The intensive care management process in patients with hematopoietic stem cell transplantation and factors affecting their prognosis

Abstract

Objectives

Hematopoietic stem cell transplantation (HSCT) recipients may require further management in intensive care unit (ICU). The ICU outcome of the HSCT recipients is claimed to have improved significantly over the last two decades. Our aim was to investigate the ICU outcome of the HSCT recipients who required management in ICU, together with the factors that are likely to affect the results.

Materials and methods

We retrospectively investigated the ICU outcome of 48 adults (≥18 years of age) who received HSCT in the bone marrow transplant unit of our hospital and required admission to ICU between 01 January 2007 and 31 December 2010. The data were retrieved from the databases of the adult bone marrow transplantation unit and the ICU.

Results

Sixty-one percent of the patients were male with a median age of 39 years (28–46.75) in the study cohort. Leukemia (54%) and lymphoma (27%) were the leading underlying disorders. The type of HSCT was autologous in 14.6% and allogeneic in 85.4% of the patients. The reason for admission to ICU was acute respiratory failure in 85.5% of the HSCT recipients and 75% had sepsis/septic shock. The mean duration of ICU stay was 104.5 (48–168) hours. Sixty-nine percent of the patients died during their ICU stay while 31% survived. Besides the several statistically significant differences between the patients who survived or died in ICU in univariate analysis, baseline Acute Physiology and Chronic Health Evaluation (APACHE II) score (odds ratio 1.38, 95% confidence interval: 1.06–1.79) and requirement of vasopressors in the ICU (odds ratio 72.29, 95% confidence interval:4.47–1169.91) were found to be independent risk factors for mortality in multivariate analysis.

Conclusion

Baseline APACHE II score and requirement of vasopressors during ICU stay were the most significant independent risk factors for mortality in HSCT recipients who required ICU management in our center.

Keywords:

• Hematopoietic stem cell transplantation
• Intensive care unit
• Mortality
• Risk factors

Introduction

Hematopoietic stem cell transplantation (HSCT) has become the therapeutic choice with curative potential in several benign and malignant hematologic diseases.¹ The outcome of stem cell transplants has improved dramatically with the advances in transplantation technologies in the last two decades leading growing number of patients to have access to HSCT. However, HSCT remains to be a procedure with significant morbidity and mortality and 5–55% of the patients might require admission to intensive care units (ICUs).^2–4 Patients requiring intensive care generally have a worse prognosis, particularly when they experience acute respiratory and/or multiple organ failure (MOF).^5,6

The limited resources in ICUs compel physicians to conduct further studies to establish appropriate patient selection criteria for ICU admission and to define set goals after admission to ICU so as to manage restricted resources properly. Various studies have intended to determine the optimal intensive care management in HSCT recipients the factors affecting their prognosis, and finally to develop prognostic models to stratify patients to risk-adapted treatment modalities.^4,5,7 However, an established and generally accepted prognostic model has not been identified yet. The purpose of the study was to investigate the ICU outcome of the HSCT recipients admitted to ICU and to determine the factors which have an impact on ICU outcome.

Materials and methods

Patients

This study was conducted at the medical ICU of Gazi University Hospital, a tertiary care academic center with 1000 beds. The nine-bed medical ICU is staffed by two full-time intensive care physicians working according to the closed-unit principle. A senior staff physician is present and consultants of different disciplines are available 24 hours a day, 7 days a week. Patients with a potentially favorable long-term prognosis (≥6 months), are admitted to this unit in case of life-threatening medical conditions (severe sepsis/septic shock, acute severe respiratory failure etc.). If our ICU beds are available, we generally accept HSCT recipient who required ICU management. If not, we give support from their clinics till ICU admission

Adult (≥18 years of age) HSCT recipients who required monitoring and inpatient management in medical ICU between 01 January 2007 through 31 December 2010 were included in the study. The first admission data of the patients with more than one admission to the ICU were used in the analysis. Patients who did not survive beyond the first 24 hours following admission to the ICU were excluded from the study.

Data collection

The demographic characteristics such as patient age, sex, comorbidities underlying diseases which required HSCT (diagnosis, diagnosis date, state of disease during transplantation, etc.), transplant-related parameters (transplant date, type, stem cell source, preparation regimen etc.), and early transplant complications were all recorded.

The reason for admission was recorded. The prognosis of ICU patients were estimated by some parameters such as Acute Physiology and Chronic Health Evaluation (APACHE II) score, Glasgow coma scale and Sequential Organ Failure Assessment (SOFA) score. APACHE II score was calculated from the worst physiological parameters and laboratory values of patient within the first 24 hours of ICU admission. The complications including infections which developed during ICU stay were also recorded. Treatment modalities implemented in the ICU (mechanical ventilation, renal replacement therapy (RRT), antibiotics, nutrition, vasopressors etc.) and the outcome of the ICU course (surviving or non-surviving) including the causes of death, were also recorded.

