Prognostic Factors for High Intraoperative Blood Loss for Multiple Myeloma-Related Bone Disease in the Spine

Abstract

Study Design

Retrospective multicenter cohort study.

Objectives

Spine surgery for multiple myeloma (MM) is associated with an increased intraoperative blood loss. Therefore, this study aims to examine prognostic factors for higher intraoperative blood loss in spine surgery for patients with MM.

Methods

In total, 158 adult patients with MM undergoing spine surgery between May 2001 and December 2021 were included. The main outcome for intraoperative blood loss was the Bleeding Index (BI), next to the visually estimated blood loss (EBL). Two separate multivariable generalized linear models (GLMs) were utilized to assess the associations between the predictors and these two outcomes.

Results

The average BI was 4.4 and average EBL was 750 mL. Compared to corpectomy with stabilization, other types of surgery (decompression with stabilization, sole decompression, sole stabilization) were associated with a lower expected BI, ranging from a 26.5% to 39% decrease. A cervical location of surgery was associated with a 40.3% reduction of expected BI compared to a lumbar location (P = 0.006). Lower platelet count (P = 0.003) and longer duration of surgery (P < 0.001) were associated with a higher expected BI. For EBL, ECOG score, surgery type, and duration of surgery were found as independent predictors.

Conclusions

This study identified lower platelet count, type of surgery, location of operated spinal levels, and a longer duration of surgery as independent predictors of higher intraoperative BI in MBD-related spine surgery. These outcomes can be relevant for preoperative screening, shared decision making, and perioperative blood transfusion deliberation or planning.

Keywords

multiple myeloma intraoperative blood loss prognostic factors oncological spine surgery

Highlights

In this study, the multiple myeloma (MM) patients had an average Bleeding Index (BI) of 4.4 and an average visually estimated blood loss (EBL) of 750 mL.

Platelet count (P = 0.003), corpectomy with stabilization and a longer duration of surgery (P < 0.001) were independent predictors for a higher expected BI. Surgery in the cervical spine was an independent predictor for a lower expected BI compared to the lumbar region (P = 0.006).

A longer duration of surgery (P < 0.001) corpectomy with stabilization, and ECOG score (ECOG 2-4 vs 0-1) (P = 0.043) were independent predictors for a higher expected EBL.

Bleeding scores were the highest for patients undergoing a corpectomy with stabilization (BI 6.7, EBL 1550 mL), followed by decompression with stabilization for BI (BI 4.3) and sole decompression for EBL (EBL 750 mL).

Surgeries in the thoracic and lumbar region had a higher BI (5.8), compared to the cervical region (BI 2.6). For EBL, the thoracic region had the highest intraoperative bleeding (900 mL), compared to the cervical region (325 mL) and the lumbar region (650 mL).

Introduction

Multiple myeloma (MM) is a hematologic malignancy with a high risk of bone involvement; around 80% of patients present with osteolytic lesions at their diagnosis.¹ Myeloma-related bone disease (MBD) is characterized by complex disturbances in the bone remodeling interaction, mainly through initiation of osteocyte apoptosis by MM cells. This disrupts the osteoclast-osteoblast balance, resulting in diffuse osteoporosis. Additionally, the bone marrow microenvironment is altered and unrestricted angiogenesis is induced, causing proliferation and preservation of MM cells.^2,3 MBD often leads to vertebral compression fractures (VCFs), which are already present in 46.8% of patients with newly diagnosed MM.⁴ New, impending or worsened VCFs can occur regardless of administered radiotherapy and can lead to aggravated spinal instability with subsequent required surgical intervention.^5,6 Additionally, patients with MBD are at high risk for skeletal-related events, spinal cord compression, and surgical or radiotherapeutic interventions.¹

Current recommendations for MBD management encompass zoledronic acid, preferably with calcium and vitamin D, or denosumab (human monoclonal antibody therapy).¹ The use of spine surgery in MBD is debated, especially due to the general poor bone quality, widespread lesions in the vertebral column, and alleged disease-specific high risk of infections.⁷ Nowadays, spine surgery in MM is only recommended in cases of vertebral instability or spinal cord compression. However, less invasive procedures such as cement augmentation or balloon kyphoplasty are often applied to treat painful VCFs.¹

Of all intraoperative complications in MBD-related spine surgery, around 40% were related to excessive blood loss.⁸ MM histology is associated with increased intraoperative blood loss in spine surgery, which is significantly higher compared to interventions for spinal metastases for solid tumors.^9–11 Knowledge of predictors for high intraoperative blood loss can be relevant for preoperative risk assessment, particularly for vulnerable oncologic patient groups like the MM population. Identifying patients at elevated risk for significant intraoperative blood loss could enable the surgical team to establish optimal operative procedures and provide sufficient transfusion resources. Therefore, this study aims to provide prognostic factors for higher intraoperative blood loss in spine surgery for patients with MM.

Methods

Study Design

This retrospective cohort study was conducted at two tertiary academic medical centers that are part of a single umbrella organization. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations were used to guarantee proper reporting of data.¹²

Patient Selection

In this study, we included adult patients diagnosed with MM undergoing operative treatment for MBD (corpectomy with stabilization, decompression surgery with or without stabilization, stabilization alone) between May 2001 and December 2021. No distinctions were made for myeloma subtypes. Only patients with complete preoperative and postoperative hemoglobin data were included. Patients undergoing vertebral augmentation procedures were excluded, as an incomparably lower amount of blood loss was expected in these minimally invasive procedures. Patients with prior cement augmentations in the surgically treated area were excluded.

