Investigating the Relationship Between High Density Lipoprotein Cholesterol and Fusion Outcomes Following Single Level Lumbar Fusion: A Propensity Matched Analysis

Abstract

Study Design

Retrospective cohort.

Objectives

Although dyslipidemia has been implicated in impaired bone healing, the role of high-density lipoprotein cholesterol (HDL-C) in spinal fusion outcomes remains poorly defined. Emerging evidence suggests that extremely elevated HDL-C levels may be associated with adverse osseous events. The objective of this study was to evaluate the association between HDL-C levels and rates of pseudoarthrosis and reoperation following single-level lumbar interbody fusion.

Methods

The TriNetX Research Network was queried for adult patients who underwent single level lumbar fusion during the 20-years period ended December 2, 2023. The high HDL-C cohort was designated by HDL-C ≥70.1 mg/dL, reflecting one standard deviation above the mean for all patients in the database. One standard deviation below the mean denoted the low HDL-C group (≤32.1 mg/dL). Propensity score matching was performed 1:1 according to demographics and comorbidities. Pseudoarthrosis diagnosis and reoperation were assessed at 6 months, 1 year, and 2 years postoperatively utilizing risk ratios (RR) and 95% confidence intervals (CI).

Results

After propensity matching, 1316 patient pairs were included. Patients with elevated HDL-C demonstrated significantly higher rates of pseudoarthrosis at 6 months, 1 year, and 2 years postoperatively. Early reoperation risk was also significantly increased in the high HDL-C cohort within 6 months, though differences were not sustained at later time points.

Conclusions

Elevated HDL-C was independently associated with increased risk of pseudoarthrosis following single-level lumbar fusion. These findings suggest that HDL-C may represent an underrecognized metabolic risk factor for impaired spinal fusion and warrants further investigation into underlying mechanisms and its possible utility in preoperative optimization programs.

Introduction

Lumbar spinal fusion is a commonly performed surgical procedure indicated for a range of degenerative, traumatic and deformity-related spinal conditions. Despite continual improvements in surgical technique and perioperative management, pseudoarthrosis remains a significant complication, occurring in 1-12% of lumbar fusion cases with variation due to complexity of cases and patient populations.^1–3 The consequences of pseudoarthrosis are multifaceted with many patients experiencing persistent axial or radicular pain alongside decreased function.^4–6 In turn, this pain and disability has been shown to be associated with delayed return to work and a prolonged dependence on narcotics use. ⁷ Further, additional surgical interventions may also be required in up to 12% of cases. ⁴ Thus, pseudoarthrosis contributes significantly to poor outcomes following spinal fusion procedures.

Numerous risk factors for pseudoarthrosis have been described including advanced age, body mass index (BMI), smoking, diabetes mellitus, and osteoporosis.^2,8–11 More recently, serum lipids have become an area of interest due to their potential role in bone metabolism and the possibility of pharmacologic modification. In particular, dyslipidemia may stimulate various signaling pathways, impairing bone regeneration and strength.^12–14 Specifically, elevated cholesterol is associated with decreased bone mineral density (BMD) alongside reduced osteoblast differentiation and markers of bone formation such as Runx2, collagen 1A, and endogenous bone morphogenic protein-2 (BMP-2).^15,16 Likewise, several large database studies have demonstrated an association between elevated low density lipoprotein (LDL-C) levels and increased rates of pseudoarthrosis following both single-level lumbar interbody fusion and anterior cervical discectomy and fusion (ACDF), suggesting lipid metabolism may contribute to spinal fusion outcomes.^17,18 This association is further evidenced by the protective effects conferred by statin therapy which is associated with increased rates of successful bony fusion following these procedures.^17,18

In contrast, the role of high-density lipoprotein cholesterol (HDL-C) in bone healing and spinal fusion is not well defined. Although HDL-C has largely been viewed as cardioprotective, emerging literature suggests that the relationship between HDL-C and clinical outcomes may not be linear, but U-shaped.^19,20 Likewise, extremely elevated levels of HDL-C have been associated with increased all-cause mortality, reduced bone mineral density (BMD), and increased fracture risk.^21,22 Experimental evidence further suggests that certain HDL-C particle subtypes (OxHDL) may reduce osteoblast mineralization through proinflammatory pathways. ²³ This raises the possibility that elevated HDL-C could adversely affect spinal fusion rates, although its utility as a modifiable risk factor in this patient population is likely limited due to its other advantageous effects. ¹⁹

