Prognostic factors for patients with multiple myeloma admitted to the intensive care unit

ABSTRACT

Objectives: This study aimed to analyze clinical characteristics and outcomes of critically ill patients with multiple myeloma (MM) admitted to the intensive care unit (ICU) and identify predictors of poor short-term prognosis.

Methods: Data for patients with MM admitted to the ICU were extracted from the Medical Information Mart for Intensive Care III database. The risk factors leading to the ICU and hospital mortality were evaluated using logistic regression analysis.

Results: Of 126 patients identified, 17 (13.5%) and 37 (29.4%) died in the ICU and hospital, respectively. Patients with ICU mortality showed higher median blood urea nitrogen (57.0 vs. 29.0) and poorer Acute Physiology Scores (APS, 70.0 vs. 46.0) than did surviving patients on the day of ICU admission. In-ICU deceased patients had higher proportion of mechanical ventilation (64.7% vs. 26.6%) and vasopressor use (64.7% vs. 17.4%) at admission and positive pathogenic culture during ICU stay (58.8% vs. 19.3%). The APS and positive pathogenic culture were independent prognostic factors for ICU mortality, while risk factors for hospital mortality included higher APS and relapsed/refractory disease.

Conclusion: The short-term prognoses for patients with MM admitted to the ICU were mainly determined by the severity of organ failure, infection, and disease status.

KEYWORDS:

• Multiple myeloma
• intensive care unit
• mortality
• hospitalization
• acute physiology score
• risk factors
• prognosis
• infections

Introduction

Multiple myeloma (MM) is the most common plasma cell disorder and second most common hematological malignancy worldwide [1]. The clinical manifestations of MM are heterogeneous. Patients can be asymptomatic like smoldering multiple myeloma (SMM), or manifested with serious conditions: (1) End-organ damage, such as anemia, hypercalcemia, kidney damage, and infection; (2) Treatment-related complications, such as bone marrow suppression, thrombotic events and transplant-related diseases; (3) Complicated with high-risk cytogenetics, plasma cell leukemia, refractory or recurrent disease, etc [2]. Thus, a number of MM patients should consider being admitted to intensive care unit (ICU) for further diagnosis and treatment.

Not all critically ill patients can benefit from ICU admission. Previous studies determined that the severity of acute diseases, mechanical ventilation, hematopoietic stem cell transplantation, and neutropenia are risk factors for the survival of patients with hematological malignancies in the ICU [3–6]. It remains unclear whether the prognostic indicators for patients with MM in the ICU differ from those of patients with other hematological malignancies, and it is necessary to determine the categories of MM patients who will benefit from being admitted to the ICU. Studies in this area are limited. Azoulay et al. investigated 75 patients with MM admitted to the ICU at a single French hospital between 1992 and 1998, and found that female sex, mechanical ventilation, and the use of vasopressor agents were negative independent prognostic risk factors, whereas disease remission predicted a better outcome [7]. Hampshire et al. analyzed hospital mortality of hematological malignancies across 178 ICUs in the United Kingdom (UK) over 12 years; while the study included 227 patients with MM, they did not perform a separate analysis of the deceased group to identify prognostic factors [8].

Prognosis prediction for patients with MM admitted to the ICU is of crucial clinical significance, as it can help doctors to better inform patients and their families and assist the rational allocation of medical resources. Therefore, we analyzed the data of all patients with MM who were admitted to the ICU as extracted from the Medical Information Mart for Intensive Care III (MIMIC-III) database to better understand their outcomes during hospitalization and identify short-term prognostic factors.

Materials and methods

MIMIC-III database

All data used in this study were retrieved from the MIMIC-III database, which is a large-scale, open-source registry based in the United States that compiled comprehensive clinical data from 38,645 adults and 7,875 neonates admitted to the ICU in the Beth Israel Deaconess Medical Center, Boston, Massachusetts between 1 June 2001 and 31 October 2012 [9]. All patients were anonymized. As data were derived from a third-party anonymized public database, the study was exempt from requiring approval and consent by the institutional review board of our organization.