Outcome Measures

Several relevant variables were examined for associations with high intraoperative blood loss in MM. In the patient’s electronic medical records, the following baseline variables were collected: age at surgery, gender, body mass index (BMI, in kg/m²), deceased at 1 year after surgery (y/n), number of spinal and skull lesions, American Spinal Injury Association (ASIA) impairment scale,¹³ back pain prior to surgery (y/n), additional comorbidities using the Age-adjusted Charlson Comorbidity Index (aCCI),¹⁴ Eastern Cooperative Oncology Group (ECOG) performance status score,¹⁵ and prior local radiation treatments and systemic therapy (y/n). The extracted surgical variables were amount of used packed red blood cells (pRBC) for blood transfusions, the type of surgery, the surgical approach (anterior, posterior, or combined) the number and location of operated spinal levels, time between first diagnosis of MM and surgery, and the duration of the procedure. Indications for surgery encompassed either spinal cord compression or significant vertebral instability with impending spinal cord compression. Preoperatively, hemoglobin, white blood cell count, platelet count, lymphocyte count, neutrophil count, alkaline phosphatase, calcium, albumin, creatinine, prothrombin test (PT), PT expressed as the International Standardized Ratio (PT/INR),¹⁶ and partial thromboplastin time (PTT) are the laboratory values that were collected.

Blood Loss

The Bleeding Index ((BI) was applied as a main outcome for intraoperative blood loss:^11,17,18

B I = p o i n t s H b d e c r e a s e ({H b}_{p r e o p} - {H b}_{p o s t o p}) + n u m b e r o f p R B C u n i t s

Where Hb_preop (g/dL) is the patient’s preoperative hemoglobin concentration and Hb_postop (g/dL) is the patient’s lowest hemoglobin concentration within 5 days postoperatively. This 5-day limit was chosen as hemoglobin drifts appear to stabilize and do not exhibit notable fluctuations 2-4 days after major surgeries.¹⁹ The pRBC units were counted when applied both intraoperatively and postoperatively prior to the lowest hemoglobin concentration. The estimated blood loss (EBL) as visually assessed by the anesthesia or operating team and reported in the operative notes, was used as a secondary bleeding outcome, as this method encompasses a risk of underestimation of blood loss.^20,21,22 No cell salvage techniques were applied in these oncologic cases.

Missing Data

All variables were examined for missing data and excluded from further analyses when the percentage was over 30%. Missing data included: BMI (3.8%), deceased at 1 year after surgery (24.1%), number of spinal lesions (10.8%) and skull lesions (11.4%), aCCI (4.4%), ASIA score (1.3%), ECOG score (14.6%), prior local radiation treatments (10.8%) and systemic therapy (0.6%), location of operated spinal levels (10.1%), time between first diagnosis and MM and surgery (1.3%), duration of surgery (12%), white blood cell count (1.3%), lymphocyte count (12%), neutrophil count (16.5%), alkaline phosphatase (8.9%), calcium (1.9%), albumin (7%), creatinine (1.3%), PT (6.3%), PT/INR (8.2%), and PTT (12%) Accordingly, no variables were removed from further analyses. Little’s MCAR test indicated that the missing data was missing completely at random (P = 0.749). Therefore, multiple imputation by chained equations was performed with 25 generated imputed datasets and 100 iterations to ensure convergence.

Statistical Analysis

After imputation and prior to multivariable modeling, several diagnostics were performed. Normality checks showed non-normal distribution of the outcome variables BI and EBL. Multicollinearity among predictor variables was assessed using an adjusted Generalized Variance Inflation Factor (GVIF) analysis, calculated as GVIF^(1/(2 × df)), where df represents the degrees of freedom of the predictor. No potentially problematic explanatory variables (values above 5) were observed. Assessment of outliers showed that extreme outliers (|z| >3) were rare after imputation.

The clinical and demographic parameters of the study population were described using descriptive statistics. Univariate analyses were performed to explore associations with the primary outcomes using Spearman’s rank correlation for continuous predictors and the Kruskal–Wallis test for categorical variables. For the multivariable analysis, variables were selected based on clinical relevance, previous research, and the univariate analysis.^9,23–25 A rule-of-thumb allowing 1 degree of freedom per ten patients (N = 158) was used. Given the positively skewed distributions of both BI and EBL, two separate multivariable generalized linear models (GLMs) with a gamma distribution and log-link function were utilized to assess the associations between the predictors and the two outcomes. To reduce the risk of overfitting while preserving model performance, simplified and parsimonious models were created and were compared to the full models using McFadden pseudo-R². In the simplified models, the df were reduced by clinically relevant dichotomization of included variables. In the parsimonious models, solely significant variables in either of the univariate analyses with strong clinical rationale were retained. Normality of the distributions of residuals was visually inspected. Influential data points were identified using Cook’s distance and leverage test and removed if required. Model goodness-of-fit was assessed in a comparison of the deviance and the Akaike information criterion (α = 0.157). All models showed a lower value next to the null models. No observations were removed after Cook’s distance and leverage test analyses before or after multivariable modeling.

All statistical analyses and predictive modeling were performed with RStudio programming language, version 2024.12.0 + 467 (Posit, https://posit.co/). Significance was defined as P < .05. All testing was 2-sided.

Results

Patient Cohort

After a manual screening of the electronic medical records of potentially eligible patients, a total of 164 patients were identified that underwent spine surgery for MBD. Six patients were excluded due to missing preoperative or postoperative hemoglobin data. No patients were removed due to previous augmentation procedures in the surgical area. Of the 158 remaining included patients, 97 were male (61.4%) (Table 1). The median age of patients at surgery was 63 years (standard deviation (SD) 10.88). The average BMI was 28.45 (SD 5.31). Most patients had no (ASIA E, 50%) to mild (ASIA D, 38%) neurologic impairment. Previous systemic therapy was administered in 73 patients (46.2%), 47 patients received radiation therapy preoperatively (29.7%) and 28 patients received both (17.7%). Most surgeries were performed in the thoracic region (76, 48.1%). The median duration of surgery was 288 minutes (Interquartile Range [IQR] 197.75 – 420) (Table 1). Corpectomy with stabilization was done in 62 patients, the most frequently performed procedure (39.2%), followed by decompression with stabilization (48, 30.4%), decompression (31, 19.6%), and stabilization (17, 10.8%). Most surgeries were performed with a posterior surgical approach (136, 86.1%). The baseline characteristics per surgery type were presented in Table A1 (see Appendix). The patient group that underwent sole decompression and sole stabilization procedures presented with skull lesions relatively frequently (38.7% and 11.8%) compared to more invasive procedures (4.8% for corpectomy with stabilization, 2.1% for decompression with stabilization). Additionally, previous systemic therapy was administered in around 65 percent in this patient group, compared to 32.3 (corpectomy with stabilization) and 45.8 percent (decompression with stabilization) for the larger procedures. Preoperative laboratory values were presented in Table 2.