The databases of adult bone marrow transplant unit and ICU and hospital's electronic patient records, patient registration files and unit discharge reports were rewieved retrospectively in order to retrieve the data.

This retrospective observational clinical study has been approved by the local ethics comitte.

Analysis of data and statistical method

SPSS (Statistical Package for Social Sciences) version 16 was used in the analysis of data obtained from the study. Statistical analysis of the patients were initially performed as a whole, and then in two groups categorized as surviving and non-surviving patients. Continuous variables (quantitative variables) are described as mean (±SD) if showed normal distribution. For the variables which did not show normal distribution, median, and interquartile ranges were provided. The categorical variables (qualitative variables) are described as counts (percentages) and was analysed with, χ² test and Fisher exact test – when needed . Student's t test was used to compare two independent groups when the continuous variables provided parametric test conditions and the Mann–Whitney U test was used when continuous variables were non-parametric. After determining the variables that may affect intensive care mortality in univariate analysis, the independent variables were used to adjust the patient charactersistics to the ICU outcome via logistic regression analysis. In all statistical analysis, P < 0.05 was accepted as statistically significant.

Results

A total of 64 HSCT recipients were admitted to our medical ICU during study period. In patients with more than one admission to ICU, the first admission data were included in the study. Patients who died within the first 24 hours following their admission to ICU were excluded from the study. As a result, a total of 48 patients were included to the study.

The median age of the patients was 39 (28–46.75) and 29 (61%) of them being male. Twenty-six (54%) patients had leukemia (acute or chronic), 13 (27%) had lymphoma (Hodgkin or non-Hodgkin), 4 (9%) had multiple myeloma, and finally 5 (11%) had non-malignant hematological diseases (aplastic anemia, polycythemia vera, primary myelofibrosis) as the underlying hematological disease. While 7 (15%) patients had received autologous stem cell transplantation, 41 (85.4%) patients had received allogeneic stem cell transplantation. Four (8.3%) allogeneic HSCT received transplants from an HLA-mismatched donor. The conditioning regimen was myeloablative in 37 (77%) and non-myeloablative in 11 (23%) patients. The disease status was complete remission in 13 (27%) patients while 10 (21%) patients were in partial remission and 3 (6%) patients had stable disease at the time of stem cell transplantation. Ten (21%) patients had relapsed disease and 12 (25%) patients had progressive disease also during stem cell transplantation. The median time to HSCT after diagnosis was 467 (210.5–955.75) days. The median time interval from HSCT to admission to ICU was 140.5 (29–298.25) days. Forty patients had platelet and neutrophil engraftment prior to admission to intensive care. Fourteen patients (29%) had suffered from hepatic sinusoidal obstruction syndrome (SOS), while 28 patients had graft-versus-host disease (GVHD; liver, skin, gastrointestinal system, or combinations thereof).

The most common cause of admission to ICU was acute respiratory failure (86% – 41 patients) and sepsis/septic shock (75% – 36 patients). Other reasons for admission were renal failure, liver failure, and altered mental status, follow-up after cardiopulmonary arrest. Some clinical characteristics and differences in the surviving and non-surviving HSCT patients before and during the ICU admission periods are shown in Table 1.

Table 1. Some clinical characteristics and differences in the surviving and non-surviving HSCT patients before and during the ICU admission periods

During the ICU stay, a total of 41 (85%) patients received mechanical ventilation support. The most common causes of initiating mechanical ventilation was pneumonia, acute respiratory distress syndrome (ARDS), deep septic shock, and acidosis. While invasive mechanical ventilation (IMV) was initiated in 36 patients (75%) (18 patients started to receive IMV before admission to intensive care), 23 patients (48%) received non-invasive mechanical ventilation (NIMV) (10 patients started to received NIMV before admission to intensive care).

While 27 patients (56%) had central venous catheters on admision to ICU, catheters were replaced in the ICU in 18 patients (38%). Central venous pressure was measured in 35 (73%) and invasive arterial blood pressure was monitored in 33 patients (69%). Vasopressors were used in a total of 33 (69%) patients (18 patients started to receive vasopressors before admission to ICU) during the ICU stay.