Nonetheless, given the emerging evidence that extremely elevated HDL-C may exert paradoxical effects on bone metabolism, we sought to evaluate whether high HDL-C levels are associated with impaired fusion outcomes following lumbar interbody fusion. The purpose of this study was to assess the relationship between HDL-C levels and rates of pseudoarthrosis and reoperation among adults undergoing single-level posterior or transforaminal lumbar interbody fusion using a large, multicenter database. We hypothesized that patients with elevated HDL-C would experience higher rates of pseudoarthrosis and early reoperation compared with those with low HDL-C.

Materials and Methods

Data Source

This study used the TriNetX Research Network database (Cambridge, Massachusetts), a global healthcare data network that compiles real-time, de-identified electronic health record data from approximately 120 healthcare organizations (HCOs) that represent over 250 million patients. ²⁴ Participating HCOs authorize the integration of their records with the TriNetX platform in exchange for access to its functional capabilities, which include clinical trial protocol design, database query for observational study purposes, and drug safety evaluation.

Ethical Approval

The TriNetX Research Network is compliant with the Health Insurance Portability and Accountability Act of 1996 (HIPAA) Security Rule and certified according to the ISO 27001:2013 standard. For these reasons, this study was deemed exempt from institutional review board approval and informed consent due to minimal risk of patient privacy compromise. This study was designed and reported in accordance with Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Cohort Selection

All data included in this analysis were obtained on December 2, 2025 from the TriNetX Research Network database within the 20-years period beginning December 2, 2003, and ending December 2, 2023. This temporal constraint assured the availability of at least 2 years of follow up data for every patient included in the study. Inclusion criteria, exclusion criteria, propensity score matching criteria, and study endpoints utilized classifications outlined by the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM), International Classification of Diseases, 10th Revision, Procedure Coding System (ICD-10-PCS), Current Procedural Terminology (CPT), and RxNorm medical billing codes. First, patients between the ages of 18 and 89 who underwent single-level posterior or transforaminal lumbar interbody fusion were identified (ICD-10-PCS 0SG00AJ, 0SG007 J, 0SG00 KJ, 0SG00JJ, 0SG03 KJ, 0SG037 J or CPT 22630, 22633). Patients with a history of multi-level fusion were excluded from the analysis (CPT 22632 or 22634). Next, patients were separated into high and low HDL-C cohorts, denoted as one standard deviation above or below the mean HDL-C of all patients included in the TriNetX database. This approach has been utilized by prior studies investigating the relationship between low density lipoprotein cholesterol (LDL-C) and spinal fusion outcomes with the TriNetX database.^17,18 The high HDL-C group contained patients with a serum HDL-C ≥70.1 mg/dL whereas the low HDL-C group included patients with a serum HDL-C ≤32.1 mg/dL. The absolute magnitude of difference between the low and high HDL-C cohorts is consistent with that identified in studies of the U-shaped association between HDL-C and cardiovascular disease, with the risk of all-cause mortality increasing above 80 mg/dL and below 40 mg/dL. ²⁵ There were 2111 subjects in the high HDL-C group and 2813 subjects in the low HDL-C group [Figure 1].

Figure 1.

Flowchart demonstrating query design. Abbreviations - PLIF, posterior lumbar interbody fusion; TLIF, transforaminal lumbar interbody fusion; HDL-C, high density lipoprotein cholesterol; PSM, propensity score matching; BMI, body mass index

Propensity Score Matching

Next, the two cohorts underwent 1:1 propensity matching to reduce the risk of confounding covariates. The baseline characteristics chosen for matching included age at index, sex, race, ethnicity, body mass index (BMI), hemoglobin A1c (HbA1c), essential hypertension, nicotine dependence, diabetes mellitus, chronic ischemic heart disease, atrial fibrillation and flutter, other cardiac arrhythmias, osteoporosis without current pathological fracture, and concomitant statin therapy. Successful balancing of the two cohorts was reflected by a standard difference less than or equal to 0.1 for each of the selected covariates.