Table 1.

Baseline Characteristics of the MM Patients

Variables	MM Patients (n = 158)	Median BI [IQR]	Median EBL (mL) [IQR]
Age at surgery (y)	63.08 (10.88)
Deceased at 1yr after surgery
Alive	112 (70.9%)	4.25 [2.5, 6.95]	725 [300, 1500]
Deceased	46 (29.1%)	5.9 [2.55, 7.85]	850 [400, 1550]
Sex
Female	61 (38.6%)	5.1 [2.5, 8.3]	775 [400, 1425]
Male	97 (61.4%)	4.30 [2.5, 6.9]	700 [300, 1600]
BMI	28.45 (5.31)
Pain
Yes	148 (93.7%)	4.55 [2.48, 7.25]	750 [375, 1550]
No	10 (6.3%)	3.05 [2.57, 4.07]	825 [521.25, 1123.75]
ECOG
0	37 (23.4%)	5.3 [2.8, 7.4]	500 [300, 950]
1	59 (37.3%)	4 [2.4, 6.5]	500 [300, 1050]
2	33 (20.9%)	3.8 [2.3, 8.8]	937.5 [512.5, 1300]
3	16 (10.1%)	5.55 [4.05, 6.73]	1175 [675, 1762.5]
4	13 (8.2%)	7.2 [6.4, 10.2]	2500 [1500, 3600]
ASIA
A	2 (1.3%)	2.75 [2.42, 3.08]	525 [462.5, 587.5]
B	2 (1.3%)	7.75 [7.47, 8.03]	2800 [2150, 3450]
C	15 (9.5%)	5.8 [3.4, 8.75]	1200 [425, 1900]
D	60 (38%)	4.45 [2.37, 7.23]	875 [450, 1600]
E	79 (50%)	4.3 [2.45, 6.9]	600 [300, 1100]
aCCI
0	60 (38%)	4.45 [2.57, 7.77]	750 [312.5, 1500]
1	66 (41.8%)	5.4 [2.4, 7.85]	600 [325, 1675]
2	5 (3.2%)	4.6 [2.5, 5.1]	775 [750, 925]
3	6 (3.8%)	6.9 [3.97, 7.12]	700 [425, 1612.5]
4	9 (5.7%)	3.1 [2.5, 3.4]	650 [500, 900]
5	8 (5.1%)	4.45 [2.03, 5.32]	800 [600, 2200]
6+	4 (2.5%)	2.35 [1.87, 4.1]	400 [250, 750]
Number of spinal lesions
1	48 (30.4%)	5.6 [2.98, 7.3]	800 [400, 1900]
2	30 (19%)	3.85 [2.5, 7.95]	900 [500, 1600]
3 or more	80 (50.6%)	4.4 [2.32, 7.1]	700 [300, 1180]
Presence of skull lesions
Yes	18 (11.4%)	4.05 [1.62, 8.75]	900 [500, 1500]
No	140 (88.6%)	4.4 [2.5, 7.13]	700 [350, 1575]
Previous local RT
Yes	47 (29.7%)	5.7 [2.6, 7.8]	700 [400, 1600]
No	111 (70.3%)	4.4 [2.45, 7.15]	750 [312.5, 1500]
Previous systemic therapy
Yes	73 (46.2%)	3.5 [2.1, 7.1]	600 [250, 1225]
No	85 (53.8%)	5.3 [3, 7.6]	900 [500, 1800]
Surgery type
Corpectomy with stabilization	62 (39.2%)	6.7 [4, 8.8]	1550 [600, 2137.5]
Decompression with stabilization	48 (30.4%)	4.3 [2.45, 6.95]	597.5 [337.5, 1000]
Decompression	31 (19.6%)	3 [2, 6.1]	750 [300, 1000]
Stabilization	17 (10.8%)	2.4 [1.8, 3.5]	200 [100, 512.5]
Surgical approach
Anterior	14 (8.9%)	6.45 [3.4, 8.95]	2000 [575, 2500]
Posterior	136 (86.1%)	4.4 [2.4, 7.13]	700 [325, 1350]
Combined	8 (5.1%)	4.95 [3.42, 6.83]	1200 [750, 1825]
Location of operated levels
Cervical	22 (13.9%)	2.6 [1.85, 3.85]	325 [200, 1137.5]
Thoracic	76 (48.1%)	5.8 [3.15, 8.35]	900 [500, 1700]
Lumbar	21 (13.3%)	5.8 [3, 7.2]	600 [175, 1750]
Combined	39 (24.7%)	3.3 [2.3, 6.1]	700 [400, 1000]
Number of operated spinal levels
1	40 (25.3%)	4.95 [2.1, 7.62]	1000 [400, 1800]
2+	118 (74.7%)	4.4 [2.5, 7.18]	700 [312.5, 1300]
Time from diagnosis to primary surgery (days)	35.5 [0, 388]
Duration of surgery (mins)	288 [197.75, 420]
Drain output (ml)^a	595 [387.5, 873.75]
Pre-operative hemoglobin (g/dl)	12.04 (1.92)
Post-operative hemoglobin (g/dl) (lowest)	8.9 [8, 10.28]
Units of pRBC	0 [0, 2]
BI		4.4 [2.5, 7.2]
EBL (ml)			750 [375, 1500]

Values are presented as number (%) and mean (standard deviation) for normally distributed data or median [Q1,Q3] for non-normally distributed data.

