A considerable improvement in outcomes is seen in critically ill cancer patients because of recent advances in the medical and surgical treatment of cancer patients^1,2). This improvement in prognosis leads to an increased number of patients living with cancer causing an increasing number of patients requiring hospitalizations and intensive care unit (ICU) admission^3,4). Admission of critically ill patients to an ICU requires vast professional, technological, and costly resources. However, at present, available ICU beds are scarce, especially in developing countries, it is important to predict cancer patients who may benefit from ICU treatment^5,6).

Brain tumor patients may require ICU admission due to raised intracranial pressure, developing brain edema, epilepsy or complications because of tumor therapy⁷⁾. A lot of studies are available on the clinical course of patients with solid tumors and hematological malignancies admitted to an ICU, but there is a shortage of data on primary brain tumor patients⁸⁾. Most of the studies in critically ill brain tumor patients are done in developed countries, and we didn’t find any study about clinical characteristics, outcomes, and factors associated with increased risk of mortality in primarily brain tumor patients admitted in to ICU from developing countries.

Therefore, we aimed to assess the clinical course and 30-day mortality with factors affecting the mortality of brain tumor patients who were admitted to the medical ICU. As far as we know, this is the first study to evaluate primarily the brain tumor population admitted in to medical ICU in India.

This study was a single-centre retrospective observational cohort study and was conducted in a medical ICU of a tertiary care center in India. A consent waiver was collected from the institutional ethics committee. The data were collected and analyzed from the records of adult brain tumor patients who were admitted in to the medical oncology ICU from January 2018 to December 2021.

We evaluated all adult patients (>18 years) with a definite diagnosis of brain tumor at ICU admission. If the patient was admitted multiple times to the ICU during his/her hospital stay, only the first admission was included in the study. Patients who had ICU stay of less than 24 h, post-operative neurosurgery patients, those in complete cancer remission for more than 5years, and those admitted from or discharged to another ICU, were excluded from the study. Data regarding demographics, baseline characteristics, clinical and laboratory data including admission of patients from the emergency room (ER) or ward, tumor characteristics, reasons for ICU admission, need for vasopressor, invasive mechanical ventilation (IMV) support and renal replacement therapy, and 30-day mortality were collected in a pre-designed proforma.

Sequential Organ Failure Assessment (SOFA) score, SAPS II (Simplified Acute Physiology Score II), and Acute physiology and chronic health evaluation II (APACHE II), APACHE III, and APACHE IV scores were calculated from the data obtained on the day of ICU admission. The data, required to calculate various scores, was collected.

Data were analyzed from 42 patients who fulfilled the inclusion criteria. In our study, 22 patients (52.4%) were admitted from ER and 20 patients (47.6%) were shifted from ward. There was no statistically significant difference among survivors and nonsurvivors according to the type of admission (p = 0.644) (Fig. 1). The two most common indications for ICU admission were altered mental status (57.1%) and sepsis (23.8%) as given in Fig. 2. There was a statistically significant difference in reasons for ICU admission between survivors and nonsurvivors (P value = 0.004). Other causes of ICU admission were respiratory distress, gastrointestinal bleeding, cardiac arrest survivor, acute kidney injury.

As shown in table 1, mean age (SD) of study group was 58.52 (13.06) years and out of them, 50% were females. Comorbidities such as hypertension and diabetes mellitus were present in 20 patients (47.6%). Metastasis from brain to other parts of the central nervous system, such as spinal cord was seen in 5 patients (11.9%). The mean time for ICU admission after diagnosis of malignancy was 26.62 ± 31.48 months and 36 (85.7%) patients were having a history of tumor resection. Mean GCS, serum sodium, serum calcium, arterial lactate, and procalcitonin were 11.62 (3.52), 132.4 (6.81), 8.32 (0.88), 2.53 (3.22) and 4.88 (16.85) respectively. Severity indices were calculated upon hospital admission and found that mean APACHE II, III, IV, SAPS II, and SOFA scores were 18.1 ± 7.1, 23.35 ± 23.86, 58.88 ± 26.87, 33 ± 14.33, and 6.1 ± 2.97 respectively. The mean duration of ICU stay was 7.5 days and hospital stay was 14.71 days. Overall, the 30-day mortality rate was 30.95%. Out of a total of 42 patients, 31% of patients required IMV support and 21.4% of patients received vasopressor support. No patient required renal replacement therapy.

Evidence of infection was found in (36 patients) 85% of patients. out of them most common infection was found to be urinary tract infection (80%), followed by bloodstream infection in 13 patients (30.9%) and respiratory tract infection in 8 patients (19%). 14.28% of patients had fungal infections (mainly candida infection). The most common organism was found to be gram-negative bacteria in the urinary and respiratory tract. Blood stream infection was mostly caused by gram-positive bacteria. 38 % of patients with urinary tract infections had mix gram-positive bacteria, gram-negative bacteria, and fungal infections.