Outcome of Interest

Primary outcomes included the diagnosis of pseudoarthrosis, as defined by ICD-10-CM M96.0, and subsequent reoperation. Mean HDL-C and LDL-C were also assessed at each time interval. Outcomes were evaluated at 6 months, 1 year, and 2 years following single-level PLIF or TLIF.

Statistical Analysis

Baseline characteristics were evaluated with descriptive statistics. Binary outcomes for each endpoint were assessed using relative risk (RR) and 95% confidence intervals (CI). The Student’s t-test was utilized to compare differences between continuous variables. All statistical procedures were automatically executed within the TriNetX research platform. A two-sided P-value <0.05 was considered to be statistically significant.

Results

There were 1386 pairs included in the final study population. Baseline characteristics and outcomes of the propensity score match are contained in Table 1. The mean age at the time of surgery was approximately 64 years. Female patients represented 72.7% and 73.4% of the high HDL-C and low HDL-C cohorts, respectively, whereas 27.2% and 26.6% of subjects were male. A majority of all matched pairs were White (76.9%) compared with 13.9% Black or African American and approximately 4.0% Hispanic or Latino. After matching, there were no differences between the two groups with respect to any of the covariates included in the propensity score except mean BMI. The mean BMI of patients in the high HDL-C cohort was 29.7 ± 6.2 kg/m² compared with 30.6 ± 6.7 kg/m² in the low HDL-C cohort. However, the absolute magnitude of this difference is likely below clinically meaningful thresholds. Rates of statin therapy were equivalent at baseline (all standard difference <0.1).

Table 1.

Baseline Demographics and Propensity Score Matching Outcomes

Characteristic	Before matching				After matching
	High HDL-C	Low HDL-C	P-value	Std diff	High HDL-C	Low HDL-C	P-value	Std diff
Demographics
Age at index	64.6 (9.7)	62.8 (11.2)	<0.001	0.169	63.9 (9.8)	64.0 (10.9)	0.761	0.012
Female	1723 (81.9)	1225 (42.7)	<0.001	0.883	1007 (72.7)	1017 (73.4)	0.669	0.016
Male	379 (18.0)	1627 (56.7)	<0.001	0.874	377 (27.2)	368 (26.6)	0.700	0.015
Black or African American	246 (11.7)	348 (12.1)	0.632	0.014	191 (13.8)	194 (14.0)	0.869	0.006
White	1686 (80.1)	2173 (75.8)	<0.001	0.105	1065 (76.8)	1067 (77.0)	0.928	0.003
Hispanic or latino	79 (3.8)	132 (4.6)	0.142	0.042	60 (4.3)	51 (3.7)	0.383	0.033
Comorbidities
Essential hypertension	1412 (67.1)	2247 (78.4)	<0.001	0.255	1017 (73.4)	1018 (73.4)	0.966	0.002
Nicotine dependence	315 (15.0)	666 (23.2)	<0.001	0.211	250 (18.0)	245 (17.7)	0.804	0.009
Diabetes mellitus	471 (22.4)	1290 (45.0)	<0.001	0.493	421 (30.4)	429 (31.0)	0.742	0.013
Chronic ischemic heart disease	340 (16.2)	848 (29.6)	<0.001	0.324	281 (20.3)	276 (19.9)	0.813	0.009
Atrial fibrillation and flutter	144 (6.8)	314 (11.0)	<0.001	0.145	111 (8.0)	118 (8.5)	0.629	0.018
Other cardiac arrhythmia	365 (17.3)	557 (19.4)	0.062	0.054	252 (18.2)	257 (18.5)	0.806	0.009
Osteoporosis without current pathological fracture	469 (22.3)	297 (10.4)	<0.001	0.327	236 (17.0)	229 (16.5)	0.722	0.014
Mean BMI (kg/m2)	28.4 (6.2)	31.9 (6.4)	<0.001	0.558	29.7 (6.2)	30.6 (6.7)	0.001	0.137
Mean HbA1c (%)	5.8 (0.9)	6.4 (1.5)	<0.001	0.474	6.0 (1.0)	6.0 (1.4)	0.847	0.009
Medications
Rosuvastatin	243 (11.5)	419 (14.6)	0.002	0.091	176 (12.7)	173 (12.5)	0.864	0.007
Simvastatin	318 (15.1)	607 (21.2)	<0.001	0.158	250 (18.0)	256 (18.5)	0.768	0.011
Fluvastatin	10 (0.5)	10 (0.3)	0.486	0.020	10 (0.7)	10 (0.7)	1	<0.001
Pravastatin	195 (9.3)	272 (9.5)	0.794	0.008	140 (10.1)	124 (8.9)	0.301	0.039
Lovastatin	42 (2.0)	60 (2.1)	0.812	0.007	28 (2.0)	29 (2.1)	0.894	0.005
Atoravastatin	646 (30.7)	1180 (41.2)	<0.001	0.219	478 (34.5)	475 (34.3)	0.905	0.005
Pitavastatin	14 (0.7)	20 (0.7)	0.892	0.004	10 (0.7)	10 (0.7)	1	<0.001