Abbreviations: aCCI, age-adjusted charlson comorbidity index; ASIA, american spinal cord injury association; BI, bleeding index; BMI, body mass index; EBL, visually estimated blood loss; ECOG, eastern cooperative oncology group; MM, multiple myeloma; pRBC, packed red blood cells; RT, radiation therapy.

^aDrains were placed in 54 patients (34.2%), values presented reflect this subgroup only.

Table 2.

Baseline Laboratory Values of the MM Patients

Lab variables	MM patients (n = 158)
Albumine (g/dL)	3.8 [3.5, 4.2]
Alkaline phosphatase (U/l)	69 [54.5, 84]
Calcium (mg/dL)	9 [8.3, 9.5]
Creatinine (mg/dL)	.81 [.69, 1.07]
Lymphocyte count (x10^⁹/L)	1.12 [0.57, 1.79]
Neutrophil count (x10^⁹/L)	5.23 [3.4, 7.42]
Platelet count (x10^⁹/L)	209.5 [149.75, 291.25]
PT	13.6 [12.9, 14.57]
PT/INR	1.1 [1, 1.17]
PTT	27 [24.8, 29.8]
White blood cell count (x10^⁹/L)	6.99 [5.17, 9.92]

Values are presented as number (%) and mean (standard deviation) for normally distributed data or median [Q1,Q3] for non-normally distributed data.

Abbreviations: MM, multiple myeloma; PT, prothrombin test; PT/INR, PT expressed as the international standardized ratio; PTT, partial thromboplastin time.

Bleeding Scores

Intraoperative blood loss was assessed using the BI and EBL. The median BI was 4.4 [IQR 2.5 - 7.2] (Table 1). Preoperatively, the average hemoglobin level was 12.04 g/dl (SD 1.92), equal to 7.47 mmol/l when a conversion factor of 0.6206 is applied. The lowest hemoglobin level within 5 days post-operatively was 8.9 g/dl (IQR 8 - 10.28) (5.52 mmol/l), on average. Drains were placed in 54 patients (34.2%), with a median output of 595 mL [IQR 387.5 - 873.75]. In patients receiving local radiation before surgery, the average BI was 5.7, compared to 4.4 in non-radiated patients. An average BI of 3.5 was reported in patients undergoing previous systemic therapy, compared to a BI of 5.3 in patients who did not. The type of surgery showed differences in the average BI. Patients who had corpectomies with stabilization had an average BI of 6.7, the highest bleeding score compared to the other types of surgery. Decompression with stabilization and sole decompression had BIs of 4.3 and 3, respectively. Sole stabilization had the lowest average BI of 2.4. The anterior surgical approach showed a higher average BI (6.45) compared to posterior (BI 4.4) or combined (BI 4.95), although solely applied in corpectomies with stabilization (Table 1, Table A1). Surgeries in the thoracic and lumbar region showed a relatively high average BI of 5.8 compared to the cervical region (BI 2.6). Surgeries with two or more operated spinal levels showed an average BI of 4.4, compared to BI 4.95 for single-level procedures.

The average median intraoperative EBL was 750 mL [IQR 375 - 1500]. Patients with an ECOG score of 4 had an average EBL of 2500 mL, next to an average EBL of 500 mL for ECOG 0 to 1. Patients undergoing systemic therapy before surgery reported an average EBL of 600 mL, compared to 900 mL for patients who did not. Corpectomies with stabilization were the type of surgery with the highest average EBL (1550 mL). Decompression had the second-highest average EBL (750 mL), followed by decompression with stabilization (597.5 mL). The lowest average EBL was reported in stabilization surgery (200 mL). Procedures in the thoracic region had the highest average EBL of 900 mL, next to 325 mL in the cervical region and 600 mL in the lumbar region (Table 1).

Prognostic Factors

In the univariate GLM analysis, significant associations with BI were found for the types of surgery in comparison to corpectomy with stabilization (decompression with stabilization P = 0.048, decompression P = 0.004, stabilization P < 0.001), combined surgical approach compared to posterior approach (P = 0.011), cervical location of operated spinal levels compared to a lumbar location (P = 0.006), and duration of surgery (P < 0.001) (Table 3). For EBL, significant associations were reported for ECOG score (ECOG 4 vs ECOG 0, P = 0.009), the types of surgery in comparison to corpectomy with stabilization (decompression with stabilization P = 0.002, decompression P = 0.002, stabilization P < 0.001), cervical location of operated spinal levels compared to a lumbar location (P = 0.041), number of operated spinal levels (P = 0.041) and duration of surgery (P < 0.001) (Table 3).

Table 3.

Univariable Generalized Linear Model (GLM)

