred to our neuro-oncology department. Forty-nine percent of patients presented without comorbitites. Twenty-nine (45 %) patients presented with medical past history of seizure during their disease. The majority (68 %) had glioma as the primary
J Neurooncol Table 1 Patients’ characteristics at the time of ICU admission Data
N
Age (years)
56 (19–85)
%
Clinical status
%
Multiple organs failures
16
9
Respiratory disorders
32
17
Hepatic dysfunction
10
42
Hemodynamic instability
25 6 28
Tumor type Grade II glioma Anaplastic glioma Glioblastoma
6 12 30
Table 2 Clinical status at the time of ICU admission (numbers are noted in % or median range)
Primary cerebral lymphoma
18
25
Renal dysfunction
Other
5
7
Hematological anomalies
Newly diagnosed (before treatment)
9
14
Controlled (CR, PR, S)
27
42
Disease status
Uncontrolled (progressive disease) 28 Previous lines of chemotherapy Before ICU admission
44
Anemia Neutropenia Thrombopenia
11 14 14
Metabolic abnormality
17
Abnormal neurological examination
87
0
7
10
Glasgow coma score
13 (5–15)
1
34
52
SPAS II score
37 (0–94)
2
9
14
[2
16
24
Variables were defined based on the severe sepsis definition (Dellinger et al., Intensive Care Medicine 2013)
Use of bevacizumab before ICU admission
12
19
Use of GliadelÒ
3
4.5
Past medical history of seizure
29
45
Absent
35
49
Cardio-vascular only
16
23
Other
8
11
Multiplesa
12
17
Comorbidities
CR complete response; PR partial response; S stable disease a
Cardiovascular and other disease
tumor, while one quarter had PCNSL. Forty-two percent of patients presented with controlled brain tumors and 14 % of them were admitted to the ICU at the time of brain tumor diagnosis. Twenty-four percent of patients had previously received more than two lines of treatment. Finally, the median patient age was 56 years (range, 19–85), which was similar to the median age of the general population admitted to the ICU. Etiologies, clinical status at admission, and treatment in the ICU (Table 2) Most ICU admissions occurred for neurological disorders and were most frequently for refractory status epilepticus. Refractory status epilepticus was associated with more frequent seizure past medical history (p = 0.007). Septic shock or respiratory failure were the causes of ICU admission in 17 and 16 % of cases, respectively (Fig. 1). Median time before first symptom of failure to ICU admission was 2 days (range 0–31). Most patients had decreased consciousness at the time of admission to the
ICU (Glasgow coma score \15) (Table 2). Thirty-two percent of them had respiratory disorders and 25 % had hemodynamic instability. Blood count abnormalities were found in 28 % of the patients. Forty-one percent of patients (n = 27) required mechanical ventilation (tracheal intubation, n = 26 and non-invasive ventilation, n = 1). The median time of tracheal intubation was 2 days (range 1–50 days). Thirteen percent of patients (n = 8) received catecholamines: adrenaline (n = 2), noradrenaline (n = 3), or dobutamine (n = 3). Nine percent of patients (n = 6) received chemotherapy in the ICU. Median duration of ICU hospitalization was 3 days (range, 1–186) and median duration of conventional hospitalization after ICU was 4 (range, 1–18). Historical comparisons between patients admitted before and after the year 2008 showed no significant difference between patient characteristics, status at admission to the ICU, and treatment approaches. Predictive factors of death in the ICU or in the hospital (Table 3) Sixteen patients (22 %) died in the ICU, while six (11 %) of them died in the ward following ICU discharge. The presence of the following factors at the time of admission was significantly associated with an increased risk of death in the ICU: non-neurological etiology of admission, multiple organ failure, respiratory, hemodynamic, hepatic, or renal dysfunction. Regarding treatment approaches, mechanical ventilation and catecholamine infusion were significantly associated with a higher risk of death in the ICU. In contrast, an abnormal Glasgow coma score as well
123
J Neurooncol
Fig. 1 Causes of ICU admission
