Ретроспективное когортное исследование по вопросу о смертности от щелочной фосфаты и СИС при множественной миеломе.

Introduction

Multiple myeloma (MM), a heterogeneous hematological malignancy, is characterized by clonal plasma cell proliferation and distinct bone microenvironment dysregulation, where osteoclast activation dominates over suppressed osteoblast function 1 – 5 . In 2020, MM accounted for 176,404 cases (14%) among 1,278,362 diagnoses of leukemia, lymphoma, and multiple myeloma globally. Its incidence has risen significantly, particularly in males, individuals aged ≥ 50 years, and populations in high-income countries 6 , MM imposes substantial burdens on both patients and healthcare systems. A comprehensive understanding of MM and its associated factors is critical for disease prevention and management, ultimately improving prognosis and quality of life in affected individuals.

Alkaline phosphatase (ALP), a key serum biomarker, provides critical insights into physiological and pathological states 7 . In MM, extremely low ALP levels may indicate severe osteoblast suppression and advanced disease, while moderate elevations could reflect compensatory bone remodeling 8 . ALP’s association with hepatic/renal dysfunction and immunomodulation may exacerbate organ failure in critically ill MM patients 9 , 10 . Notably, studies in chronic kidney disease have demonstrated a U-shaped relationship between ALP and mortality 11 , 12 , suggesting threshold-dependent biological effects.

MM is characterized by osteolytic lesions resulting from osteoclast activation and osteoblast inhibition, leading to pathologic fractures and hypercalcemia 13 . Unlike solid tumors where ALP elevation typically signals bone repair, ALP suppression in MM correlates with osteoblast dysfunction—a hallmark of myeloma bone disease 14 . This paradoxical behavior positions ALP as a unique prognostic marker in MM, reflecting both tumor burden and skeletal complications 14 .

ICU admission in MM patients often signifies acute deterioration due to sepsis, renal failure, or hyperviscosity syndrome, with mortality rates remain high 15 . The role of ALP in this critical context remains unclear. ALP not only serves as a routine health assessment parameter but also functions as a monitoring indicator for MM patients. Deviations from established reference ranges may signal underlying pathological imbalances necessitating clinical intervention.

Currently, insufficient evidence exists regarding the relationship between ALP and ICU mortality. This retrospective cohort study of 161 MM patients in the MIMIC database aims to investigate the potential nonlinear association between ALP levels and ICU mortality, thereby advancing understanding in this field.

Materials and methods

Mimic-IV database and participants

This was a retrospective study based on the MIMIC-IV database (version:2.2) 16 , which contains data on more than 40,000 patients admitted to the ICU at Beth Israel Deaconess Medical Center in Boston, Massachusetts, from 2008 to 2019. Access to the database was granted to certified researchers 17 (Qian-hui Wang’s certification number is 52,968,321), and ethical approval was waived due to the anonymized nature of the data 18 .

Patients who meet the criteria are screened in the MIMIC-IV database. The inclusion criteria were as follows: Adult patients (≥ 18 years) with a primary diagnosis of multiple myeloma (MM), defined by ICD-9 codes 20,300, 20,301, 20,302 or ICD-10 codes C9000, C9001, C9002. The exclusion criteria were as follows: Among the 2857 patients initially extracted from the MIMIC-IV database, Missing ICU admission records (n = 2674); Multiple ICU admissions (n = 20; only the first admission was retained to avoid confounding from repeated measurements); Missing anion gap values (n = 1); Duplicate records (n = 1). Regarding the handling of missing data, cases with missing values were directly excluded. The final cohort included 161 MM patients (Fig. 1 ). All information can be extracted from the corresponding tables in the MIMIC database 19 . Fig. 1 Flowchart of study population.

Covariates

The exposure variable in this study was ALP, with a normal level of approximately 35–130 IU/L 20 . ALP value measured within 24 h of ICU admission, categorized into tertiles (22–62 IU/L, 63–99 IU/L, 100–659 IU/L) based on clinical relevance and distribution. The outcome variable for this study was in-hospital mortality after ICU admission. Covariates in the database adjusted for demographics (age, sex, ethnicity), vital signs (highest heart rate, respiratory rate, systolic/diastolic blood pressure, temperature, lowest SpO2), laboratory parameters (lowest hematocrit, hemoglobin, platelet count; highest white blood cell count, anion gap, sodium, creatinine), and severity scores (SAPS II, OASIS). SAPS II and OASIS were included to comprehensively assess organ dysfunction and acute illness severity 21 , 22 .

Statistical methods

The descriptive data are presented as median and quartiles. Group comparisons were performed using the Mann–Whitney U test (non-normal data) or independent t-test (normal data) for continuous variables, and chi-square or Fisher’s exact test for categorical variables. The continuous variables with normal distribution and homogeneity are presented as Mean ± SD . The continuous variables with non-normal distribution are expressed as median with interquartile range. Categorical variables are expressed as frequencies (%). And the results are presented as mortality odds ratio (OR) with 95% confidence intervals (CI). A P value < 0.05 was considered to indicate a statistically significant difference.

The tertile categorization of ALP was based on two main considerations. Firstly, from a statistical perspective, the tertile division aimed to minimize within—group variance while maximizing between—group discrimination, as confirmed via analysis of variance. Secondly, we explored alternative categorization methods, such as quintiles. However, the quintile categorization led to groups with relatively small sample sizes in some intervals, which could increase the statistical uncertainty and reduce the power of the analysis. Moreover, the clinically meaningful thresholds for ALP in the context of ICU in—hospital mortality were not well—established, and the tertile division provided a balanced approach that combined clinical insights with statistical considerations. We ultimately selected a tertile 23 ALP (22–62 IU/L, 63–99 IU/L, 100–659 IU/L) and divided the patients into three groups.