Forty-seven patients (98%) received antibiotics due to suspected/documented infections before ICU admission. During ICU stay, nine patients (19%) developed pneumonia/ventilator-associated pneumonia (VAP), five patients (10%) developed blood/catheter infection, and two patients (4%) developed urinary tract infection. The most frequently agents isolated were Gram negative bacterial agents and particularly Acinetobacter spp. (in 10 patients). Tigecycline and Colistin were the most commonly added antibiotics to the initial regimen. While severe sepsis/septic shock were observed in 24 patients (50%) on admission to the ICU, 16 patients (33%) developed severe sepsis/septic shock in the ICU. Fourteen patients (29%) required renal replacement therapy (hemodialysis and/or continuous venovenous hemodialysis) in ICU. While blood/blood product replacements were required in 40 patients (83%), 42 patients (88%) received enteral and/or parenteral nutritional supports (Tables 1 and 2).

Table 2. Clinical characteristics of HSCT recipients, who survived or died in the ICU course

The median ICU stay of the patients was 104.5 (48–168) hours. Thirty-three patients (69%) died and 15 patients (31%) survived beyond the ICU course. The most common causes of death were sepsis/septic shock and MOF (62%). While only 1 out of the 15 surviving patients was discharged directly to home from the ICU, 14 patients were transferred to other clinics. Two of the patients who were transferred to clinics died in those clinics, while three of them died following readmission to the ICU. Moreover, hospital mortality rate was 79.2% for HSCT recipients in our study.

Patients were categorized as survivors and non-survivors according to ICU outcomes and the characteristics of the patients were compared accordingly. The possible factors affecting mortality were also investigated. The surviving and the non-surviving patients showed similar characteristics such as sex, age, underlying hematological disease, HSCT type, HLA compatibility, conditioning regimen, and comorbidities. Patients went to HSCT while primary diseases relapsed or progressive, had a higher ICU mortality. Mortality was also higher in patients who had a history of SOS after transplantation (Table 1).

Baseline APACHE II scores, SOFA scores on admission and discharge, number of organ failure on admission and discharge, vital signs on admission, laboratory parameters on admission and discharge were significantly worse in the patients who died in the ICU (Table 2).

The frequency of patients who were already on vasopressors before admission to ICU was similar in the two groups (20% vs. 45%, P = 0.233), whereas the patients requiring vasopressors after admission to ICU was more common in the non-survivor group (20% vs. 90.9%; P < 0.001) (Tables 1 and 2).

Requirement of mechanical ventilation and IMV were both statistically more frequent in the non-surviving patient group (P < 0.001). Mechanical ventilation as a result of septic shock was significantly higher in the non-surviving patient group (P = 0.039). The duration of mechanical ventilation was similar in the two groups (P = 0.119).

Intubation rate was significantly higher (P < 0.001) while the extubation rates were reciprocally low (P = 0.001) in the non-survivor group. Central venous pressure and invasive arterial blood pressure monitoring rates were also significantly higher (P < 0.001) in the non-survivor group.

The interval between HSCT to admission to ICU was 279 days (105–491) vs. 102 days (26–239), the time interval between engraftment and admission to ICU was 174 days (88–452) vs. 146 days (56–245) and the time between admission to the hospital and admission to ICU was 3 days (1–20) vs. 17 days (5–25) in surviving and non-surviving groups, respectively. The time interval from HSCT to ICU admission was significantly longer (P = 0.023) while the time between admission to the hospital and ICU admission was found statistically shorter (P = 0.035) in the surviving patient group compared to the non-surviving group. The time between engraftment and ICU admission was similar in the two groups (P = 0.202).

The presence of infection on admission to ICU was similar in the groups whereas accompanying septic shock was significantly higher in the non-surviving patients (13.3% vs. 51.5%; P = 0.024). While none of the surviving patients developed infection during ICU stay, non-surviving patients developed pneumonia/VAP and septic shock.

Acute renal failure, MOF and septic shock were all significantly higher in the non-surviving patient group (Table 2). Renal replacement therapy was required more often in the non-surviving patient group (0 vs. 42.4%; P = 0.002).

Enteral nutritional support (P < 0.001) was provided more commonly in the surviving patient group, whereas blood/blood products support (P = 0.008) was required more often in the non-surviving group.

The ICU mortality rate of 48 HSCT recpients was calculated as 69%. The most frequent causes of death were noted as septic shock and MOF. The duration of ICU stays was 94 hours (41–144) in the surviving and 114 hours (49–207) in the non-surviving group (P = 0.29).

Logistic regression analysis showed baseline APACHE II score and initiation of vasopressor in the ICU were indicated to be the most statistically significant parameters on mortality (Table 3).