Variables	BI			EBL
Variables	Exp (β)	95% CI	P-value	Exp (β)	95% CI	P-value
Age at surgery (y)	1.005	.994, 1.016	.394	.993	.975, 1.011	.428
Deceased at 1yr after surgery
Alive	reference			reference
Deceased	1.036	.777, 1.382	.808	.85	.526, 1.374	.508
Sex
Female	1.075	.822, 1.407	.598	1.105	.704, 1.733	.666
Male	reference			reference
BMI	.991	.966, 1.017	.503	.992	.952, 1.034	.71
Pain
Yes	1.243	.749, 2.064	.401	1.276	.497, 3.279	.613
No	reference			reference
ECOG
0	reference			reference
1	1.044	.746, 1.459	.804	1.419	.82, 2.454	.214
2	1.184	.792, 1.771	.413	1.464	.774, 2.77	.244
3	1.121	.692, 1.816	.643	1.421	.655, 3.085	.376
4	1.439	.872, 2.372	.157	2.939	1.326, 6.511	.009
ASIA
A	.427	.107, 1.698	.23	.425	.037, 4.917	.495
B	1.493	.563, 3.96	.423	2.796	.485, 16.11	.253
C	1.347	.827, 2.192	.234	1.587	.66, 3.813	.304
D	1.25	.948, 1.648	.116	1.374	.83, 2.277	.22
E	reference			reference
aCCI
0	reference			reference
1	1.054	.801, 1.388	.707	.975	.621, 1.533	.914
2	.929	.228, 3.79	.918	.395	.04, 3.876	.427
3	1.283	.388, 4.242	.684	2.542	.345, 18.728	.362
4	.438	.062, 3.087	.417	.28	.022, 3.595	.336
5	.913	.251, 3.322	.891	.669	.081, 5.547	.711
6+	1.448	.361, 5.808	.602	1.229	.13, 11.602	.858
Number of spinal lesions
1	reference			reference
2	1.105	.723, 1.687	.646	1.152	.597, 2.222	.674
3 or more	.851	.645, 1.123	.257	.672	.434, 1.042	.078
Number of skull lesions
Yes	1.181	.651, 2.143	.585	1.02	.384, 2.708	.969
No	reference			reference
Previous local RT
Yes	1.2	.918, 1.57	.186	1.135	.723, 1.779	.583
No	reference			reference
Previous systemic therapy
Yes	.837	.64, 1.094	.195	.699	.438, 1.114	.135
No	reference			reference
Albumine (g/dL)	.946	.777, 1.151	.578	.95	.69, 1.307	.752
Alkaline phosphatase (U/l)	1	.997, 1.004	.832	1	.994, 1.007	.905
Calcium (mg/dL)	1.068	.948, 1.203	.28	1.058	.869, 1.288	.575
Creatinine (mg/dL)	1.088	.927, 1.278	.305	1.033	.793, 1.347	.808
Lymphocyte count (x10^⁹/L)	1.034	.903, 1.182	.631	1.059	.844, 1.33	.619
Neutrophil count (x10^⁹/L)	.982	.959, 1.005	.132	.99	.953, 1.029	.616
Platelet count (x10^⁹/L)	.999	.998, 1	.105	1	.998, 1.003	.73
PT	1.007	.901, 1.126	.901	1.157	.964, 1.389	.121
PT/INR	1.37	.432, 4.343	.594	2.683	.407, 17.714	.308
PTT	.995	.972, 1.019	.697	1.004	.968, 1.043	.819
White blood cell count (x10^⁹/L)	1.007	.976, 1.039	.664	1.011	.961, 1.064	.663
Surgery type
Corpectomy with stabilization	reference			reference
Decompression with stabilization	.752	.569, .995	.048	.506	.333, .771	.002
Decompression	.539	.359, .811	.004	.372	.204, .678	.002
Stabilization	.404	.258, .635	< .001	.181	.093, .353	< .001
Surgical approach
Anterior	1.394	.948, 2.049	.094	1.676	.878, 3.201	.12
Posterior	reference			reference
Combined	.441	.238, .817	.011	.406	.144, 1.143	.091
Location of operated levels
Cervical	.533	.344, .825	.006	.470	.23, .96	.041
Thoracic	.906	.636, 1.291	.587	.792	.441, 1.421	.436
Lumbar	reference			reference
Number of operated spinal levels
1	reference			reference
2+	.806	.613, 1.06	.126	.633	.41, .977	.041
Time from diagnosis to primary surgery (days)	1	1, 1	.699	1	1, 1	.143
Duration of surgery (mins)	1.002	1.001, 1.003	< .001	1.003	1.002, 1.004	< .001
Drain output (ml)^a	1	1, 1.001	.096	1	1, 1.001	.092

Abbreviations: aCCI, age-adjusted charlson comorbidity index; ASIA, american spinal cord injury association; BI, bleeding index; BMI, body mass index; CI, confidence interval; EBL, visually estimated blood loss; ECOG, eastern cooperative oncology group; PT, prothrombin test; PT/INR, PT expressed as the international standardized ratio; PTT, partial thromboplastin time; RT, radiation therapy.

Bold indicates statistical significance (P < 0.05).

^aDrains were placed in 54 patients (34.2%), values presented reflect this subgroup only.

The following potential clinically relevant predictors that showed significance in the univariate analysis were used for the full multivariable GLM analyses: ECOG score, type of surgery, number and location of operated levels (comparison of only cervical, thoracic, and lumbar regions), and duration of surgery. Previous research established BMI⁹ and coagulation abnormalities such as platelet count and PT/INR 2 as additional potential predictors^24,25. Time from diagnosis to primary surgery was added as an indication of disease burden and duration (next to ECOG score). Although surgical approach was significant in the univariate analysis, it was largely determined by surgery type and therefore not included in the multivariable models (Table A1).

In the simplified GLM models, the ECOG score (ECOG 0 and 1 vs ECOG 2, 3, and 4) was dichotomized. The parsimonious GLM models consisted of the ECOG score (dichotomized), type of surgery, number and location of operated levels (comparison of only cervical, thoracic, and lumbar regions), and duration of surgery. The simplified and parsimonious models yielded similar model performance to the full models (Δpseudo-R² < 0.02). To retain full explanatory performance while avoiding the risk of overfitting, the simplified models were applied. For transparency, the full and parsimonious models were presented in the Appendix (Table A2-A5).

The simplified multivariable GLM model with BI identified platelet count, type of surgery, location of operated spinal levels (cervical vs lumbar), and duration of surgery as independent predictors of expected BI (Table 4). Each platelet count unit (x10^9/L) decrease was associated with a 0.2% increase in expected BI (exp(b) = 0.998, 95% CI 0.997-0.999, P = 0.003). Compared to corpectomy with stabilization, decompression with stabilization was associated with a 26.5% reduction (exp(b) = 0.735, 95% CI 0.579 - 0.933, P = 0.013) of expected BI, decompression was associated with a 37.3% reduction (exp(b) = 0.627, 95% CI 0.439 - 0.897, P = 0.012) and stabilization with a 39% reduction (exp(b) = 0.61, 95% CI 0.405 - 0.917, P = 0.02) (Table 4). Surgery in the cervical spine was associated with a 40.3% reduction of expected BI compared to surgery in the lumbar region (exp(b) = 0.597, 95% CI 0.417 - 0.857, P = 0.006). A longer duration of surgery was positively associated with blood loss, with each 10-minute increase corresponding to a 2% increase in expected BI (exp(b) = 1.020, 95% CI: 1.011 - 1.025, P < 0.001) (Table 4).