Univariate analysis showed that factors associated with higher 30-day mortality were age, higher lactate, procalcitonin, and need for IMV support or vasopressor support and higher APACHE II, III, IV, SAPS II, SOFA SCORE during ICU stay as seen in table 2. Out of them only 3 factors, Age and need for IMV support or vasopressor support, were found to be directly associated with 30-day mortality in multivariate analysis (Table 3).

This is the first descriptive analysis of clinical course and outcomes in brain tumor patients admitted to a medical ICU in India and compared various scoring systems to guide decision-making. Our 30-day mortality rate was 30.95%. It corresponded to the ICU mortality rate in cancer patients and hospital mortality rate in patients with non-oncological diseases as described by previous studies^9,10). It is lower than studies done on primary brain tumors by Maxen et al. (73%), Bernhard et al. (60%)^7,11). It is higher than studies on PBT done by Kang et al. (20%), and 2 French studies (22–23%)^10,12,13). This might be because the patients with primary brain tumors in previously published cohorts were less severely ill than patients in our cohort as they received less vasopressor support and had lower apache scores and lower hospital stay, in contrast to our study.

A most common reason for ICU admission in brain tumor patients was altered mental status similar to Kang et al. Altered mental status can be because of developing brain edema, seizures, encephalopathy, or complications derived from the tumor therapy. New-onset seizures can be the first clinical symptom of a brain tumor¹⁴⁾. Sepsis was second most common cause of ICU admission in our study cohort. Sepsis is a major cause of ICU admission that also affects the outcome of the cancer patient. Cancer patients are particularly at risk for severe sepsis than the general population because of immunosuppression caused by the malignancy itself or its treatment, organ involvement by malignant cells, infections from medical devices, such as catheters, nutritional deficiencies, or adverse effects of cancer therapy¹⁵⁾.

The mean age of our study cohort was higher in non-survivor group in our study cohort. This may be because of higher comorbid conditions and higher metastasis rates in elderly patients. Previous studies also showed that comorbidities metastasis and increased age were associated with cancer death^16,17).

Mean lactate and procalcitonin were higher in the non-survivor in our study cohort. Lactate levels can be estimated via point-of-care testing at the bedside. High lactate level is a guide to poor tissue perfusion and organ dysfunction in critically ill patients. However, it is a non-specific biomarker and may be raised in other clinical conditions also. Previous studies showed that high lactate is associated with increased mortality in patients with sepsis and it should be used as a biomarker for risk stratification in suspected sepsis^18-21).

Serum procalcitonin is a biomarker for the early diagnosis of sepsis. It helps monitor the antimicrobial treatment regimen and can be used as a tool for antimicrobial stewardship. However, its level can also be increased in non-infectious processes such as tumorous diseases like medullary thyroid cancer, lung cancer with neuroendocrine component, or liver metastases. Previous brain tumor studies didn’t compare lactate and procalcitonin levels in cancer patients^22-24).

ICU admission is usually expensive, so prediction of patients who may benefit from ICU treatment is paramount; especially in developing countries. Intensivists use their clinical judgment or various scoring systems to triage cancer patients for ICU admission. However, clinical judgment is often inaccurate to triage cancer patients for ICU admission. Information derivable from mortality-predicting tools may help guide physician in evidence-based decision-making along with their clinical judgement. So we calculated and compared various severity scores and found that severity scores were significantly higher in non-survivors.

The requirement of invasive mechanical ventilator support and vasopressor support was significantly higher in non-survivors and directly correlated to mortality similar to previous studies^25-27). Infectious complications are common in a cancer patient because of their immunocompromised state and neutropenia. Gram-negative bacteria are a common cause of infection and sepsis in patients with cancer similar to our studies^28-30).

Our study has several limitations. Because of the retrospective design, there can be data collection bias or patient selection bias. However, our data were extracted from a prospectively managed database, which warrants a certain degree of reliability and it is not feasible to conduct prospective studies in this group of patients because of the rarity of this disease. Our study population was small due to strict inclusion criteria. It is a single-centre study. The collective experience of intensivists from multicenters and specialized ICUs in dedicated brain tumor centers at various places can contribute to favorable outcomes. Because of the retrospective design and the single-center scope which may limit generalizability.

However, our study has some major advantages. We evaluated a homogeneous group of patients with brain tumors admitted in to the medical ICU. We also evaluated the effect of lactate and procalcitonin on survival. We also compared multiple different scoring systems between survivors and nonsurvivors to predict mortality in brain tumor patients.

To conclude, advancing age and the need for IMV or vasopressor support may be associated with worse prognosis in brain tumor patients admitted to ICU. A scoring system could be used along with clinical judgement to triage brain tumor patients for ICU admission. A larger retrospective or prospective study is needed to validate and explore potential predictors of survival.