Table 3. Independent risk factors affecting mortality of HSCT patients in ICU according to logistic regression analysis

None of the patients with a history of SOS, who developed septic shock and pneumonia/VAP in the ICU, who required RRT in the ICU or who developed MOF in the ICU survived. Mortality associated with these variables was 100%.

Discussion

HSCT is a highly effective treatment modality with curative potential performed in several benign and malignant hematological diseases. However, it remains to have significant morbidity and mortality particularly when the referral to ICU is inevitable. Effective management of the limited resources of ICUs is mandatory which compels the physicians to develop risk stratified models in ICU admission and management. Underlying hematological diseases and their status at the time of ICU admission, transplantation types, conditioning regimen, additional comorbidities, presence of transplant-related complications such as SOS or GVHD and immunosuppressive agents account for these severe clinical manifestations requiring intensive care support in these patient.^2–10

Requirement for ICU admission is reported to vary between less than 5% to more than 55%.^2–4 A report where data from relevant studies were compiled, requirement of intensive care management was indicated as 15.7% on average.¹¹ The need for intensive care support in our HSCT recipient cohort was 13% (our transplantation unit performed mean 90 HSCT annually, half of them autologous transplantation), and was in the range of previous reports. The diverse rate of ICU admission rate may either be the consequence of varying transplant dynamics (patient selection, the rate of high risk patients) or by different criteria in deciding the transfer of patients to ICU.

Our data show that several characteristics of patients such as age, sex, primary hematological disease, transplantation type and conditioning regimen had no impact on ICU mortality and was in accordance with the study by Afessa et al.^2,11 Yet, in some studies ICU mortality was higher in allogeneic HSCT recipients (15–40% vs. 5–10%).^5,12–14 Contrast to the these reports which indicated the presence of HLA mismatch, GVHD and graft rejection as the causes of increased mortality in allogeneic HSCT recipients who require ICU admission,¹⁵ we did not demonstrate any differences in ICU mortality between allogeneic and autologous HSCT recipients (42.9% vs. 73.2%, P = 0.123). The small sample size, the low number of autologous HSCT recipients, and HLA mismatched patients within the cohort might explain the diverse results. Nevertheless, Scales et al. demonstrated higher mortality in allogeneic HSCT recipients, where 217 of the 543 patients who required intensive care had received autologous HSCT.¹⁶ However, it should be noted that only 4% of our autologous HSCT patients required intensive care during the study period while 23% of the allogeneic HSCT patients required ICU management.

The mortality was higher in patients with relapsed and progressive disease and in patients who had a history of SOS in our study. These results are also in accordance with the previous literature.^4,5,7 However, GVHD was not demonstrated as a prognostic factor associated with higher mortality in our study in contrast to the reports by Pene et al.¹⁴ and Huaringa et al.¹⁷ The small sample size, retrospective nature of the study might be responsible for the discrepancy. The improvement in supportive therapies and modifications in prophylactic anti-microbial agents and implementation of respiratory muscle training and physiotherapy to patients with GVHD in our institution which we started 5 years ago might have decreased the rate of ICU admission in patients with GVHD. However, these variables are beyond the scope of this study.

The dismal prognostic impact of high APACHE II and SOFA scores in our study demonstrated similarities with numerous previous studies.^{2,4,8,12,13,18,19} However, there are also concerns regarding the value of these prognostic models in predicting ICU mortality in this special cohort of patients. Mortality measured through some prognostic models such as APACHE II, Simplified Acute Physiology Score II and Mortality Prediction Model II failed to predict actual mortality in some previous reports.^2,4,8 These generic prognostic models are developed through studies including various patient groups and the different dynamics of the HSCT patients is claimed to requisite prognostic models which are new and specific to this group. Even in a report by Jackson et al.,⁴ where APACHE II score has been considered to be insufficient in estimating ICU mortality in the HSCT recipients, values over 45 were associated with 100% mortality. Afessa et al.,¹⁸ on the other hand, have demonstrated that the mortality measured via APACHE III and the actual mortality were 42 and 46%, respectively. These results show that current ICU prognostic scoring systems, such as APACHE III remains to have an important role in guiding intensive care physicians in the rational use of expensive and limited intensive care support modalities.