as tumor type, tumor control status, and the number of previous courses of chemotherapy were not associated with an increased risk of death in the ICU (Table 3). No patient receiving chemotherapy in the ICU died there, while only one of them died in the conventional unit. Finally, a longer duration between symptom onset and ICU admission tended to be associated with a higher risk of death in the ICU. Patients who died in the ICU were admitted with a median time of 4.5 days vs. 1 day for patients who did not die, and only 31 % of patients who died in the ICU were admitted within 24 h, vs. 40 % (p = 0.083). According to multivariate analysis adjusted for tumor type and disease control, etiology of admission [OR 5.4; 95 % confidence interval (CI) 1.041–28.051; p = 0.045], multiple organ failure [OR 8.0; 95 % CI 1.368–47.112; p = 0.021], respiratory failure [OR 9.6; 95 % CI 1.897–48.733; p = 0.006] and hemodynamic instability [OR 10.1; 95 % CI 1.827–55.952; p = 0.008] remained significantly associated with higher mortality in the ICU. Analyses of predictive markers of death during the hospital stay were the same as those for the ICU except for etiology, hyperbilirubinemia, and renal dysfunction (Online Table 2). A worse KPS was significantly associated with in-hospital death but not with death in the ICU. Functional status (Fig. 2) To analyze the potential functional impact of ICU admission, we compared the baseline patient KPS (before the onset of the first symptom related to ICU admission) with the KPS between 2 and 4 months after ICU discharge. Among the 49 patients alive after ICU discharge, KPS before and after ICU hospitalization could be assessed in 35 patients: 9 patients were admitted to the ICU at the time
123
of initial diagnosis and complete data were not available for 5 patients. Among these 35 evaluable patients, 27 (77 %) presented with a stable or improved KPS after ICU discharge. Survival analyses (Online Table 3) The 3-, 6-month-, 1-, and 2-year OS rate were 35, 24, 13, and 6 %, respectively. Factors that poorly impacted OS were as follows: advanced age, uncontrolled disease, number of previous line of chemotherapy prior to ICU admission, and KPS \ 70. Histological subtype tended to influence OS.
Discussion Despite the neuro-oncological specificity of patients, and in the context of recent improvements and increases in available therapies [12, 13, 15], to our knowledge, no evaluation of ICU management in this specific setting has been reported to date. In the present study, neuro-oncology patients admitted to the ICU represented only 2.6 % of patients referred to our neuro-oncology department. At present, the ICU mortality rate of our patients is 22 %, increasing to 33 % during hospitalization. This rate is close to the range reported for other ICU patients [8, 9]. In other cancer patients referred to an ICU, mortality is frequently higher, close to 50 % in the literature [4, 7]. This difference can be explained in part by the specificity of neuro-oncology patients admitted to the ICU. In our experience, half of patient admissions were because of neurological causes, mainly status epilepticus, which appeared to be a favorable factor associated with a shorter duration of stay in the ICU
J Neurooncol Table 3 Predictive factors of death in the ICU according to univariate analyses
Factors
Death in ICU risk p value
Relative risk
Qualitative analyses Etiologies (neurological versus other)
0.022
0.370
Multiple organ failure
\0.001
5.455
Respiratory disorder
\0.001
6.260
Hepatic dysfunction
0.003
4.920
Hyperbilirubinemia
0.016
5.250
\0.001
4.907
Renal dysfunction
0.023
4.575
Hematological abnormalities
0.054
Anemia
0.794
Neutropenia Thrombopenia
0.437 0.197
Diabetes mellitus
0.597
Hemodynamic instability
Metabolic abnormality
0.245
Impaired consciousness
0.205
Tracheal intubation
0.001
7.407
Catecholamine use
0.005
7.375
Chemotherapy in ICU
0.326
Histology
0.678
Glioblastoma versus other glioma
1.000
Disease control
0.380
Previous line(s) of chemotherapy (B1 versus [1)
0.409
Bevacizumab use
1.000
Past medical history (yes versus no)
0.949
Continuous variables (quantitative analyses) Age
0.978 a
Time from first symptom to IC admission Duration of IC hospitalization
\0.001
SAPS II score
a
0.083 0.873
Duration of ventilation
0.734
Karnofsky Performance Status before ICU
0.081
Which motivated ICU admission
Fig. 2 KPS evolution before and after ICU admission
(median: 3 days). These results differ from those reported in other studies, where neurological causes were very rare, with mortality rates often higher than 50 % [4, 5, 7, 16,
17]. Neurological failure is a marker of severity in these patients and is rarely the primary cause of ICU admission. In these reports, the major causes of ICU admissions were septic shock and respiratory failure. Moreover, in our study, only 14 % of admitted patients had newly diagnosed cancer. This rate grew to 71 % in some studies, including patients with other malignancies [18, 19]. Taken together, these distinct characteristics could explain, in part, the relatively low ICU mortality of our patients compared with that of other cancer patients reported in the literature. Predictive factors for ICU mortality in our series were close to those reported in other oncology ICU studies. We observed that respiratory, hemodynamic, hepatic, and renal dysfunction; therefore, multiple organ failure, were predictive of a higher mortality rate at the time of ICU