To explore the relationship between ALP and in-hospital mortality after ICU admission, the stability of the relationship was determined through Cox univariate and multivariate regression analysis, curve fitting, and inflection point analysis. Univariate analysis: Initial screening of covariates associated with ICU mortality ( P < 0.05). Multivariate models:Model I: Adjusted for age and race (minimally adjusted model). Model II: Based on clinical experience and adjusted for covariates with > 10% change in effect size upon inclusion (age, race, heart rate [HR], respiratory rate [RR], temperature, SpO2, platelet count [PLT], white blood cell count [WBC], sodium, and SAPS II). In addition, restricted cubic spline (RCS) regression was performed with 4 knots at the 5th, 35th, 65th, and 95th percentiles of ALP to assess linearity and examine the dose–response curve between ALP and ICU in—hospital mortality after adjusting variables in Model 2 (Fig. 2 ). We used a three-piece-wise logistic regression model with smoothing to analyze the association threshold between ALP and ICU in—hospital mortality after adjusting the variables in Model 2. The likelihood-ratio test method was used to determine inflection points (Table 4 ). Furthermore, potential modifications of the relationship between ALP and ICU in—hospital mortality were assessed, including the following variables: sex, age (< 60 vs. ≥ 60 years).Heterogeneity among subgroups was assessed by multivariate logistic regression, and interactions between subgroups and ALP were examined by likelihood ratio testing (Fig. 3 ). Results were presented as hazard ratios (HR) and 95% confidence intervals (CIs). Fig. 2 The nonlinear relationship between ALP and ICU in-hospital mortality in Multiple Myeloma in ICU. Adjusted for all covariates as model 2 (ALP takes the upper limit of 95.6%). Fig. 3 Forest plots illustrate the hazard ratios (HRs) for in-hospital mortality across ALP groups using multiple models. For each stratification factor, adjustments were made for all other covariates (including age, race, heart rate, respiratory rate, platelet count, white blood cell count, SAPS II, and OASIS scores), in addition to the stratification component itself.

Given the relatively small sample size (n = 161) in our regression modeling, we took several measures to mitigate the risk of overfitting. Firstly, we carefully selected the covariates for adjustment based on prior clinical knowledge and the strength of their association with the outcome in univariate analysis (Wald’s test and LR-test). Secondly, we adjusted for covariates with > 10% change in effect size upon inclusion.

All statistical analyses were conducted using R Statistical Software (Version 4.2.2, http://www.R-project.org , The R Foundation) and Free Statistics Analysis Platform (Version 1.9, Beijing, China, http://www.clinicalscientists.cn/freestatistics) 24 .

Results

Finally, a total of 161 patients with MM admitted to the ICU were identified in the MIMIC-IV database (Fig. 1 ) according to the inclusion exclusion criteria. Table 1 summarizes the basic demographic characteristics of participants stratified by ALP levels. The mean age of the participants was 71.0 ± 11.2 years, 101 (62.7%) were males and 60 (37.3%) were females. ALP levels were positively associated with lowest SPO2, lowest HCT. The mean SAPSII score on the first day of ICU admission was 50.2 ± 14.1, and the mean Oasis score was 33.6 ± 9.0. Table 1 Baseline demographic characteristics of the study population stratified by ALP. Variables Total (n = 161) ALP IU/L p 22–62 (n = 54) 63–99 (n = 53) 100–659 (n = 54) Gender, n (%) 0.768 Female 60 (37.3) 20 (37) 18 (34) 22 (40.7) Male 101 (62.7) 34 (63) 35 (66) 32 (59.3) Age, y 71.0 ± 11.2 71.1 ± 10.8 70.0 ± 11.3 71.9 ± 11.7 0.688 Race, n (%) 0.9 White 102 (63.4) 33 (61.1) 32 (60.4) 37 (68.5) Black 30 (18.6) 11 (20.4) 11 (20.8) 8 (14.8) Other 29 (18.0) 10 (18.5) 10 (18.9) 9 (16.7) Highest.HR, 108.2 ± 19.9 105.8 ± 20.9 107.2 ± 16.8 111.6 ± 21.5 0.287 Highest SBP, 141.1 ± 19.8 141.5 ± 19.8 142.8 ± 21.6 139.2 ± 18.1 0.632 Highest DBP, 87.9 ± 16.3 88.3 ± 17.8 86.8 ± 14.8 88.5 ± 16.3 0.837 Highest RR, 30.3 ± 7.1 29.0 ± 6.7 30.7 ± 7.3 31.1 ± 7.2 0.283 Highest temperature, 37.4 ± 0.7 37.5 ± 0.8 37.4 ± 0.8 37.3 ± 0.6 0.339 Lowest.SPO2, 89.8 ± 11.3 90.1 ± 12.1 90.4 ± 7.9 88.8 ± 13.4 0.731 Lowest HCT, 26.1 ± 5.8 24.2 ± 4.6 26.9 ± 6.3 27.2 ± 5.9 0.012 Lowest HB, 8.6 ± 1.9 8.0 ± 1.5 8.8 ± 2.0 8.9 ± 2.0 0.025 Lowest PLT 124.0 (60.0, 179.0) 118.0 (55.8, 171.8) 133.0 (68.0, 179.0) 129.5 (53.8, 180.8) 0.623 Highest WBC 7.8(4.0,12.1) 6.2(3.5,11.4) 8.0(6.1,12.3) 8.1(3.9,12.6) 0.265 Highest AG, 17.4 ± 5.6 17.0 ± 6.7 17.0 ± 4.8 18.1 ± 5.2 0.484 Lowest sodium, 136.2 ± 6.8 134.9 ± 9.0 136.5 ± 5.2 137.2 ± 5.4 0.187 ALP 80.0 (56.0, 111.0) 47.5 (36.0, 55.5) 80.0 (72.0, 85.0) 137.0(111.8, 197.2) < 0.001 Highest Cre 1.6 (1.0, 3.1) 1.6 (0.9, 3.2) 1.8 (1.1, 2.8) 1.6 (1.0, 4.2) 0.765 SAPS II, 50.2 ± 14.1 50.6 ± 16.1 48.4 ± 14.1 51.7 ± 12.0 0.478 OASIS, 33.6 ± 9.0 32.4 ± 9.3 32.5 ± 9.2 35.9 ± 8.1 0.069 Data presented are mean ± SD, median (IQR), or N (%). SD: standard deviation; IQR: interquartile range; HR: heart rate; SBP: systolic blood pressure; DBP: diastolic blood pressure; RR: Respiratory Rate; HCT: hematocrit; HB: hemoglobin; PLT: Platelet; WBC: white blood cell; AG: Anion gap; ALP: alkaline phosphatase Cre: creatinine; SAPS II: Simplified Acute Physiology Score II; OASIS: Oxford acute severity of illness score.