Patients

The inclusion criteria were (1) ≥18 years old and (2) patients identified with MM based on the International Classification of Diseases, ninth revision (ICD-9) codes. We excluded patients (1) with missing data related to diagnosis, treatment, hospitalization, discharge records, and other relevant items; (2) with non-plasma cell disorders; and (3) with other plasma cell disorders, such as monoclonal immunoglobulinemia with unknown significance (MGUS), SMM, and primary light chain amyloidosis (AL). Among 150 patients screened from the MIMIC-III database using the ICD-9 codes, two were excluded owing to the lack of diagnosis and treatment information and hospitalization records. Two patients without comorbid plasma cell disorders, six with SMM, one with MGUS, and one with AL were also excluded. Then twelve patients whose reason for admission were considered as monitoring were excluded (Supplementary Table S1). Ultimately, 126 patients with MM were included in this study (Supplemental Figure S1).

Demographics and laboratory variables

For patients with MM being admitted to the ICU multiple times, only data from the first ICU admission were included. The following data pertaining to all included patients were extracted: age, sex, admission type, lowest platelet count, white blood cell count, hemoglobin and albumin, highest blood calcium, creatinine, blood urea nitrogen (BUN), bilirubin, lactate dehydrogenase, Sequential Organ Failure Assessment (SOFA) score, and Acute Physiology Score (APS) [10, 11] on the first day of ICU admission. Moreover, any use of endotracheal intubation, vasopressors, and/or renal replacement therapy were noted, as were the patient’s inhospital time prior to ICU admission, length of ICU and hospital stay, comorbidities, pathogenic culture results [12], and severe sepsis [13, 14]. The primary endpoint of this study was ICU mortality, while the secondary was hospital mortality, which included in-ICU deceased patients and those who survived their ICU stay but still died in the hospital. There were missing data for blood calcium, albumin, bilirubin, and lactate dehydrogenase, and their processing was based upon the methods in literature [15].

The specific treatments received at admission were shown in Supplementary Table S2. Patients’ disease states were classified into the following six groups: (1) newly diagnosed, (2) under the first-line chemotherapy, (3) under transplantation, (4) remission, (5) relapsed or refractory disease and (6) unknown status.

Statistical methods

The R 3.5.3 software (https://www.r-project.org/) was used for data analysis. Numerical variables were described using means ± standard deviations (SDs) or medians (interquartile ranges [IQRs]). Categorical variables are described using frequencies and percentages. T-tests or Mann–Whitney U-tests were performed to compare the clinical characteristics and laboratory indicators of the mortality and survival groups. Univariate and multivariate logistic regression models were used to identify prognostic factors; variables with P-values <0.05 on univariate analysis were included in the regression model, and stepwise regression was performed to obtain the final independent factors. The results are expressed using odds ratio (ORs) and 95% confidence interval (CIs). A two-tailed P-value <0.05 indicated statistical significance.

Results

Patients

A total of 126 MM patients with a mean age of 70.8 years (SD, 11.7 years) were identified within the MIMIC-III database (Supplemental Table S3), among whom 83 (65.9%) were men. Upon ICU admission, 50 patients (39.7%) had thrombocytopenia, 116 (92.1%) anemia, and the median BUN was 32.5 (IQR, 19.3–58.0) mmol/L, which was significantly above the reference levels. The median APS score on the first day of ICU admission was 49.5 (IQR, 39.0–62.0).

The most common indications for ICU admission were shock and respiratory failure accounting for 58 (46.0%) and 43 (34.1%) of the patients, respectively (Supplemental Table S1). Other reasons for ICU admission included disturbed consciousness (11.9%) and renal failure (7.94%). Infection was not only an important reason of admission, but also a main complication observed in ICU. Sixty-seven (53.2%) patients met the Angus’ diagnostic criteria for severe sepsis, and 31 (24.6%) tested positive on pathogenic cultures (Supplemental Table S3).

Seventeen patients (13.5%) died in the ICU, while 37 (29.4%) died in the hospital. The median inhospital time prior to ICU admission was 0.05 (IQR, 0.00–2.57) days, while the median length of ICU and hospital stay were 2.31 (IQR, 1.42–5.01) days and 10.3 (IQR, 5.81–19.3) days respectively (Supplemental Table S3).