Table 4.

Simplified Multivariable Generalized Linear Model (GLM) for the Association Between Selected Predictors and BI

Variables	Exp (b)	95% CI	P-value
Intercept	5.903	1.895, 18.392	.003
BMI	.994	.974, 1.015	.566
ECOG score
ECOG 0 + 1	reference
ECOG 2 + 3 + 4	1.027	.824, 1.278	.815
Platelet count (x10^9/L)	.998	.997, .999	.003
PT/INR	.895	.333, 2.405	.827
Surgery type
Corpectomy with stabilization	reference
Decompression with stabilization	.735	.579, .933	.013
Decompression	.627	.439, .897	.012
Stabilization	.61	.405, .917	.02
Location of operated levels
Cervical	.597	.417, .857	.006
Thoracic	1.039	.767, 1.407	.807
Lumbar	reference
Number of operated levels
1	reference
2+	1.132	.884, 1.449	.33
Time from diagnosis to primary surgery (days)	1	1, 1	.588
Duration of surgery (mins)	1.002	1.001, 1.003	< .001

Abbreviations: BI, bleeding index; BMI, body mass index; CI, confidence interval; ECOG, eastern cooperative oncology group; PT/INR, prothrombin test (PT) expressed as the international standardized ratio (INR).

Bold indicates statistical significance (P < 0.05).

For EBL, corpectomy with stabilization was also associated with a higher expected EBL compared to the other types of surgery; decompression with stabilization was associated with a 47.6% reduction (exp(b) = 0.524, 95% CI 0.36 - 0.762, P = 0.001), decompression with a 51.7% reduction (exp(b) = 0.483, 95% CI 0.281 - 0.832, P = 0.01), and stabilization with a 67.1% reduction (exp(b) = 0.329, 95% CI 0.176 - 0.614, P = 0.001) (Table 5). A longer duration of surgery was positively associated with blood loss, with each 10-minute increase corresponding to a 3% increase in expected EBL (exp(b) = 1.030, 95% CI: 1.017 - 1.038, P < 0.001) (Table 5). Unlike BI, a higher ECOG score (2-4 compared to 0-1) was associated with a higher expected EBL (exp(b) = 1.426, 95% CI 1.015 - 2.002, P = 0.043), while no signification associations were found for the surgical location (Table 5).

Table 5.

Simplified Multivariable Generalized Linear Model (GLM) for the Association Between Selected Predictors and EBL

Variables	Exp (b)	95% CI	P-value
Intercept	216.205	39.685, 1177.894	<.001
BMI	1.004	.973, 1.036	.814
ECOG score
ECOG 0 + 1	reference
ECOG 2 + 3 + 4	1.426	1.015, 2.002	.043
Platelet count (x10^⁹/L)	.999	.997, 1.001	.224
PT/INR	2.049	.466, 9.018	.345
Surgery type
Corpectomy with stabilization	reference
Decompression with stabilization	.524	.36, .762	.001
Decompression	.483	.281, .832	.01
Stabilization	.329	.176, .614	.001
Location of operated levels
Cervical	.632	.362, 1.102	.109
Thoracic	1.28	.798, 2.051	.309
Lumbar	reference
Number of operated levels
1	reference
2+	1.106	.751, 1.628	.611
Time from diagnosis to primary surgery (days)	1	1, 1	.414
Duration of surgery (mins)	1.003	1.001, 1.004	< .001

Abbreviations: BMI, Body Mass Index; CI, confidence interval; EBL, visually estimated blood loss; ECOG, Eastern Cooperative Oncology Group; PT/INR, prothrombin test (PT) expressed as the International Standardized Ratio (INR).

Bold indicates statistical significance (P < 0.05).

Discussion

Several factors associated with high intraoperative blood loss for patients undergoing MBD-related spine surgery in the spine were assessed in this study. With an average BI of 4.4, the platelet count, type of surgery, the location of operated spinal levels, and a longer duration of surgery were found to be significantly associated with higher intraoperative blood loss. Knowledge on factors that lead to higher intraoperative blood loss can be relevant during preoperative patient screening. Furthermore, knowledge of a potentially increased intraoperative bleeding risk in certain patients with MBD, or in this patient population in general, could facilitate the anesthesia and operating team in considering preventive measures and blood transfusion preparations.

To our knowledge, this is the first study examining prognostic factors for intraoperative blood loss specifically for MBD-related spine surgery. Kumar et al (2016) examined influencing factors for blood loss in metastatic spinal tumor surgery, mainly consisting of lung cancer metastases (30%) and breast cancer metastases (19%).²⁶ In that study, 13 percent of the patients that underwent surgery had MM or lymphoma lesions, grouped as hematologic metastases. Comparable to our results, a longer duration and type of surgery were significantly associated with higher intraoperative blood loss. In (thoracolumbar) posterior instrumentation and decompression, they reported more than 600 mL extra blood loss compared to minimally invasive approaches (not included in our study). Similarly, our results show increased bleeding risk with larger or more invasive procedures (such as corpectomies with stabilization). This was not found in our previous research when intraoperative bleeding in MM was compared to spinal metastases.¹¹ These extensive procedures were relatively prevalent in our patient cohort, owing to the tertiary care setting and exclusion of augmentation procedures. In an assessment of associated factors for perioperative complications on increased intraoperative bleeding in spine surgery by Lange et al, a longer duration of surgery and open procedures were considered significant intraoperative risk factors.²⁷ For 30-day postoperative complications, factors such as aCCI, number of operated levels and performance status were found to be additional significant risk factors.²⁷ Although ECOG score and time from diagnosis to surgery were not found significant univariate predictors for high BI, patients in a worse disease stage or that suffered from MM for a longer time were expected to have a higher risk of perioperative complications such as bleeding.²⁷ A possible explanation is that patients with a more advanced stage of disease were treated with less invasive procedures with concomitant less intraoperative bleeding.⁸ This could possibly coincide with the differences in baseline characteristics per surgery type (see Table A1). A higher presence of skull lesions and administered previous systemic therapy were found in patients undergoing less invasive procedures, suggesting progressed disease.