123
J Neurooncol
discharge. Again, as reported in other cohorts, tracheal intubation and catecholamine use were also predictive of a higher mortality rate [7, 11, 17, 18]. In addition, the length of time between the first symptom onset and ICU admission tended to predict ICU mortality in the present study. This factor was one of the most adverse reported by the ‘‘prospective multicentric study of the groupe de recherche respiratoire en reanimation onco-he´matologique’’ study [20]. This trial evaluated 1011 patients admitted to the ICU, mainly for acute respiratory failure and/or shock. A time to ICU admission of less than 24 h appeared to be associated with better hospital survival, while organ dysfunction was associated with higher hospital mortality. These results were in line with other literature reports in which earlier ICU admission appeared to be one of the most important prognostic factors for patients requiring ICU admission [5, 21]. Compared with results of the large ICON study of ICUs around the world, patients with primary tumors had comparable prognostic factors to those of general ICU patients [9]. In this study, we observe that almost all patients requiring respiratory assistance required tracheal intubation. This is contrary to results showing that non-invasive ventilation for cancer patients may improve the outcome [22]. However, tracheal intubation was related mainly to patients’ neurological status rather than severity of hypoxemia. The tumor type and disease control status did not influence the ICU mortality rate in our study, underlining the significant distinction between predictive factors of ICU mortality and disease-related prognostic factors of overall outcome. This is consistent with the results reported in the context of hematological malignancies. In contrast, other studies found that disease status was a principal determinant of ICU mortality in patients treated for solid tumors [4, 7, 18]. This finding underlines the need to assess ICU management in a disease-specific manner. The expected impairment of QoL is sometimes considered a limitation to the admission of oncology patients to the ICU. Since KPS was a functional status evaluation, which could reflect one part of the QoL, our results rather go against this statement, since the majority of patients (77 %) did not experience altered functional status following ICU admission. This suggests that ICU admission was not detrimental to surviving patients. This finding is in agreement with the results of a multicenter, prospective study evaluating health-related QoL after ICU admission. The authors reported that 80 % of survivors had no alteration in QoL 3 months after ICU admission [20]. Our study had limitations related to its design as it was a monocentric and retrospective study. A selection bias related to patient admission is possible. However, the final decision regarding ICU admission was always discussed between a senior intensivist and a senior neuro-oncologist.
123
The finding that the population of ICU neuro-oncology patients was comparable between two time periods suggests that our criteria did not change over time. Some patients were assessed as too severe by both senior physicians to benefit from ICU care, mostly due to uncontrolled brain tumor development. Another limitation might be that some patients were not included in the database. To avoid this bias, we performed multiple searches in our whole-patient population database with the help of the statistical department. All retrieved files were checked for brain tumor malignancy.
Conclusions The mortality rate of neuro-oncology patients admitted to the ICU was comparable to that of other populations of ICU patients, indicating that selected patients could benefit from ICU admission. The main cause of ICU admission was neurological worsening. Predictive factors of death appeared to be similar to those described for other malignancies, except for histological type and disease-control status, which did not impair patient outcome. Finally, ICU admission was not associated with a subsequent alteration of functional status. Acknowledgments This work was completed in the SIRIC of Marseille, grant INCa-DGOS-Inserm 6038. ARTCsud. Authors’ Contribution ET conceived the study, participated in its design, acquired and analysed data and drafted the manuscript. RD conceived the study, participated in its design, analyzed data and helped to draft the manuscript. SB, MB, CB, and DA participated in the acquisition of data. SB help to draft the manuscript. OC and NB participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript. Compliance with ethical standards Conflict of interest No potential conflict of interest relevant to this article was reported.