Comparison between patients who died in the ICU and survivors

There were no significant differences between patients who died in the ICU and those who survived regarding age, sex, ethnicity, and disease states. Compared to surviving patients, those who died stayed longer in the unit (4.38 vs. 2.22), and had significantly higher median BUN (57.0 vs. 29.0), proportion of mechanical ventilation (64.7% vs. 26.6%), and vasopressor use (64.7% vs. 17.4%) on the day of ICU admission. Scores for critical illness prognosis were significantly elevated in the deceased group. Infection was also closely associated with the ICU mortality rate. Patients with severe sepsis (35.3% vs. 13.8%, OR 3.42) and positive pathogenic culture had significantly poorer prognoses than the other patients (58.8% vs. 19.3%, OR 5.99).

Factors associated with hospital mortality included platelet count on the first day of ICU admission; SOFA and APS scores; severe sepsis; and disease status (Supplemental Table S4). In addition, inhospital deceased patients had a longer inhospital time prior to ICU admission, and longer length of ICU and hospital stay.

Analysis of factors related to patient prognosis

Logistic regression analysis revealed that the APS score (AOR 1.05) as well as positive pathogenic culture (AOR 9.26) were independent prognostic factors for ICU mortality among patients with MM admitted to the ICU (Supplemental Table S5). While relapsed or refractory disease (AOR 11.4) and the APS score (AOR 1.05) were independent prognostic factors of hospital mortality. According to the logistic regression model of ICU mortality, we obtained the formula to predict the death of MM patients in ICU: P (Y = ICU mortality) = $(1/1+{\mathrm{e}}^{\text{-}5.95+0.05\ast \mathrm{A}\mathrm{P}\mathrm{S}\mathrm{I}\mathrm{I}\mathrm{I}+1.96\ast \mathrm{p}\mathrm{a}\mathrm{t}\mathrm{h}\mathrm{o}\mathrm{g}\mathrm{e}\mathrm{n}\mathrm{i}\mathrm{c}\mathrm{c}\mathrm{u}\mathrm{l}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{u}\mathrm{l}\mathrm{t}})$, and the value of pathogenic culture result was 0 or 1.

Finally, we compared the causes of death between in-ICU deceased patients and ICU survivors who died in the hospital (Supplemental Table S6). The median time from ICU discharge to death was 20.39 (IQR, 25.50) days. Severe infection was the most common reason of death for both groups. Compared with patients who died in the ICU, in-hospital deceased patients had a higher proportion of deaths due to late-stage disease or complications such as cachexia, amyloidosis, and extramedullary infiltration (33.3% vs. 17.7%), but lower proportion of deaths due to respiratory and circulatory failure (14.3% vs. 29.4%). The most common treatments received in admission were PIs and IMiDs. In-hospital deceased patients had a higher proportion of transplantation (29.4% vs. 14.3%).

Discussion

Here, we evaluated clinical characteristics and outcomes of critically ill patients with MM admitted to the ICU and we identified predictors of poor short-term prognosis. Overall, 17 and 37 patients in our study died in the ICU and hospital, respectively. There were significant differences in the BUN, SOFA and APS scores, and infection between patients who died in ICU and their surviving counterparts. Among these factors, the APS score and positive pathogenic culture were independent predictors of ICU mortality. Conversely, the independent predictors of hospital mortality were APS score and disease status, whereby patients with relapsed or refractory disease showed poorer hospital outcomes.

Our findings indicated that the ICU and hospital mortality rates of patients with MM admitted to the ICU (13.5% and 29.4%, respectively) were lower than those reported in previous studies. A multi-center study in the UK found that the ICU and hospital mortality rates of patients with MM admitted to the ICU were 38.0% and 60.1%, respectively [8]. Azoulay et al. found that the overall and 30-day ICU mortality rates of such patients were 49.33% and 57.23%, respectively [7]. On the one hand, these differences could be attributed to the different study groups, study lengths, admission and discharge criteria, implementation of end-of-life decisions, and treatment methods. On the other hand, the difference could also be related to the different clinical characteristics of the patients. In our study, only 31.7% of the patients required endotracheal intubation, which was a markedly lower rate than that in the UK study (52.8%). Moreover, the number of patients who experienced complications and organ failure in our study was relatively low, which was reflected in the lower SOFA scores among our patients than in those of the two aforementioned studies.