Legend - High HDL-C cohort includes patients with serum HDL-C ≥70.1 mg/dL. Low HDL-C cohort includes patients with serum HDL-C ≤32.1 mg/dL. Std diff <0.1 indicates appropriate balance after matching. Abbreviations - HDL-C, high density lipoprotein cholesterol; std diff, standard difference; BMI, body mass index.

Outcomes of the primary analysis are included in Table 2. Patients with elevated preoperative HDL-C demonstrated consistently higher incidence of pseudoarthrosis at each time interval assessed. Within 6 months postoperatively, the rate of pseudoarthrosis was 7.5% in the high HDL-C cohort compared with 4.5% in the low HDL-C group (RR 1.677, 95% CI 1.236-2.277, P = 0.001). Further, at final 2-years follow-up, the likelihood of pseudoarthrosis was nearly 40% greater amongst patients with high HDL-C (RR 1.358, 95% CI 1.072-1.719, P = 0.011). In contrast, HDL-C demonstrated less robust correlations regarding reoperation risk. Whereas the incidence of subsequent reoperation was more than doubled in the high HDL-C vs low HDL-C cohort within 6 months (RR 2.188, 95% CI 1.216-3.934, P = 0.007), no significant differences were observed at 1 or 2 years postoperatively. Lastly, mean serum LDL-C levels did not differ between the two groups at any time interval and there was a persistent delta in serum HDL-C of approximately 40 mg/dL maintained across each analysis [Table 3].

Table 2.

Comparison of Pseudoarthrosis and Reoperation Rates Between High and Low HDL-C Cohorts

Complication	Risk in high HDL-C	Risk in low HDL-C	Risk difference	Risk ratio (95% CI)	P-value
Pseudoarthrosis
Six months	104 (7.5%)	62 (4.5%)	3.0%	1.677 (1.236-2.277)	0.001
One year	109 (8.3%)	74 (5.6%)	2.7%	1.473 (1.108-1.959)	0.007
Two years	148 (10.7%)	109 (7.9%)	2.8%	1.358 (1.072-1.719)	0.011
Reoperation
Six months	35 (2.5%)	16 (1.2%)	1.3%	2.188 (1.216-3.934)	0.007
One year	47 (3.6%)	34 (2.6%)	1.0%	1.382 (0.895-2.135)	0.142
Two years	75 (5.4%)	62 (4.5%)	0.9%	1.210 (0.871-1.679)	0.255

Legend – risk is expressed as the number of patients with the outcome (proportion of cohort). Bold-face type indicates statistical significance. High HDL-C cohort includes patients with serum HDL-C ≥70.1 mg/dL. Low HDL-C cohort includes patients with serum HDL-C ≤32.1 mg/dL. Abbreviations - HDL-C, high-density lipoprotein cholesterol; CI, confidence interval.

Table 3.

Comparison of Mean HDL-C and LDL-C Between High and Low HDL-C Cohorts

Mean HDL-C	High HDL-C cohort	Low HDL-C cohort	P-value
Six months	62.2 (30.4)	15.8 (20.0)	<0.001
One year	65.8 (30.5)	16.4 (19.9)	<0.000
Two years	67.3 (29.1)	18.8 (21.1)	<0.001

Legend - values expressed as mean (standard deviation) in mg/dL. Bold-face type represents statistical significance. Abbreviations - High HDL-C cohort includes patients with serum HDL-C ≥ 70.1 mg/dL. Low HDL-C cohort includes patients with serum LDL-C ≤ 32.1 mg/dL. Abbreviations - HDL-C, high-density lipoprotein cholesterol; LDL-C, low density lipoprotein cholesterol; CI, confidence interval.

