Evaluation of factors associated with survival in allogeneic stem cell-transplanted patients admitted to the intensive care unit (ICU)

ABSTRACT Introduction: There are conflicting results concerning the outcome of patients after an allogeneic hematopoietic stem cell transplantation (allo-HSCT) who required treatment in the intensive care unit (ICU). The aim of this study was to evaluate the outcome and prognostic parameters in terms of patient survival after allo-HSCT and admission to the ICU within the first 30 days after transplantation. Methods: Patients after allo-HSCT, who were ≥18 years and admitted to the ICU after the initiation of conditioning therapy and within the first 30 days after allo-HSCT at the University Hospital of Bonn between January 2017 and April 2021, were analysed retrospectively. Baseline data, laboratory parameters, established scoring systems, vital parameters, and outcome were collected. Results: 44 patients (median age of 63 years) were analysed. The 90-day survival rate was 50% (N = 22) and the 1-year survival rate was 27% (N = 12). The 90-day and 1-year survival rates of patients who required MV were 38% (N = 13) and 18% (N = 6). There was a significant correlation between increased mortality and an APACHE-Score ≥20 (p = 0.03), a SAPS-II-Score ≥60 (p = 0.04) and a SOFA-Score ≥9 (p = 0.03). Invasive mechanical ventilation (p = 0.05) and vasopressor support (p = 0.03) showed a negative correlation with the outcome. Conclusion: This study found several parameters (APACHE-II-Score, SAPS-II-Score, SOFA-Score, MV and vasopressor support) associated with increased mortality after allo-HSCT and admission to the ICU. The outcome of allo-HSCT patients admitted to the ICU is not as poor as previously reported. Even older patients under long-term ventilation may benefit from intensive care therapy.


Introduction
The option of an allogeneic haematopoietic stem cell transplantation (allo-HSCT) extends the potentially curative treatment options for malignant and non-malignant haematologic diseases [1].Since 2000, various studies have shown that about 10-20% of all patients receiving an allo-HSCT require intensive care treatment during the peritransplant setting [2,3].In the 1980s/1990s the mortality rate was up to 90%, especially for patients requiring invasive mechanical ventilation (MV) [4][5][6].In the last decades, the outcome of patients with allo-HSCT has improved significantly.This progress is based on enhanced transplant procedures (reduced-intensity conditioning regimes), better antimicrobial, antifungal therapies, and improved intensive care treatment (for example, non-invasive ventilation or lung protective ventilation strategies) [7][8][9][10].
The most frequent cause of admission to the intensive care unit (ICU) in the peritransplant setting is sepsis, followed by respiratory failure in need of noninvasive respiratory supports such as high flow nasal cannula or continuous positive airway pressure (CPAP), MV, RRT, and circulatory instability requiring vasopressors [11].Overall, ICU admission after allo-HSCT is associated with a significant increase in mortality.In the current literature, the one-year survival of allogeneic stem cell-transplanted patients in the ICU varies between 5% and 33% [12][13][14][15].Most studies do not distinguish between very early ICU and late ICU in allo-HSCT.However, these collectives are different due to their chemotherapy side effects, immune suppression, blood count regeneration, and potential GvHD.To date, data on patients treated in the ICU in the peritransplant setting are scarce.
This study provides a retrospective analysis of patients after allo-HSCT who required treatment in the ICU after the initiation of conditioning therapy and within the first 30 days after allo-HSCT treatment at a single University Hospital in Germany.The study focusses on the patient characteristics admitted to the ICU in the peritransplant setting in correlation with the outcome.The aim of this study was to identify parameters that could serve as biomarkers that are associated with the outcome.This could ultimately support clinical decision-making.
Parts of this work were presented as a lecture at the 2022 annual meeting of the German Society of Hematology and Oncology in Vienna [16].

Study population
Data from all patients aged 18 years or older admitted to the ICU in the peritransplant period after allo-HSCT at the University Hospital of Bonn between January 1, 2017, and April 31, 2021, were retrospectively analysed.The peritransplant period was defined from the beginning of conditioning therapy until 30 days after allo-HSCT.For patients admitted to the ICU more than once, only the first stay was considered.
Concerning the diagnosis of sepsis, the definition from 2016 was used [17].