Table 2 presents the univariate analyses of risk factors related to ICU in-hospital mortality in patients with severe MM, reports as HR and 95% CI. Univariate analysis showed that Age, HR, RR, SPO2, HCT, HB, AG, SAPSII and Oasis score were significantly associated with ICU in-hospital mortality ( P < 0.05). Conversely, factors such as gender, race, platelets, WBC, sodium and other laboratory results, were not significantly associated with ICU in-hospital mortality (Table 2 ). In Cox multivariate regression analysis (Table 3 ), the relationship between ALP and mortality was not significant when it was used as a continuous variable. However, when considered as a categorical variable, tertiles of ALP (100–659 IU/L) was used as a baseline reference, and lowest ALP (22–62 IU/L category compared with 63–99 IU/L category) was associated with a lower risk of ICU in-hospital mortality (crude HR 0.32; 95% CI 0.11–0.95; P = 0.039). After adjusting for the potential confounders in Table 2 (model 2), this association is weakened. It was still statistically significant (HR 0.32, 95% CI 0.11–0.95 , P = 0.039). In several subgroups, stratified analyses were performed to assess the potential impact of the relationship between the serum ALP and ICU mortality. No significant interactions were found by gender and age subgroups (Fig. 3 ). Table 2 Association of covariates and in-hospital morality in patients with MM. Item HR (95% CI) P value (Wald’s test) P value (LR-test) Gender: Male vs Female 0.63 (0.3, 1.3) 0.21 0.217 Age 1.04 (1, 1.07) 0.045 0.038 Race: ref. = White 0.589 Black 1.08 (0.4, 2.95) 0.877 Other 1.55 (0.69, 3.49) 0.293 Highest HR 1.02 (1, 1.04) 0.025 0.024 Highest SBP 0.98 (0.96, 1) 0.099 0.09 Highest DBP 0.9926 (0.9714, 1.0143) 0.501 0.497 Highest RR 1.05 (1.01, 1.09) 0.021 0.027 Highest temperature 1.11 (0.69, 1.79) 0.659 0.663 Lowest SPO2 0.98 (0.97, 1) 0.016 0.04 Lowest HCT 1.13 (1.07, 1.19) < 0.001 < 0.001 Lowest HB 1.39 (1.18, 1.63) < 0.001 < 0.001 Lowest PLT 0.9999 (0.9955, 1.0043) 0.966 0.966 Highest WBC 1.03 (0.99, 1.06) 0.142 0.165 Highest AG 1.07 (1.01, 1.14) 0.014 0.022 Lowest sodium 1.002 (0.9993, 1.0048) 0.148 0.186 ALP 1.02 (0.88, 1.18) 0.809 0.812 Highest Cre 0.97 (0.92, 1.03) 0.341 0.354 SAPS II 1.02 (1.01, 1.04) 0.007 0.015 OASIS 1.04 (1.01, 1.08) 0.014 0.016 Significant values are in bold. Table 3 Relationship between different ALP levels and ICU in-hospital morality in different models. ALP Non-adjusted Model Model 1 Model 2 HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value ALP(IU/L) 1 (1–1) 0.148 1 (1–1.01) 0.068 1 (1–1) 0.574 22–62 IU/L 0.32(0.12–0.86) 0.024 0.32 (0.12–0.89) 0.028 0.32 (0.11–0.95) 0.039 63–99 IU/L 0.52(0.23–1.15) 0.105 0.52 (0.23–1.17) 0.112 0.47 (0.18–1.21) 0.117 100–659 IU/L Reference Reference Reference Data presented are HRs and 95% CIs. Model I: adjusted for age, race. Model II: adjusted for age, race, HR, RR, temperature, SPO2, PLT, WBC, sodium and Sapsii. HR: heart rate; RR: Respiratory; PLT: Platelet; WBC, white blood cell.

Restricted cubic splines analysis in the fully adjusted models revealed a nonlinear association between alkaline phosphatase levels and the risk of mortality in the ICU (P for nonlinearity = 0.021, as shown in Fig. 2 ). By employing a three-piecewise linear regression model, we identified two inflection points at 50.682 IU/L and 178.001 IU/L for Alkaline phosphatase. (Table 4 ) When the Alkaline phosphatase level was below 50.682 IU/L, the mortality in the ICU decreased as the Alkaline phosphatase level increased (OR 0.918, 95% CI 0, Inf). In contrast, when the Alkaline phosphatase level was above 50.682 IU/L but below 178.001 IU/L, a positive association between Alkaline phosphatase and mortality in the ICU was observed. Table 4 Threshold effect analysis of ALP level and the ICU in-hospital mortality of MM. HR (95% CI) P value Turning point 1 50.682 (46.183, 55.182) Turning point 2 178.001(174.57, 181.431) ALP < 50.682 IU/L 0.918 (0, Inf) < 0.001 50.682 IU/L < = ALP < = 178.001 IU/L 1.016 (1.001, 1.032) 0.039 ALP > 178.001U/L 2.824 (0, Inf) 0.998 Likelihood ratio test – < 0.001 Non-linear test*1 – 0.048 Non-linear test*2 – 0.003 Data presented are HRs and 95% CIs. The table using Cox regression models. Adjusted for age, race, HR, RR, temperature, SPO2, PLT, WBC, sodium and Sapsii.