The most common cause of admission to the ICU was acute respiratory failure in our cohort, which is consistent with the previous reports.^2,3,8,20,21 Acute lung injury related to pneumonia and sepsis and ARDS present themselves as the most frequent causes of acute hypoxemic respiratory failure.²¹ Requirement of IMV in HSCT recipients on admission to ICU or during their stay in ICU varies between 42 and 88%, respectively.^2,3,5,20,21 The need for IMV was found to be associated with higher mortality in our study, similar to previous reports that indicate more than 80% mortality.^2,3,5,20,21 Timely implementation of NIMV has been demonstrated to decrease mortality in selected immunocompromised patients with respiratory failure.^22,23 In contrast to IMV, the rate of NIMV was similar in surviving and non-surviving patient groups. Randomized clinical trials have demonstrated that requirement for endotracheal intubation and related complications can be reduced via early implementation of NIMV.^22,23 We cannot comment on whether NIMV had an impact on reducing intubation frequency or intubation-related complications and mortality in our cohort due to the retrospective nature of our study. The duration of mechanical ventilation is also among the controversial prognostic variables. While Denardo et al.³ and Torrecilla et al.²⁴ have demonstrated a mechanical ventilation duration of more than one week was associated with 100% mortality, the results of several other studies did not support these results.^4,7,19 We also could not demonstrate an impact of mechanical ventilation duration on the outcome.

The second common cause of admission to the ICU was sepsis and hemodynamic instability associated with sepsis again in line with the existent data.¹¹ Hemodynamic instability may be the result of blood loss, restricted oral intake, secondary hypovolemia, cardiac rhythm disorders, or septic shock. Septic shock presents itself as the most common cause for hemodynamic instability and shock requiring the use of vasopressors in multiple studies as in our study.^4,8,11–13 We found that the need for vasopressors especially in ICU staying was a significant prognostic parameter and demonstrates that septic shock acted as the cause of both admissions to ICU and implementation of mechanical ventilation in our patient population. Similarly, the most common cause of shock, requiring vasopressor support has been demonstrated to be septic shock in this patient cohort with a clear association with increased mortality.^5,25 There was double or multi organ failure in 80% of the patients on admission to the ICU while MOF developed in eight patients (17%) during intensive care management. The rate of MOF is reported to vary from 22 to 81% in the literature and the discrepancies are explained by different organ failure descriptions among centers.¹¹ Development of MOF in the ICU was found to be associated with 100% mortality and is accordance with various similar studies.^{4,5,8,13,14,26}

Development of ARF in the ICU was found to be associated with increased mortality while 100% of the patients receiving RRT died in our series. Similarly, Scales et al. have demonstrated that requirement of hemodialysis in HSCT recipients rendering hemodialysis as one of the two most important independent variable with an impact on mortality.¹⁶

Our results show that pneumonia/VAP and development of septic shock also affected mortality. These results are similar to the study by Naeem et al.²⁷ and Afessa et al.¹¹ in which pneumonia and severe sepsis was associated with increased mortality, respectively.

We also determined that baseline APACHE II score and initiation of vasopressor in the ICU were significant independent risk factors on mortality. In several studies where intensive care monitoring was examined in HSCT patients, early-term mortality (which is called the first 30-day mortality following hospital or ICU discharge) was found between 54 to 96%.^2–13 Long-term mortality, represented the mortality seen over 6 months or a year after ICU admission, was found between 67 to 96%.^19,25,28,29 In our study, we accepted the short-term mortality as ICU mortality with 69%. Hospital mortality rate was also recorded. It was 79.2%. This study displaying almost similar mortality rates with the ones in the literature is of great importance as it is one and only study examining ICU mortality in HSCT patients in Turkey.

The retrospective nature of this study is its major limitation. As we based our assessment on laboratory values within the last 24 hours during the admission to ICU and the initial vital signs recorded by nurse observations during ICU admission, the values at the initial time of transfer decision of these patients to the ICU might have displayed slight differences. Procedures and therapies provided within the transfer period to ICU might have caused far better or worse values in laboratory results and clinical findings of the patient and might be confounding the picture. Retrospective nature of the study, also limits us from examining the effects of intensive care management methods, such as NIMV, and early goal-directed therapy strategies in sepsis on mortality. The small sample size and analyzing the results of a single-center are the other limitations of our study. The number of transplants is increasing dramatically in our country with the introduction of new transplant centers. Despite all the above mentioned limitations, our study is the one and only study in Turkey examining the results of ICU support in this particular patient group.

In conclusion; the presented study summarizes the ICU dynamics of HSCT recipients and demonstrates that a considerable amount of patients survive beyond ICU. Baseline APACHE II score and requirement of vasopressors in the ICU were found to be independent variables indicating mortality in HSCT recipients who required management in ICU. Being a single-center study and its retrospective nature with a small sample size limits us to draw solid conclusions from our current data. Howbeit, we believe the major contribution of our study would be to inspire the researchers in the field to conduct prospective multicenter studies with larger number of patients to determine the optimal timing of ICU admission and also indicate necessary intervention and therapies which will reduce mortality.