The APS III score collected in the MIMIC-III database is an important component of the Acute Pathologic and Chronic Health Evaluation (APACHE) Ⅲ scoring system, which is mainly used to evaluate the severity of organ failure. Previous studies showed that the APACHE II was an independent prognostic predictor in critically ill patients with hematological malignancies [8, 16]. Our data also indicated that the APS could reflect the short-term prognosis of patients with MM who were admitted to the ICU. However, other studies have also suggested that scoring systems like APACHE II, SOFA did not perform well in this particular patient group, as they overestimated the mortality rate of survivors and underestimated that of non-survivors [17, 18]. This implies that the scoring systems should not be used alone when assessing the conditions of patients with MM or making treatment decisions, and that other factors that predict ICU outcomes should also be considered.

Patients with cancer, especially those with hematological malignancies were more prone to severe infection compared to patients with non-malignant diseases [19]. This is related to the fact that such patients often have neutropenia, or malignant transformation of immune cells and loss of normal immune function. Multiple factors contribute to the immunodeficiency of MM: (1) Disease factors include hypogammaglobulinemia, impaired functions of lymphocytes, NK cells and dendritic cells, and neutropenia caused by bone marrow infiltration; (2) Treatment factors include steroid-related immunosuppression, Neutropenia secondary to chemotherapy and mucosal barrier damage; (3) Host factors include age, comorbidities and inability [20]. Infection is clearly associated with poor prognosis for ICU patients [21]. Van Beers et al. found that positive pathogenic cultures upon ICU admission were associated with a higher risk of 28-day mortality after ICU admission [22]. Our study showed that a positive pathogenic culture after ICU admission was associated with a higher ICU mortality rate among patients with MM, thus suggesting that it is necessary to actively seek evidence of infection in such patients when admitted to the ICU. Early identification and treatment of sepsis may improve their prognoses.

Studies have found that the short-term prognosis in patients with malignant tumors admitted to the ICU was mainly determined by the severity of the acute disease and not the characteristics of the primary disease [23]. Nevertheless, a recent study of patients with lymphoma admitted to the ICU found that the post-ICU survival of relapsed/refractory patients was worse than that in patients in remission, as their median survival was only 2.9 months [24]. Our findings suggest that the disease state of patients with MM at the time of ICU admission cannot predict their ICU survival but is closely associated with their hospital survival. This implies that ICU survivors who are not in remission may still have a poor short-term prognosis. Patients with relapsed or uncontrolled disease who are referred to the ICU may require alternative treatments beyond the ICU.

Some previous studies have shown that endotracheal intubation and vasopressors were independent prognostic risk factors for patients with hematological malignancies admitted to the ICU [3, 6]. However, we found that this was not the case for patients with MM, which may be because these two indicators neither fully reflect patient’s various organ functions nor accurately assess the severity of the acute disease. Previous studies also suggest that leukopenia, especially neutropenia, was an independent risk factor for death [3, 6], but this was not supported by our data. This may have been because the mechanisms of immunodeficiency and intensive treatment regimen for MM differs from that for leukemia. Previous studies have suggested that inhospital time prior to ICU admission was an independent factor affecting the prognosis of patients in ICU [3,8]. In this population of MM, the median inhospital time prior to ICU admission of patients who died in hospital was longer than that of patients who survived; however, this factor did not show an independent effect in multivariate analysis. This may be related to different study populations and sample size. Future studies with larger sample are needed.

This study has the following limitations: Firstly, it was a retrospective, single-center observational study, which limited the applicability of our findings to other patient populations. Second, owing to constraints in the purpose of our study, we did not analyze the long-term prognoses of patients admitted to the ICU. Last but not least, due to the lack of relevant data in the MIMIC-III database, this study did not include the line of therapy that patients received and some important disease risk factors such as ISS, cytogenetics, plasma cell leukemia, primary refractory disease, etc. for analysis.

In summary, we recommend that the clinical conditions of patients with MM admitted to the ICU ought to be stringently assessed, as should the reversibility of the primary disease. Upon ICU admission, it is necessary to actively screen for potential infections and correct for any organ failure while also actively communicating with the hematologists to jointly determine the appropriate treatment strategy.

Acknowledgments

We would like to thank Editage (www.editage.cn) for English language editing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.