Discussion

This study is the first to identify a nonlinear relationship between ALP levels and ICU mortality in MM patients, with the lowest risk observed at 50.682 IU/L. When analyzed as a continuous variable, ALP showed no statistically significant association with mortality ( P > 0.05). When ALP is used as a categorical variable, when using ALP levels of 100–659 IU/L as the reference, ALP levels of 22–62 IU/L (compared to ALP levels of 63–99 IU/L) are associated with lower mortality rates. Through curve fitting, we can reveal the non-linear relationship between ALP and mortality rate of multiple myeloma patients in the ICU. Further improve the inflection point analysis, and when the mortality rate is the lowest, the value of ALP level is 50.682 IU/L. Our study demonstrates that ALP serves as an independent prognostic factor for in-hospital mortality among multiple myeloma (MM) patients admitted to ICU. This finding aligns with and corroborates previous research indicating an independent association between ALP tertiles and adjusted all-cause mortality risk ratios in critical care populations 25 . Thomson et al. 26 through prospective observational studies, it is believed that ALP is an important factor in critically ill patients and is associated with ICU mortality outcomes and clinical events. Clinicians should consider ALP monitoring alongside standard severity scores (e.g., SAPS II) to optimize risk stratification.

Elevated bone-specific ALP (BAP) and other bone turnover markers during bortezomib therapy for MM have been documented 27 . Zangari et al. 28 retrospectively demonstrated a correlation between BAP changes and myeloma treatment response. Clinical studies further suggest that bone-related markers, including ALP, may predict both therapeutic efficacy and disease progression timelines 29 . Notably, cabozantinib-induced serum ALP elevation correlates with optimal tumor response in MM 28 . These findings position ALP as a valuable biomarker for MM risk stratification and therapeutic monitoring.

In previous studies exploring the relationship between the mortality of ICU patients and nucleated red blood cells (NRBCs), correlation analysis has shown that the NRBC count is positively correlated with ALP. Moreover, the persistently high level of the NRBC count is risk factor for in-hospital mortality among ICU patients 30 . In our study, we supplemented the non-linear relationship between ALP and mortality in multiple myeloma patients in the ICU, and further explored the inflection point value to make the relationship between the two more specific and clear. And by looking at ALP levels in ICU-hospitalized MM patients, the quality of life of patients could be better observed and subsequent treatment measures adjusted.

Although an observational retrospective study did not directly investigate the relationship between low ALP and reduced mortality in MM, it provided evidence supporting the unique significance of low ALP in MM 14 . A secondary retrospective analysis of a subset of MM patients from a large randomized controlled trial (RCT) partially reflects the association between varying ALP levels and survival outcomes in MM 31 . Furthermore, histomorphometric studies and investigations into the biological mechanisms of myeloma cell-induced osteoblast inhibition provide a theoretical basis for understanding the relationship between low ALP and impaired osteoblast activity in MM 32 . The characteristic of multiple myeloma is bone metabolism disorder, and the occurrence of osteolytic bone disease is a key clinical feature of multiple myeloma, which is attributed to the disruption of the subtle balance between bone formation and bone resorption 33 . The combined effect of osteogenic differentiation disorders and osteoclast dysfunction leads to bone destruction, while ALP mainly reflects the activity of osteoblasts 34 . The drugs used to treat multiple myeloma mainly have related effects on bone metabolism in relapsed/refractory MM patients by reducing osteoblast inhibition and inducing bone formation 33 . ALP can to some extent reflect the disease progression of patients with multiple myeloma.

The SAPSII score is an important component of the acute pathology and chronic health assessment scoring system, which is mainly used to assess the severity of organ failure 18 . SAPSII score can be used as a predictor of mortality in critically ill patients 35 . Previous studies have shown that the Oxford Acute Illness Severity Score is useful in assessing severity and prognosis in critically ill patients and is readily available in clinical practice 36 . In clinical practice, observing the changing trend of SAPSII and OASIS can better predict the subsequent quality of life of patients, help clinicians appropriately adjust treatment regimens, and improve clinical prognosis. However, because we found that OASIS did not have an adequate effect on mortality in the ICU, the data were not included in the covariates of our data 37 . Therefore, in our multivariate study, we only adjusted for the effect of SAPSII scores on disease status in patients with MM.

In our study, the ICU in-hospital mortality rate (20.5% 33/161) in ICU MM patients was lower than reported in previous studies. Azoulay et al. 38 observed a 30-day mortality rate of 57.23% in this group. These differences may be due to different databases, different groups of participants, the duration of the study, family financial situation, implementation of end-of-life decisions, and treatment methods. However, compared to other clinical studies, our included population is very small, but for this study, it is already the largest amount of data that can be included in the MIMIC database.

The advantage of this study is that appropriate statistical methods were used to determine the curvilinear relationship between ALP levels and ICU in-hospital mortality, rather than a simple linear relationship.

In addition, we found the inflection point for the smooth curve fit, rather than the manual grouping.

These results suggest that ICU physicians should be aware of ALP levels in the face of MM, especially true if ALP levels are above the lowest inflection point.