Scoring systems
In the following, the scoring systems are briefly presented and explained.
EBMT: The European Group for Blood and Marrow Transplantation Score assesses the individual risks and opportunities of haemato-oncological stem cell transplantation for patients.The score is made up of 5 factors (age of the patient, stage of the disease, donor type and donor-recipient and time for diagnosis).The lowest risk is 0 point and the highest risk is 7 points [18].
HCT-CI: The Haematopoietic Cell Transplantationspecific Comorbidity Index identifies comorbidities before allogeneic stem cell transplantation to evaluate  [19,20].SOFA: The Sepsis-Related Organ Failure Assessment Score (SOFA) assesses different organ systems (blood, blood pressure, liver, kidney and neurology) to evaluate the mortality risk of patients in the ICU [21].
qSOFA-Score: The quick Sepsis-Related Organ Failure Assessment Score (qSOFA) is the simplified form of the SOFA score and is used for the initial assessment in the emergency room or ICU [22].
PICAT-Score: The prognostic index for intensive care after allogeneic haematopoietic stem cell transplantation (PICAT) evaluates the outcome for these patients by different parameters (time to ICU, LDH, bilirubin, albumin, respiratory failure as the reason for ICU, PT-INR, myeloablative conditioning, age of the patient, HCT-CI) by adding coefficients to calculate the individual risk [23].
SAPS-II-Score: The Simplified Acute Physiology Score is used to assess the physiological health status and general condition of the patient.The following parameters (neurology, blood pressure, age of the patient, chronic diseases, heart rate, body temperature, oxygenation index during ventilation PaO2/FiO2, urine volume and laboratory values (serum urea, leucocytes, potassium, sodium, bilirubin and bicarbonate)) are included in the calculation.The score is calculated from 0 to 8.63 as the highest risk [24].
APACHE-II-Score: The Acute Physiology And Chronic Health Evaluation (APACHE) assesses the severity of the disease and the likelihood of the survival of the patients with different values (body temperature, arterial mean pressure, heart rate, breathing rate, oxygenation, arterial pH, sodium, potassium, bicarbonate, leucocytes, serum creatinine and haematocrit, neurology, chronic Health score, and the age of the patient) [25].

Statistical analyses
Univariate analyses with the Kaplan-Meier method were performed.Factors with a p-value <0.05 were significant.A multivariate regression analysis was not possible due to the limited number of patients.
Patients who survived were censored in April 2022.If patients were lost to follow-up, they were censored on the last day of physician contact.
The baseline characteristics were reported by using median and range.The amount was used by N and percentage.OS analyses were done by Kaplan-Maier and log-rank tests.
Statistical analyses were performed by SPSS Statistics (IBM Corp., Armonk, NY) version 27 for MAC OS.

Patient characteristics
Data of all patients after allo-HSCT, ≥ 18 years, admitted to the ICU, within the first 30 days after transplantation at the University Hospital of Bonn between January 2017 and April 2021, were analysed retrospectively.Baseline characteristics of all patients are summarized in Table 4.A total of 44 (26%) out of 170 patients who received an allo-HSCT between January 2017 and April 2021 required intensive care support in the peritransplant period.
As this study exclusively focuses on the peritransplant period, all patients were aplastic and highly immunosuppressed.There was one patient with serious graft versus host reactions (GvHD) in our cohort.The most common GVHD was mild cutaneous GvHD I-II°(n = 10; 23%).

Survival and early mortality during intensive care
The median duration of treatment in the ICU was 24 days (range 3-105 days).Thirteen patients (30%) died during the first 14 days in the ICU.The 90-day survival rate was 50% (N = 22) and the 1-year survival rate was 27% (N = 12).The median survival of all patients was 88 days (range 4-1347 days).
In the following, attention is focused on the influence of intensive care measures on survival.A total of 34 patients (77%) received invasive MV.In this group, the 90-day survival rate was 38% (N = 13) and the 1-year survival rate was 18% (N = 6).The median survival was 58 days (range 4-1347).A total of 13 of 34 (38%) invasively ventilated patients underwent tracheostomy.Out of this group, 70% (N = 9) were still alive after 90 days and the 1-year survival rate was 23% (N = 3).Ten patients did not require MV.In this group, the 90-day survival rate was 90% (N = 9) and 60% (N = 6) were still alive after one year.The single patient, who received an ECMO, had a survival time of 1174 days and is still alive (ECMO for 16 days and mechanical ventilation for 71 days).33 patients (75%) required RRT.Patients with RRT had a 90-day survival rate of 42% (N = 14) and a 1-year survival rate of 18% (N = 6).The median time of survival was 67 days (range 4-1347).The median duration of RRT was 17 days (Table 6).