It is relevant to distinguish MBD-related spine surgery from other oncologic indications in preoperative bleeding risk assessment, since these are often grouped together.^11,28 Our previous research that compared myeloma lesions with spinal metastases from solid tumors found significantly higher intraoperative bleeding scores in MM (BI 5.6 in MM vs BI 3.9 in spinal metastases) after propensity score matching analysis.¹¹ Building on these results, the results of the current study demonstrate distinctive bleeding behaviors across MBD-related spine surgeries, highlighting the importance of researching MBD as a separate entity. Quidet et al (2018) found an average blood loss of 795.2 mL in 55 patients undergoing spine surgery for symptomatic MM-lesions in their case series on surgical success.⁷ This is comparable to the average EBL found in our study (750 mL). Although MM-specific factors (eg, coagulation and hemostatic abnormalities, hypercalcemia, renal function loss) were included in this study, vertebral and surgery-specific factors played a more dominant role.

In our results, some disparities between BI and EBL were reported. For example, a lower platelet count was significantly associated with a higher expected BI (although clinically modest), but not with EBL. ECOG score (dichotomized) showed an association with EBL, while this was not found for BI. Decompression with stabilization had the second-highest bleeding of all surgery types when the average BIs are compared. However, decompression (without stabilization) had a higher average EBL than the former (750 mL vs 597.5 mL). These different results highlight the difficulty of accurate visual bleeding estimations with residual (unseen) blood loss, especially thwarted when suction canisters or surgical sponges are used.²²

Limitations and Recommendations for Future Research

There are several limitations to this study. Due to the retrospective study design, the availability of patient-specific and surgical explanatory variables was dependent of the assessment and documentation of the caregiving team. It was unclear if any protocols were used to examine the EBL, impeding its already reduced applicability. The sample size of this study was a limiting factor for including more variables in the multivariable analysis. However, this is one of the largest cohorts to date on this specific, but clinically distinct, patient population. Future research should assess whether other factors play a role in predicting high intraoperative bleeding as well. Clinical guidelines or thresholds should be established to improve clinical decision-making.

Clinical Implications

Creating awareness on significant prognostic factors for both intraoperative bleeding and other perioperative complications can assist spine surgeons and anesthesiologists in their decision-making and preparation before surgery. While the indications for surgical intervention may not be modifiable, the findings in this study can assist in considering two surgeons to reduce the duration of surgery or improving intraoperative hemostasis, for example. However, as visual bleeding estimates are thought to be imprecise, the accuracy of EBL should be analyzed critically. In this day, with an increasingly complex patient population, more studies should enquire other methods to accurately assess bleeding, such as the BI.^21,22,29 Early identification of patients at high risk for significant intraoperative blood loss can be helpful for selecting the optimal setting for surgery, creating pathways to optimize preoperative anemia, and ensuring coordination with blood bank for availability of product. Additionally, minor or minimally invasive surgical alternatives could be considered. Of note, daratumumab is a common agent used in MM treatment and can interfere with type and screen testing which may cause delays in availability of blood products.³⁰ Coordination with the preoperative clinic and the anesthesiologist assigned to the case on high-risk patients could mitigate the potential morbidity and mortality associated with high intraoperative blood loss.

Conclusions

Supplemental Material

Supplemental material - Prognostic Factors for High Intraoperative Blood Loss for Multiple Myeloma-Related Bone Disease in the Spine

Supplemental material for Prognostic Factors for High Intraoperative Blood Loss for Multiple Myeloma-Related Bone Disease in the Spine by Jens P. te Velde, MD, Hester Zijlstra, MD, PhD, Daniël de Reus, MD, Robertus J.B. Pierik, MD, Amanda S. Xi, MD, Ganesh M. Shankar, MD, PhD, Barend J. van Royen, MD, PhD, Diederik H.R. Kempen, MD, PhD, Joseph H. Schwab, MD, MS and Daniel G. Tobert, MD in Global Spine Journal

Footnotes

ORCID iDs

Jens P. te Velde

Hester Zijlstra

Daniël de Reus

Amanda S. Xi

Daniel G. Tobert

Consent to Participate

Informed consent for retrospective analysis of de-identified data was waived.

Author Contributions

All authors were involved in the design. JPV, HZ, DR and RJBP performed the data extraction. JPV and HZ performed the data analysis. All authors reviewed and edited the manuscript.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

IRB Approval

Investigation performed at Massachusetts General Hospital, Boston, USA. Local Institutional Review Board (IRB) approval was obtained for this study (registration number 2018P000688).

Supplemental Material

Supplemental material is available online.

References

Terpos

Zamagni

Lentzsch

Bone Working Group of the International Myeloma Working Group , et al.. Treatment of multiple myeloma-related bone disease: recommendations from the bone working group of the international myeloma working group. Lancet Oncol. 2021;22:e119-e130. doi:10.1016/S1470-2045(20)30559-3

Terpos

Ntanasis-Stathopoulos

Dimopoulos

. Myeloma bone disease: from biology findings to treatment approaches. Blood. 2019;133:1534-1539. doi:10.1182/blood-2018-11-852459

Ria

Reale

De Luisi

Ferrucci

Moschetta

Vacca

. Bone marrow angiogenesis and progression in multiple myeloma. Am J Blood Res. 2011;1:76-89.