However, several limitations need to be acknowledged. First, this is a retrospective study, and the diagnosis of the disease was selected by the diagnosis code in the MIMIC database, but ICD coding inaccuracies and selection bias cannot be excluded. Second, the generalizability of this study is inherently limited by its exclusive focus on U.S. residents. Single-center U.S. data may not extrapolate to global populations. Future research should prioritize assembling multiethnic, multinational cohorts to validate and refine these associations. Third, the scope of our study is MM patients admitted to the ICU to assess their quality of life, and whether this inflection point applies to all MM patients at any time remains to be examined. Fourthly, due to the nature of the simulation database, there is a lack of specific information on patients with multiple myeloma. However, we used a severity scoring system, the OASIS score and the SAPSII score, to assess the severity of MM and comorbidities, among other things. Fifthly, Unmeasured factors (e.g., bisphosphonate use, cytogenetic risk) may influence ALP-mortality associations. Sixthly, the cohort (n = 161), though the largest MM subset in MIMIC-IV, the sample size was relatively small, thereby constraining the statistical power and the ability to explore potential interactions comprehensively. After inclusion in this study, the sample size of the data is still very small. In this study, we identified two turning points, with relatively few high-level ALP values in the collected data. In the future, more high-value ALP indicators need to be collected to continue improving the investigation of this study. Curve fitting analysis revealed that the majority of ALP values in our dataset clustered around 0–150 IU/L, with limited observations exceeding this threshold. Given that the second identified inflection point (178.001 IU/L) exceeded 150 IU/L and considering the sparse distribution of higher ALP values, we retained only the first inflection point (50.682 IU/L) in our final analysis. This conservative approach was adopted to ensure scientific rigor, minimize potential confounding from outliers or missing data, and enhance the robustness of our statistical conclusions. The potential existence of the second turning point still has certain clinical significance, and there is no clear biological explanation for this biphasic trend. Further discussion is needed on the possible mechanisms and the convenience and guidance that may be brought to clinical practice.

This study represents the first endeavor to elucidate the relationship between ALP levels and the risk of Mortality in the ICU using data derived from MIMIC. The research approach adopted in this study was authentic, showcasing significant novelty and potential therapeutic implications. Additionally, the presented risk estimates should be interpreted judiciously, taking into account biological plausibility, as the possibility of spurious findings due to chance arising from multiple comparisons cannot be overlooked. Further research is needed to validate our results and delve into the intricate relationship and potential underlying mechanisms.

Conclusions

In this retrospective cohort study of Multiple Myeloma patients in the MIMIC database, we have demonstrated that ALP is independently associated the risk of mortality in the ICU. Importantly, we observed a dose–response relationship, indicating that ALP is related to ICU mortality in a non-linear relationship (P for nonlinearity = 0.021), with the curves showing a shallower slope when ALP levels were 50.682 IU/L. When dividing ALP into tertile, when using ALP levels of 100–659 IU/L as the reference, ALP levels of 22–62 IU/L (compared to ALP levels of 63–99 IU/L) are associated with lower mortality rates.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Abbreviations

Intensive care unit

Multiple myeloma

Alkaline phosphatase

Standard deviation

Interquartile range

Heart rate

Respiratory rate

Systolic blood pressure

Diastolic blood pressure

Hematocrit

Hemoglobin

Platelet

White blood cell

Anion gap

Creatinine

Simplified acute physiology score II

Oxford acute severity of illness score

Diabetes mellitus

Ischemic heart disease

Nucleated red blood cell

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Acknowledgements

We thank Dr. Liu Jie (People’s Liberation Army of China General Hospital, Beijing, China) ,Dr. Yang Qilin(The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou, Guangdong, China) and Dr. Ni wentao (Peking University People’s Hospital, Beijing, China) for helping in this revision.

Author contributions

QH W conducted data analysis and wrote the manuscript, collection and data interpretation. GX D designed the study and HB C reviewed the manuscript. All authors reviewed and edited the report and approved the final draft.

Funding

None.

Data availability

Data in the article can be obtained from mimic-IV data- base ( https://mimic.physionet.org/ ).

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The establishment of this database was approved by the Massachusetts Institute of Technology (Cambridge, MA) and Beth Israel Deaconess Medical Center (Boston, MA), and consent was obtained for the original data collection. Therefore, the ethical approval statement and the need for informed consent were waived for this manuscript.

Consent for publication

There was no requirement of individual informed consent to extract data from the mimic-IV database because mimic-IV database information was publicly available and all patient data were deidentified.

References

1. Kastritis E The role of the bone microenvironment in multiple myeloma Blood Rev. 2017 31 3 213 223 10.1016/j.blre.2017.01.001 Kastritis, E. et al. The role of the bone microenvironment in multiple myeloma. Blood Rev. 31 (3), 213–223. 10.1016/j.blre.2017.01.001 (2017).28215731

2. Roodman GD Mechanisms of bone metastasis N. Engl. J. Med. 2004 350 16 1655 1664 10.1056/NEJMra030831 15084698 Roodman, G. D. Mechanisms of bone metastasis. N. Engl. J. Med. 350 (16), 1655–1664. 10.1056/NEJMra030831 (2004). 15084698 10.1056/NEJMra030831

3. Giuliani N Multiple myeloma bone disease: Pathophysiology of osteoblast inhibition Blood 2006 108 6 2020 2027 10.1182/blood-2006-05-026112 16917004 Giuliani, N. et al. Multiple myeloma bone disease: Pathophysiology of osteoblast inhibition. Blood 108 (6), 2020–2027. 10.1182/blood-2006-05-026112 (2006). 16917004 10.1182/blood-2006-05-026112

4. Mansour A Emerging roles of osteoclasts in the modulation of bone microenvironment and immune suppression in multiple myeloma Front. Immunol. 2017 8 954 10.3389/fimmu.2017.00954 28848556 PMC5554508 Mansour, A. et al. Emerging roles of osteoclasts in the modulation of bone microenvironment and immune suppression in multiple myeloma. Front. Immunol. 8 , 954. 10.3389/fimmu.2017.00954 (2017). 28848556 10.3389/fimmu.2017.00954 PMC5554508