Survival functions
The need for mechanical ventilation was associated with a significantly decreased overall survival (p = 0.049) (Figure 1(A)).Renal replacement had no impact on survival (p = 0.059) (Figure 1(B)).Both administration of vasopressors and the number of vasopressors administered were associated with a significant decrease in survival (p = 0.029) (Figure 1(C)) and (p = 0.002) (Figure 1(D)).Furthermore, out of the patients requiring mechanical ventilation, there was a survival benefit in patients, who underwent tracheostomy (p = 0.013) (Figure 1(E)).
The initial GCS (p = 0.6) at the time of admission to the ICU, the qSOFA (p = 0.3) and the PICAT Score (p = 0.2) had no impact on OS.
In addition, remission status before allogeneic stem cell transplantation, donor type, and conditioning chemotherapy did not significantly affect survival.

Discussion
So far, there is only a little data on patients undergoing allo-HSCT admitted to the ICU during the peritransplant period.This study provides a retrospective analysis of a total of 44 patients admitted to the ICU in the peritransplant period in a German single centre.At the University Hospital of Bonn, intensive care procedures are not performed in the transplant unit.
The admission criteria of potential intensive care patients who have received allo-HSCT are difficult to define.Each admission is a case-by-case decision and should be discussed at an early stage with an experienced intensivist and haemato-oncologist.
Our patient population is of special interest due to its increased median age of 63 years and the outstanding number of serious comorbidities (median HCT-CI 3).Due to epidemiological developments, especially in the Western world, the interest concerning the outcome of older patients is currently coming to the fore.Due to the increasing number of older and fit patients, this patient group has become increasingly relevant.Despite considerable progress in the treatment and outcome of patients receiving allogeneic stem cell transplantation in recent decades, the choice of therapy for elderly patients remains controversial and outcomes still are unsatisfactory.Previous studies reported significantly younger patient populations.Lueck et al. reported a median age of 51 years and the oldest published population was by Borrega et al. with a median of 59 years [12,13].
Our data revealed a 90-day survival rate of 50% and a 1-year survival rate of 27%.In contrast, Borrega et al. reported a 90-day survival rate of only 37% and a 1- year survival rate of 16% [23].Lueck et al. reported a 1year survival rate of 14-32%.Recent published studies reported 1-year survival rates between 16% and 33% [12,14,15].Therefore, our data showed a significantly longer 90 days and a longer 1-year survival than the population evaluated by Borrega et al. and is almost in common with Lueck et al.
Our favourable survival data could be because our ICU is part of the Department of Haemato-Oncology and is exclusively run by experienced haemato-oncologists specialized in 'intensive care medicine'.Regarding patients in need of invasive MV, we found a 90-day survival of 38% and a 1-year survival of 18%.Compared to prior studies, the data are promising [4,5,13].
Borrega et al.only reported a 1-year survival of ventilated patients at 5%.This is significantly lower than our 1-year survival of 18%.Unfortunately, the 90-day survival was not calculated in this study.In contrast, Lengliné et al. (median age of the collective 48 years) described a 90-day survival of ventilated post-allo-HSCT patients between 16% and 34%.The 1-year survival was not described [26].Lindgaard et al. (median age = 48 years) reported a 1-year survival of 11% [27].
In our study, the performance of tracheostomy for better weaning or for expected long-term ventilation correlates with an improvement in overall survival (p = 0.013) compared to patients receiving invasive MV without tracheostomy.The data should be interpreted with great caution since a tracheostomy is only performed on a patient with an expected favourable outcome.Our findings are in line with previous data.Tanaka et al. showed better outcomes for patients with coronavirus, who underwent tracheostomy caused by prolonged MV [28].To date, there are no data on this topic in recipients of allo-HSCT admitted to ICU.
We found a significantly worse OS (p = 0.030) when two or more vasopressors were administered.Kew et al. have also shown that vasopressor use is associated with a decreased OS for patients after allo-HSCT admitted to the ICU.Our data are in line with Kew et al., who also reported a significant decrease in OS when vasopressors were administered.
One single patient in our study received extra-corporal membrane oxygenation (ECMO) support.This patient showed a survival of > 1000 days.In general allo-HSCT patients are high-risk ECMO patients; nevertheless, these should be discussed with the ECMO team in a timely manner.
In general, the outcome of patients with cardiac arrest after cardiopulmonary resuscitation is still poor.
In our study, none of the patients in need of CPR survived.This finding is consistent with Borrega et al.They performed 17% CPR for patients after allo-HSCT in the ICU and none was alive after 1 year.
Patients during the peritransplant period are very vulnerable as most patients are aplastic and seriously immunocompromised.Thus, there is a risk of fulminant infection and septic shock.In our population, 36 patients (82%) developed sepsis or septic shock.In this group, the 30-day mortality was 36% and the 90day mortality was 56%.Neumann et al. showed a mortality rate of 56% for septic shock patients after allo-HSCT during the ICU stay [29].Bauer et al. described a 30-day mortality of sepsis patients, who were not immunocompromised, of 27% and a 90-day mortality of 39% [30].
It remains difficult to predict the survival of this particular patient population with scoring systems.The PICAT-Score was developed for this purpose [31].Our data did not show a significant correlation between PICAT-Score and mortality (p = 0.444).Several studies, such as Michel et al., also failed to show a significant benefit from the PICAT-Score [32].The HCT-CI Score is a generally established score to predict survival after HSCT.HCT-CI data for patients in the intensive care unit after allo-HSCT are inconsistent.Our data are in line with Borrega et al. and did not show any correlation between the HCT-CI Score and OS.Bayraktar et al. reported a significant decrease in OS in patients with HCT-CI ≥ 2 [33].In contrast, our data showed that the APACHE-II and SOFA-Scores are more predictive for OS.There was a significant correlation between an APACHE-II-Score ≥ 20 and mortality (p = 0.03) and SOFA score ≥ 9 and mortality (p = 0.03).The data of Michel et al. underline our data, which showed a strong correlation between APACHE-II-Score and mortality during the peri-transplant period [32].Our data also suggest a significant correlation between SAPS-II-Score ≥ 60 and mortality (p = 0.04).This is in line with Turki et al. who also showed a significant correlation between survival and SAPS-II-Score at admission [34].These outcomes underline that these scoring systems can be applied to allogeneic stem cell-transplanted patients in the ICU.
The current study has its limitations as it is a singlecentre retrospective study and contains only a limited number of patients.Furthermore, the increased median age and comorbidities are a limitation in this study.All parameters were collected retrospectively.Further studies should be conducted preferably in a multicentre setting to confirm the results of our study.Patient characteristics, i.e. older age, and adapted transplantation procedures, such as reduced conditioning and alternative donor grafts continue to play an important role in outcomes.For patients who underwent allo-HSCT and who need intensive care support, highly qualified surveillance is required due to the persistently restricted prognosis.Therefore, these patients should only be treated in specialized centres with profound expertise in the field of allo-HSCT.