Discussion

Pseudoarthrosis is a devastating complication that can result in persistent pain, immobility, and additional surgical intervention.^2,3,6,7 For these reasons, modifiable risk factors for pseudoarthrosis are of particular interest to orthopaedic practitioners.^2,8–11 Moreover, elevated LDL-C and HDL-C have been shown to have deleterious effects on bone metabolism with the former associated with increased rates of pseudoarthrosis in both lumbar interbody fusion and ACDF.^{17,18,22,26,27} Herein we extend this work by examining the relationship between serum HDL-C and nonunion risk amongst patients who underwent posterior or lateral lumbar interbody fusion. In this large, propensity-matched analysis, elevated HDL-C was associated with higher rates of pseudoarthrosis following single-level lumbar interbody fusion. Patients with high HDL-C demonstrated a consistently increased risk of nonunion across all evaluated time points, as well as a significantly higher incidence of early reoperation. Interestingly, the magnitude of risk elevation between high HDL-C and unsuccessful fusion is comparable to that of more generally accepted risk factors in this population according to recent meta-analyses, such as smoking, age, and the number of adjacent levels fused.^28,29 In addition, the correlation between HDL-C and early reoperation only is likely explained by the differing mechanisms of fusion failure at various postoperative intervals. For example, osteoporosis is a known risk factor for early unplanned reoperation following lumbar fusion. ³⁰ Therefore, given the negative relationship between HDL-C and BMD, patients with high HDL-C may be at heightened risk of reoperation for reasons related to poor bone quality, such as hardware failure, in the near-term postoperative period whereas persistent degenerative disease often contributes to longer-term reoperation rates.^22,30,31 Notably, these associations persisted despite equivalent LDL-C levels and statin exposure between cohorts which also influence union rates in this patient population.^17,18

These findings are consistent with the existing literature. While there is wide variation in the incidence of pseudoarthrosis following single level lumbar fusion, our data are in line with a 2016 meta-analysis by Noshchenko et al which reported a 5.5% radiographic non-union rate. ³² Operative non-union rates are lower, approximately 1.0% within 2 years, whereas the 2-years reoperation rate in our study was roughly 5.0%. ³³ This difference may be attributed to broader indications for reoperation, as we include patients who underwent an additional procedure for numerous indications, which include adjacent segment disease and hardware failure as well as symptomatic pseudoarthrosis. ³⁴ Additionally, a similarly designed study by Lavu et al 2023 investigating LDL-C and nonunion risk following single level lumbar fusion reported associations of similar magnitude, ranging between a 1.6 and 1.7-fold risk elevation compared with the 1.4 to 1.7-fold elevation in our study. ¹⁷ Interestingly, despite similar methodology the absolute rate of non-union was approximately doubled in patients with high or low LDL-C, suggesting that HDL-C may represent a lower yield risk factor that is also less readily modifiable than LDL-C. ¹⁷

The idea that high as opposed to low HDL-C may convey poor outcomes for patients who undergo lumbar spinal fusion is somewhat paradoxical given the relationship between HDL-C and risk of cardiovascular events. ²¹ Although HDL-C has traditionally been regarded as cardioprotective, accumulating evidence suggests that the relationship between HDL-C and clinical outcomes is nonlinear.^21,35 The underlying mechanism of these observations is yet to be fully elucidated, but they are believed to be at least partially explained by wide variability in HDL-C surface protein expression, some of which are prothrombotic or proinflammatory. ³⁶ It is thus plausible that elevated HDL-C may confer a global inflammatory state which is known to impair bony union. ³⁷ HDL-C particles are also responsible for removing oxysterols from the systemic circulation, some of which work in tandem with BMP-2 to induce osteoblastogenesis.^38,39 Likewise, excessive circulating amounts of HDL-C are prone to oxidation which reduces osteoblast differentiation while promoting osteoclast activity.^40,41 This relationship is further evidenced by various clinical studies in which HDL-C is a risk factor for both reduced BMD and incidence of fracture.^22,42 For example, Hussain et al 2023 reported a linear association between serum HDL-C and fracture incidence, with risk elevation beginning above 58 mg/dL. ²² This closely mirrors our data, as patients in the high HDL-C cohort had a mean HDL-C between 60 and 70 mg/dL. These findings provide a plausible biologic explanation for the increased pseudoarthrosis rates observed in patients with elevated HDL-C in the present study.