5. Rajkumar SV Dimopoulos MA Palumbo A Blade J Merlini G Mateos MV Kumar S Hillengass J Kastritis E Richardson P Landgren O International myeloma working group. Updated criteria for the diagnosis of multiple myeloma Lancet Oncol. 2014 15 12 e538 e548 10.1016/S1470-2045(14)70442-5 25439696 Rajkumar, S. V. et al. International myeloma working group. Updated criteria for the diagnosis of multiple myeloma. Lancet Oncol. 15 (12), e538–e548. 10.1016/S1470-2045(14)70442-5 (2014). 25439696 10.1016/S1470-2045(14)70442-5

6. Huang J Chan SC Lok V The epidemiological landscape of multiple myeloma: A global cancer registry estimate of disease burden, risk factors, and temporal trends Lancet Haematol. 2022 9 9 e670 e677 10.1016/S2352-3026(22)00165-XIF:15.4Q1 35843248 Huang, J. et al. The epidemiological landscape of multiple myeloma: A global cancer registry estimate of disease burden, risk factors, and temporal trends. Lancet Haematol. 9 (9), e670–e677. 10.1016/S2352-3026(22)00165-XIF:15.4Q1 (2022). 35843248 10.1016/S2352-3026(22)00165-X

7. Haarhaus M Cianciolo G Barbuto S Alkaline phosphatase: an old friend as treatment target for cardiovascular and mineral bone disorders in chronic kidney disease Nutrients 2022 14 10 2124 10.3390/nu14102124 35631265 PMC9144546 Haarhaus, M. et al. Alkaline phosphatase: an old friend as treatment target for cardiovascular and mineral bone disorders in chronic kidney disease. Nutrients 14 (10), 2124. 10.3390/nu14102124 (2022). 35631265 10.3390/nu14102124 PMC9144546

8. Stewart JP Shaughnessy JD Jr Role of osteoblast suppression in multiple myeloma J. Cell Biochem. 2006 98 1 1 13 10.1002/jcb.20774IF:3.0Q3 16440324 Stewart, J. P. & Shaughnessy, J. D. Jr. Role of osteoblast suppression in multiple myeloma. J. Cell Biochem. 98 (1), 1–13. 10.1002/jcb.20774IF:3.0Q3 (2006). 16440324 10.1002/jcb.20774

9. Tutak SA Relationship between mortality and albumin, alkaline phosphatase, phosphorus and calcium values that may change with renal function and nutrition in intensive care setting Anesth. Crit. Care 2021 3 74 82 Tutak, S. A. Relationship between mortality and albumin, alkaline phosphatase, phosphorus and calcium values that may change with renal function and nutrition in intensive care setting. Anesth. Crit. Care 3 , 74–82 (2021).

10. Mao Y Zhu S Geng Y Association between serum calcium and in-hospital mortality in critical patients with multiple myeloma: A cohort study Hematology 2022 27 1 795 801 10.1080/16078454.2022.2095948 35820067 Mao, Y., Zhu, S. & Geng, Y. Association between serum calcium and in-hospital mortality in critical patients with multiple myeloma: A cohort study. Hematology 27 (1), 795–801. 10.1080/16078454.2022.2095948 (2022). 35820067 10.1080/16078454.2022.2095948

11. Kalantar-Zadeh K Shah A Duong U Hechter RC Dukkipati R Kovesdy CP Kidney bone disease and mortality in CKD: Revisiting the role of vitamin D, calcimimetics, alkaline phosphatase, and minerals Kidney Int. Suppl. 2010 117 S10 S21 10.1038/ki.2010.189 PMC5494176 20671739 Kalantar-Zadeh, K. et al. Kidney bone disease and mortality in CKD: Revisiting the role of vitamin D, calcimimetics, alkaline phosphatase, and minerals. Kidney Int. Suppl. 117 , S10–S21. 10.1038/ki.2010.189 (2010). 10.1038/ki.2010.189 PMC5494176 20671739

12. Sumida K Molnar MZ Potukuchi PK Prognostic significance of pre-end-stage renal disease serum alkaline phosphatase for post-end-stage renal disease mortality in late-stage chronic kidney disease patients transitioning to dialysis Nephrol. Dial. Transplant. 2018 33 2 264 273 10.1093/ndt/gfw412 28064159 PMC5837395 Sumida, K. et al. Prognostic significance of pre-end-stage renal disease serum alkaline phosphatase for post-end-stage renal disease mortality in late-stage chronic kidney disease patients transitioning to dialysis. Nephrol. Dial. Transplant. 33 (2), 264–273. 10.1093/ndt/gfw412 (2018). 28064159 10.1093/ndt/gfw412 PMC5837395

13. Irisawa H Bone disease in multiple myeloma Nihon Rinsho 2015 73 1 42 46 25626302 Irisawa, H. Bone disease in multiple myeloma. Nihon Rinsho 73 (1), 42–46 (2015). 25626302

14. Annibali O Petrucci MT Santini D Alkaline phosphatase (alp) levels in multiple myeloma and solid cancers with bone lesions: Is there any difference? J. Bone Oncol. 2020 26 100338 10.1016/j.jbo.2020.100338 33304804 PMC7708946 Annibali, O. et al. Alkaline phosphatase (alp) levels in multiple myeloma and solid cancers with bone lesions: Is there any difference?. J. Bone Oncol. 26 , 100338. 10.1016/j.jbo.2020.100338 (2020). 33304804 10.1016/j.jbo.2020.100338 PMC7708946