Conclusion
We found that the outcome of our allo-HSCT patients admitted to the ICU is not as poor as previously reported.Even older patients (median age of 63 years) under long-term ventilation may benefit from extensive intensive care therapy.Sepsis remains a major threat in this setting.SOFA, APACHE-II and SAPS-II have a significant impact on survival in this patient population (very early ICU in allo-HSCT) and can be also safely applied here.

Figure 1 .
Figure 1.(A).Overall survival according to the necessity of MV.The need for mechanical ventilation showed a marginally significant disadvantage in overall survival.(B).Overall survival according to the necessity of RRT.The use of renal replacement showed no significant disadvantage in survival.(C).Overall survival according to the necessity of vasopressors.The use of vasopressors showed a significant disadvantage in survival (p = 0.029).(D).Overall survival according to the number of vasopressors.The number of vasopressors also showed a significant disadvantage on OS. (E).Overall survival according to tracheostomy.The use of a tracheal cannula shows a higher overall survival.

Figure 2 .
Figure 2. (A).Overall survival according to APACHE-Score.There was a significant correlation between an APACHE-Score ≥ 20 and mortality.(B).Overall survival according to SAPS-II-Score.There was a significant correlation between SAPS-II-Score ≥ 60 and mortality.(C).Overall survival according to SOFA-Score.There was a significant correlation between SOFA score ≥ 9 and mortality.

Table 3 .
Laboratory parameters (during the first 24 h after admission).

Table 4 .
Patient characteristics admitted to the ICU.