References 1. Kostakou E, Rovina N, Kyriakopoulou M, Koulouris NG, Koutsoukou A (2014) Critically ill cancer patient in intensive care unit: issues that arise. J Crit Care 29:817–822 2. Kulkarni AP, Divatia JV (2013) A prospective audit of costs of intensive care in cancer patients in India. Indian J Crit Care Med 17:292–297 3. Wigmore TJ, Farquhar-Smith P, Lawson A (2013) Intensive care for the cancer patient—unique clinical and ethical challenges and outcome prediction in the critically ill cancer patient. Best Pract Res Clin Anaesthesiol 27:527–543 4. Aygencel G, Turkoglu M, Turkoz Sucak G, Benekli M (2014) Prognostic factors in critically ill cancer patients admitted to the intensive care unit. J Crit Care 29:618–626
J Neurooncol 5. de Montmollin E, Tandjaoui-Lambiotte Y, Legrand M et al (2013) Outcomes in critically ill cancer patients with septic shock of pulmonary origin. Shock Augusta Ga 39:250–254 6. Lamia B, Hellot M-F, Girault C, Tamion F, Dachraoui F, Lenain P, Bonmarchand G (2006) Changes in severity and organ failure scores as prognostic factors in onco-hematological malignancy patients admitted to the ICU. Intensive Care Med 32:1560–1568 7. Azoulay E, Moreau D, Alberti C, Leleu G, Adrie C, Barboteu M, Cottu P, Levy V, Le Gall JR, Schlemmer B (2000) Predictors of short-term mortality in critically ill patients with solid malignancies. Intensive Care Med 26:1817–1823 8. Le Gall JR, Lemeshow S, Saulnier F (1993) A new simplified acute physiology score (SAPS II) based on a European/North American multicenter study. JAMA 270:2957–2963 9. Vincent J-L, Marshall JC, Namendys-Silva SA et al (2014) Assessment of the worldwide burden of critical illness: the intensive care over nations (ICON) audit. Lancet Respir Med 2:380–386 10. Thie´ry G, Azoulay E, Darmon M, Ciroldi M, De Miranda S, Le´vy V, Fieux F, Moreau D, Le Gall JR, Schlemmer B (2005) Outcome of cancer patients considered for intensive care unit admission: a hospital-wide prospective study. J Clin Oncol Off J Am Soc Clin Oncol 23:4406–4413 11. Ferra` C, Marcos P, Misis M et al (2007) Outcome and prognostic factors in patients with hematologic malignancies admitted to the intensive care unit: a single-center experience. Int J Hematol 85:195–202 12. Chinot OL, Wick W, Mason W et al (2014) Bevacizumab plus radiotherapy-temozolomide for newly diagnosed glioblastoma. N Engl J Med 370:709–722 13. Soussain C, Hoang-Xuan K, Taillandier L et al (2008) Intensive chemotherapy followed by hematopoietic stem-cell rescue for refractory and recurrent primary CNS and intraocular lymphoma: Socie´te´ Franc¸aise de Greffe de Moe¨lle Osseuse-The´rapie Cellulaire. J Clin Oncol Off J Am Soc Clin Oncol 26:2512–2518 14. Dellinger RP, Levy MM, Rhodes A et al (2013) Surviving sepsis campaign: international guidelines for management of severe sepsis and septic shock: 2012. Crit Care Med 41:580–637
15. Sampson JH, Heimberger AB, Archer GE et al (2010) Immunologic escape after prolonged progression-free survival with epidermal growth factor receptor variant III peptide vaccination in patients with newly diagnosed glioblastoma. J Clin Oncol Off J Am Soc Clin Oncol 28:4722–4729 16. Chang Y, Huh J-W, Hong S-B, Lee DH, Suh C, Kim S-W, Lim C-M, Koh Y (2014) Outcomes and prognostic factors of patients with lung cancer and pneumonia-induced respiratory failure in a medical intensive care unit: a single-center study. J Crit Care 29:414–419 17. Soares M, Silva UVA, Teles JMM, Silva E, Caruso P, Lobo SMA, Dal Pizzol F, Azevedo LP, de Carvalho FB, Salluh JIF (2010) Validation of four prognostic scores in patients with cancer admitted to Brazilian intensive care units: results from a prospective multicenter study. Intensive Care Med 36:1188–1195 18. Soares M, Toffart A-C, Timsit J-F et al (2014) Intensive care in patients with lung cancer: a multinational study. Ann Oncol Off J Eur Soc Med Oncol ESMO 25:1829–1835 19. Kim YJ, Kim M-J, Cho Y-J et al (2014) Who should be admitted to the intensive care unit? The outcome of intensive care unit admission in stage IIIB-IV lung cancer patients. Med Oncol Northwood Lond Engl 31:847 20. Azoulay E, Mokart D, Pe`ne F et al (2013) Outcomes of critically ill patients with hematologic malignancies: prospective multicenter data from France and Belgium—a groupe de recherche respiratoire en re´animation onco-he´matologique study. J Clin Oncol Off J Am Soc Clin Oncol 31:2810–2818 21. Larche´ J, Azoulay E, Fieux F, Mesnard L, Moreau D, Thiery G, Darmon M, Le Gall J-R, Schlemmer B (2003) Improved survival of critically ill cancer patients with septic shock. Intensive Care Med 29:1688–1695 22. Azoulay E, Alberti C, Bornstain C, Leleu G, Moreau D, Recher C, Chevret S, Le Gall JR, Brochard L, Schlemmer B (2001) Improved survival in cancer patients requiring mechanical ventilatory support: impact of noninvasive mechanical ventilatory support. Crit Care Med 29:519–525
123