15. Diao X Cai R Luo J Zheng Z Zhan H Prognostic factors for patients with multiple myeloma admitted to the intensive care unit Hematology 2020 25 1 433 437 10.1080/16078454.2020.1845502 33210963 Diao, X., Cai, R., Luo, J., Zheng, Z. & Zhan, H. Prognostic factors for patients with multiple myeloma admitted to the intensive care unit. Hematology 25 (1), 433–437. 10.1080/16078454.2020.1845502 (2020). 33210963 10.1080/16078454.2020.1845502

16. Goldberger AL PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals Circulation 2000 101 23 E215 220 10.1161/01.cir.101.23.e215 10851218 Goldberger, A. L. et al. PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation 101 (23), E215-220. 10.1161/01.cir.101.23.e215 (2000). 10851218 10.1161/01.cir.101.23.e215

17. Huimin SO Jing WA Chang HU Chang LI Jianguo LI Effects of cholestasis and hypoxic hepatitis on prognosis of ICU patients: A retrospective study based on MIMIC III database J. South. Med. Univ. 2020 40 6 771 777 10.12122/j.issn.1673-4254.2020.06.01 PMC7321284 32895209 Huimin, S. O., Jing, W. A., Chang, H. U., Chang, L. I. & Jianguo, L. I. Effects of cholestasis and hypoxic hepatitis on prognosis of ICU patients: A retrospective study based on MIMIC III database. J. South. Med. Univ. 40 (6), 771–777. 10.12122/j.issn.1673-4254.2020.06.01 (2020). 10.12122/j.issn.1673-4254.2020.06.01 PMC7321284 32895209

18. Full article: Association between serum calcium and in-hospital mortality in critical patients with multiple myeloma: a cohort study (no date). Available at: https://www.tandfonline.com/doi/full/10.1080/16078454.2022.2095948 (Accessed: 19 April 2024). 10.1080/16078454.2022.2095948 35820067

19. Johnson AE ‘The MIMIC code repository: Enabling reproducibility in critical care research J. Am. Med. Inform. Assoc. JAMIA 2018 25 1 32 39 10.1093/jamia/ocx084 29036464 PMC6381763 Johnson, A. E. et al. ‘The MIMIC code repository: Enabling reproducibility in critical care research. J. Am. Med. Inform. Assoc. JAMIA 25 (1), 32–39. 10.1093/jamia/ocx084 (2018). 29036464 10.1093/jamia/ocx084 PMC6381763

20. Zumárraga JP Serum values of alkaline phosphatase and lactate dehydrogenase in osteosarcoma Acta Ortopedica Brasileira 2016 24 3 142 146 10.1590/1413-785220162403157033 27217815 PMC4863862 Zumárraga, J. P. et al. Serum values of alkaline phosphatase and lactate dehydrogenase in osteosarcoma. Acta Ortopedica Brasileira 24 (3), 142–146. 10.1590/1413-785220162403157033 (2016). 27217815 10.1590/1413-785220162403157033 PMC4863862

21. Le Gall, J. R., Lemeshow, S., Saulnier, F. A new simplified acute physiology score (SAPS II) based on a European/North American multicenter study [published correction appears in JAMA 1994 May 4;271(17):1321]. JAMA 270 (24), 2957-2963. 10.1001/jama.270.24.2957 (1993 ) . 10.1001/jama.270.24.2957 8254858

22. Wang, N., Wang, M., Jiang, L., Du, B., Zhu, B. & Xi, X. The predictive value of the oxford acute disease severity score for clinical outcomes in patients with acute kidney injury: A secondary analysis of a large prospective observational study (2021).

23. Hwang SD Kim SH Kim YO Serum alkaline phosphatase levels predict infection-related mortality and hospitalization in peritoneal dialysis patients PLoS ONE 2016 11 6 e0157361 10.1371/journal.pone.0157361 27310428 PMC4911047 Hwang, S. D. et al. Serum alkaline phosphatase levels predict infection-related mortality and hospitalization in peritoneal dialysis patients. PLoS ONE 11 (6), e0157361. 10.1371/journal.pone.0157361 (2016). 27310428 10.1371/journal.pone.0157361 PMC4911047

24. Yang Q Association between preadmission metformin use and outcomes in intensive care unit patients with sepsis and type 2 diabetes: A cohort study Front. Med. 2021 10.3389/fmed.2021.640785 33855034 PMC8039324 Yang, Q. et al. Association between preadmission metformin use and outcomes in intensive care unit patients with sepsis and type 2 diabetes: A cohort study. Front. Med. 10.3389/fmed.2021.640785 (2021). 33855034 10.3389/fmed.2021.640785 PMC8039324

25. Tonelli M Curhan G Pfeffer M Sacks F Thadhani R Melamed ML Wiebe N Muntner P Relation between alkaline phosphatase, serum phosphate, and all-cause or cardiovascular mortality Circulation 2009 120 1784 1792 19841303 10.1161/CIRCULATIONAHA.109.851873 Tonelli, M. et al. Relation between alkaline phosphatase, serum phosphate, and all-cause or cardiovascular mortality. Circulation 120 , 1784–1792 (2009). 19841303 10.1161/CIRCULATIONAHA.109.851873

26. Thomson SJ 'Liver function tests’ on the intensive care unit: A prospective, observational study Intensive Care Med. 2009 35 8 1406 1411 10.1007/s00134-009-1511-7IF:27.1Q1 19513695 Thomson, S. J. et al. “Liver function tests” on the intensive care unit: A prospective, observational study. Intensive Care Med. 35 (8), 1406–1411. 10.1007/s00134-009-1511-7IF:27.1Q1 (2009). 19513695 10.1007/s00134-009-1511-7