Zijlstra

Wolterbeek

Ponds

NHM

, et al. The incidence of vertebral compression fractures and spinal instability in newly diagnosed multiple myeloma patients. J Orthop. 2023;38:62-67. doi:10.1016/j.jor.2023.03.008

te Velde

Zijlstra

Lans

, et al. Fracture rate after conventional external beam radiation therapy to the spine in multiple myeloma patients. Spine J. 2023;24:137-145. doi:10.1016/j.spinee.2023.09.009

Milavec

Ravikumar

Syn

Soekojo

Chng

Kumar

. Surgical management of multiple myeloma with symptomatic involvement of the spine. Int J Spine Surg. 2020;14:785-794. doi:10.14444/7112

Quidet

Zairi

Boyle

, et al. Evaluation of the relevance of surgery in patients with multiple myeloma harboring symptomatic spinal involvement: a retrospective case series. World Neurosurg. 2018;114:e356-e365.

Zijlstra

Pierik

Crawford

, et al. Analysis of complications and revisions after spine surgery in 270 multiple myeloma patients with spinal involvement. Eur Spine J. 2023;32:4335-4354. doi:10.1007/s00586-023-07903-1

Mohme

Mende

Pantel

, et al. Intraoperative blood loss in oncological spine surgery. Neurosurg Focus. 2021;50:E14. doi:10.3171/2021.2.FOCUS201117

10.

Chen

Tai

Nayak

, et al. Blood loss in spinal tumour surgery and surgery for metastatic spinal disease: a meta-analysis. Bone Jt J. 2013;95-B:683-688. doi:10.1302/0301-620X.95B5.31270

11.

te Velde

Zijlstra

Pierik

RJB

, et al.

Is surgery for multiple myeloma-related spinal involvement associated with higher blood loss compared to spinal metastases?

Eur Spine J 2025;Oct 7. doi:10.1007/s00586-025-09294-x

41053490

12.

von Elm

Altman

Egger

Pocock

Gøtzsche

Vandenbroucke

STROBE Initiative . The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61:344-349. doi:10.1016/j.jclinepi.2007.11.008

13.

Rupp

Biering-Sørensen

Burns

, et al. International standards for neurological classification of spinal cord injury. Top Spinal Cord Inj Rehabil. 2021;27:1-22. doi:10.46292/sci2702-1

14.

Charlson

Szatrowski

Peterson

Gold

. Validation of a combined comorbidity index. J Clin Epidemiol. 1994;47:1245-1251. doi:10.1016/0895-4356(94)90129-5

15.

Oken

Creech

Tormey

, et al. Toxicity and response criteria of the eastern cooperative oncology group. Am J Clin Oncol. 1982;5:649-655.

16.

van den Besselaar

AMHP

Chantarangkul

Tripodi

. Thromboplastin standards. Biologicals. 2010;38:430-436. doi:10.1016/j.biologicals.2010.02.012

17.

Lawton

Morrey

. The use of heparin in patients in whom a pulmonary embolism is suspected after total hip arthroplasty. JBJS. 1999;81:1063-1072.

18.

Haighton

Kempen

DHR

Wolterbeek

Marting

van Dijk

Veen

RMR

. Bridging therapy for oral anticoagulation increases the risk for bleeding-related complications in total joint arthroplasty. J Orthop Surg Res. 2015;10:145. doi:10.1186/s13018-015-0285-6

19.

George

Beaty

Kilic

, et al. Hemoglobin drift after cardiac surgery. Ann Thorac Surg. 2012;94:703-709. doi:10.1016/j.athoracsur.2012.03.038

20.

Jaramillo

Montane-Muntane

Capitan

, et al. Agreement of surgical blood loss estimation methods. Transfusion (Arlingt). 2019;59:508-515. doi:10.1111/trf.15052

21.

Donovan

Lostis

Jones

Whitehouse

. Estimation of blood volume and blood loss in primary total hip and knee replacement: an analysis of formulae for perioperative calculations and their ability to predict length of stay and blood transfusion requirements. J Orthop. 2021;24:227-232. doi:10.1016/j.jor.2021.03.004

22.

Guinn

Broomer

White

Richardson

Hill

. Comparison of visually estimated blood loss with direct hemoglobin measurement in multilevel spine surgery. Transfusion (Arlingt). 2013;53:2790-2794. doi:10.1111/trf.12119

23.

Cui

Shi

, et al. Risk factors of total blood loss in the posterior surgery for patients with thoracolumbar metastasis. BMC Muscoskelet Disord. 2021;22:898. doi:10.1186/s12891-021-04789-2

24.

Uhl

Assmann

Hamza

Harrison

Gernsheimer

Slichter

. Laboratory predictors of bleeding and the effect of platelet and RBC transfusions on bleeding outcomes in the PLADO trial. Blood. 2017;130:1247-1258. doi:10.1182/blood-2017-01-757930

25.

Coppola

Tufano

Di Capua

Franchini

. Bleeding and thrombosis in multiple myeloma and related plasma cell disorders. Semin Thromb Hemost. 2011;37:929-945. doi:10.1055/s-0031-1297372

26.

Kumar

Zaw

Khine

, et al. Blood loss and transfusion requirements in metastatic spinal tumor surgery: evaluation of influencing factors. Ann Surg Oncol. 2016;23:2079-2086. doi:10.1245/s10434-016-5092-8

27.

Lange

Stadtmüller

Scheibel

, et al. Analysis of risk factors for perioperative complications in spine surgery. Sci Rep. 2022;12:14350. doi:10.1038/s41598-022-18417-z

28.

Amelot

Terrier

L-M

Le Nail

L-R

, et al. Multiple myeloma spinal lesion care: management of a primary bone malignancy rather than a spinal metastasis. World Neurosurg. 2023;176:e680-e685. doi:10.1016/j.wneu.2023.05.118

29.

Gerdessen

Meybohm

Choorapoikayil

, et al. Comparison of common perioperative blood loss estimation techniques: a systematic review and meta-analysis. J Clin Monit Comput. 2021;35:245-258. doi:10.1007/s10877-020-00579-8

30.

Arnall

Maples

Harvey

Moore

. Daratumumab for the treatment of multiple myeloma: a review of clinical applicability and operational considerations. Ann Pharmacother. 2022;56:927-940. doi:10.1177/10600280211058754

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.28 MB