The observed relationship between HDL-C and non-union rates following single-level lumbar fusion warrants careful consideration by the orthopaedic practitioner. Although it is true that pseudoarthrosis is most clinically relevant when it indicates reoperation, radiographic non-unions still influence patient outcomes. ³² Amongst patients with non-union following single-level lumbar interbody fusion that did not require revision surgery, Oswestry Disability Index (ODI) scores and Numeric Rating Scales back pain scores were substantially poorer than those who achieved union. ³² From a clinical standpoint, elevated HDL-C may represent a previously underappreciated metabolic risk factor for impaired fusion following lumbar interbody procedures. While HDL-C itself is not readily modifiable and should not be targeted for reduction, awareness of elevated HDL-C may help identify patients at increased risk for pseudoarthrosis. In such patients, surgeons may consider heightened vigilance, optimization of other modifiable risk factors, or selective use of osteoinductive adjuncts.^43,44 However, these findings do not support routine alteration of surgical strategy based solely on HDL-C levels, and prospective studies are needed before such measures can be broadly recommended.

Our study has several distinct qualities that improve the robustness of its findings. First, we employed a de-identified database that incorporates electronic health records from both academic and non-academic HCOs that are located in numerous countries across the world. This allows for broad generalizability of our results to various different patient populations and demographics. The inclusion of thousands of patients lowers the likelihood of Type II error. We employed a rigorous propensity score matching criteria and achieved balance across several notable risk factors for nonunion following lumbar fusion including age, diabetes, osteoporosis, statin use, and smoking status.^17,28,29,45

However, this retrospective analysis also confers numerous inherent limitations that necessitate explicit acknowledgement. First, our study is purely observational in nature which precludes causal claims. Second, although propensity-score matching was performed, there is a possibility of additional confounding variables that the matching process is unable to account for. Specifically, other patient-specific characteristics that may influence pseudoarthrosis rates include chronic steroid use, underlying bone quality, postoperative compliance with surgeon and rehabilitative protocols, baseline nutritional status, and the exact segmental location at which the fusion occurred. ² Mean BMI was not successfully balanced after matching, although the absolute magnitude of this difference is lower than that which has been shown to increase pseudoarthrosis risk. ⁹ We also cannot account for variation in surgical technique which may influence outcomes, such as the incorporation of adjunctive materials, addition of bone grafting, or surgeon experience. Moreover, the TriNetX platform is dependent upon medical billing codes, which render it subject to misclassification bias particularly with respect to the use of both ICD-10-PCS and CPT codes to identify single-level lumbar fusion candidates. It is also possible that exclusive definition of pseudoarthrosis using its ICD-10-CM code underestimates the true rate of pseudoarthrosis, as its diagnosis would require explicit documentation in the electronic medical record following radiographic or clinical identification. The definition of nonunion in this manner may also impair applicability of our results, as the exact diagnostic criteria were not available. Additionally, HDL-C measurements were treated as static exposures, and temporal changes in lipid profiles or HDL-C particle functionality could not be assessed. Therefore, future studies are necessary to identify the exact threshold at which elevated HDL-C confers elevated risk of unsuccessful fusion.

Conclusions

Elevated HDL-C was associated with increased rates of pseudoarthrosis following single-level lumbar interbody fusion and with higher risk of early reoperation. Although causality cannot be ascertained, these findings contribute to a growing body of evidence suggesting that extremely elevated HDL-C may exert deleterious effects on bone metabolism and fusion biology. Future prospective studies incorporating lipid subfraction analysis, inflammatory biomarkers, and radiographic fusion assessment are needed to clarify causality and guide perioperative risk stratification.

ORCID iD

Kyle Stump https://orcid.org/0009-0006-3878-5183

Ethical Considerations

This study was waived for Institutional Review Board approval due to the use of a de-identified database.

Data Availability Statement

Data is available upon request.*