27. Heider U Treatment of bortezomib increases osteoblast function in patients with multiple myeloma Blood 2005 106 3457 3457 Heider, U. et al. Treatment of bortezomib increases osteoblast function in patients with multiple myeloma. Blood 106 , 3457–3457 (2005).16037393

28. Zangari M Alkaline phosphatase variation during carfilzomib treatment is associated with best response in multiple myeloma patients Eur. J. Haematol. 2011 86 6 484 487 10.1111/j.1600-0609.2011.01602.x 21477075 Zangari, M. et al. Alkaline phosphatase variation during carfilzomib treatment is associated with best response in multiple myeloma patients. Eur. J. Haematol. 86 (6), 484–487. 10.1111/j.1600-0609.2011.01602.x (2011). 21477075 10.1111/j.1600-0609.2011.01602.x

29. Zangari M Predictive value of alkaline phosphatase for response and time to progression in bortezomib-treated multiple myeloma patients Am. J. Hematol. 2007 82 9 831 833 10.1002/ajh.20961 17546639 Zangari, M. et al. Predictive value of alkaline phosphatase for response and time to progression in bortezomib-treated multiple myeloma patients. Am. J. Hematol. 82 (9), 831–833. 10.1002/ajh.20961 (2007). 17546639 10.1002/ajh.20961

30. Wang J Su Z Zhao X Investigation of the possible causes of NRBC in ICU patients and the dynamic trend of NRBC count in survival and death patients or with different underlying diseases: A retrospective study Evid. Based Complement Altern. Med. 2023 2023 9076647 10.1155/2023/9076647 PMC9867577 36691599 Wang, J. et al. Investigation of the possible causes of NRBC in ICU patients and the dynamic trend of NRBC count in survival and death patients or with different underlying diseases: A retrospective study. Evid. Based Complement Altern. Med. 2023 , 9076647. 10.1155/2023/9076647 (2023). 10.1155/2023/9076647 PMC9867577 36691599

31. Berenson J Survival in patients with multiple myeloma receiving zoledronic acid: Stratification by baseline bone alkaline phosphatase levels JCO 2006 24 7505 7505 10.1200/jco.2006.24.18_suppl.7505 Berenson, J. et al. Survival in patients with multiple myeloma receiving zoledronic acid: Stratification by baseline bone alkaline phosphatase levels. JCO 24 , 7505–7505. 10.1200/jco.2006.24.18_suppl.7505 (2006).

32. Giuliani N Rizzoli V Roodman GD Multiple myeloma bone disease: Pathophysiology of osteoblast inhibition Blood 2006 108 13 3992 3996 10.1182/blood-2006-05-026112 16917004 Giuliani, N., Rizzoli, V. & Roodman, G. D. Multiple myeloma bone disease: Pathophysiology of osteoblast inhibition. Blood 108 (13), 3992–3996. 10.1182/blood-2006-05-026112 (2006). 16917004 10.1182/blood-2006-05-026112

33. Terpos E Daratumumab improves bone turnover in relapsed/refractory multiple myeloma; phase 2 study “REBUILD Cancers 2022 14 11 2768 10.3390/cancers14112768 35681747 PMC9179322 Terpos, E. et al. Daratumumab improves bone turnover in relapsed/refractory multiple myeloma; phase 2 study “REBUILD. Cancers 14 (11), 2768. 10.3390/cancers14112768 (2022). 35681747 10.3390/cancers14112768 PMC9179322

34. Wang Y Man Q-H Li C Clinical efficacy of Thalidomide combined with PAD Regimen in patients with multiple Myeloma Pak. J. Med. Sci. 2023 39 5 1243 1248 10.12669/pjms.39.5.6815 37680825 PMC10480731 Wang, Y., Man, Q.-H. & Li, C. Clinical efficacy of Thalidomide combined with PAD Regimen in patients with multiple Myeloma. Pak. J. Med. Sci. 39 (5), 1243–1248. 10.12669/pjms.39.5.6815 (2023). 37680825 10.12669/pjms.39.5.6815 PMC10480731

35. Tekin B The comparison of scoring systems: SOFA, APACHE-II, LODS, MODS, and SAPS-II in critically ill elderly sepsis patients J. Infect. Dev. Ctries. 2024 18 1 122 130 10.3855/jidc.18526 38377099 Tekin, B. et al. The comparison of scoring systems: SOFA, APACHE-II, LODS, MODS, and SAPS-II in critically ill elderly sepsis patients. J. Infect. Dev. Ctries. 18 (1), 122–130. 10.3855/jidc.18526 (2024). 38377099 10.3855/jidc.18526

36. Zhang M Wang Z Hong X Zheng S Fang X Xie L The value of Oxford acute severity score in severity assessment of critically Ill patients: A Single-center analysis based on 470 cases China J. Emerg. Med. 2017 26 2 197 201 Zhang, M. et al. The value of Oxford acute severity score in severity assessment of critically Ill patients: A Single-center analysis based on 470 cases. China J. Emerg. Med. 26 (2), 197–201 (2017).

37. Performance on the APACHE II, SAPS II, SOFA and the OHCA score of post-cardiac arrest patients treated with therapeutic hypothermia|PLOS ONE (no date). Available at: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0196197 (Accessed: 19 April 2024). 10.1371/journal.pone.0196197 PMC5933749 29723201

38. Hampshire PA Admission factors associated with hospital mortality in patients with haematological malignancy admitted to UK adult, general critical care units: A secondary analysis of the ICNARC Case Mix Programme Database Crit. Care 2009 13 4 137 10.1186/cc8016 19706163 PMC2750195 Hampshire, P. A. et al. Admission factors associated with hospital mortality in patients with haematological malignancy admitted to UK adult, general critical care units: A secondary analysis of the ICNARC Case Mix Programme Database. Crit. Care 13 (4), 137. 10.1186/cc8016 (2009). 19706163 10.1186/cc8016 PMC2750195