The influence of osteoporosis on mechanical complications in lumbar fusion surgery: a systematic review

Highlights • Patients with degenerative spinal disease often have underlying osteoporosis.• Osteoporosis is an important risk factor for mechanical complications after lumbar fusion.• Routine preoperative screening allows interventions that can reduce complication risk.• Alternative screening recommendations may be necessary for surgical spine patients.• Standardized methods of bone health evaluation and complication reporting are needed.


Background
Osteoporosis is a highly prevalent, age-related skeletal disorder characterized by a progressive loss of bone mass and increased susceptibility to fractures [ 1 ].As the global population ages, the prevalence and significance of osteoporosis and other age-related degenerative conditions will continue to rise [ 1 , 2 ].This demographic shift is of particular significance to the spine surgeon, who will be increasingly faced with the challenge of treating patients with degenerative spinal pathologies, poor bone quality, and sequelae of osteoporosis.These conditions are frequently comorbid, with one recent systematic review estimating 79% of spine surgery patients over the age of 50 have osteoporosis or low bone mass [ 3 ].Previous studies have reported higher rates of complications, longer hospitalizations, more frequent readmissions and reoperations, and increased total healthcare costs in osteoporotic patients following spine surgery [4][5][6].Osteoporosis has also been suggested to be an independent risk factor for mechanical complications like cage subsidence (CS), pedicle screw loosening (SL), pseudarthrosis, vertebral compression fracture (VCF), and proximal junctional kyphosis (PJK) or failure (PJF) [7][8][9][10][11].
However, despite this prevalence and association with poor outcomes, osteoporosis remains profoundly underdiagnosed and undertreated in the spine surgery population [ 12 , 13 ].Although many surgeons anecdotally recognize the challenges of instrumenting osteoporotic bone and may modify their surgical plan in the setting of this diagnosis, dedicated bone health assessments are infrequently performed preoperatively [14][15][16].Underutilization of osteoporosis screening may be related to a variety of issues including logistical difficulties, concerns about the accuracy of lumbar T-scores in the degenerative spine, or lack of consensus regarding the implications of low bone density for surgical management.In the absence of clear specialty-specific guidelines addressing bone density in elective lumbar fusion, many surgeons may feel uncomfortable assuming the responsibility for osteoporosis screening and treatment [ 17 , 18 ].Inadequate insurance coverage and reimbursement practices can also discourage providers from ordering diagnostic testing or prescribing pharmacologic therapies, and may make patient adherence to treatment cost-prohibitive [ 19 , 20 ].
Consequently, while an association between osteoporosis and surgical complications may seem intuitive, it is not reflected in current practices, specialty guidelines, or healthcare policies.Addressing this gap will require engagement of patients, surgeons, and policymakers regarding the importance of bone health in spine surgery and the utility of preoperative screening for preventing complications.The purpose of this manuscript is therefore to summarize existing literature on osteoporosis in lumbar fusion, focusing on mechanical complications that can be attributed to poor bone health.Rather than concentrating on a single outcome in isolation, the authors felt a comprehensive review that encompasses the spectrum of osteoporosis-related complications is necessary to put the significance of this condition into perspective and advocate for changes in the standard of care.Moving forward, these findings can serve as a reference to inform current practices, identify areas in need of further study, and ultimately provide for more consistent, effective, and evidence-based patient care.

Literature search strategy
PubMed, Embase, and Web of Science databases were searched for original research articles reporting on the risk of specific mechanical complications of lumbar fusion surgery in relation to bone mineral density (BMD), or surrogate measurement.Details on individual search strategies are provided in Supplementary File, Table S1.Articles were independently screened for eligibility by 2 reviewers (A.F.+ A.B.) using the criteria in Table 1 .

Data extraction and outcome measures
Data was extracted by the first author (A.F.) and validated by 2 additional authors (K.H. + J.R.).Variables included information related to (1) study design and setting; (2) patient demographics; (3) treatment characteristics: surgical procedures and perioperative anti-osteoporosis therapies; (4) prognostic factor assessment: imaging modality, anatomical site(s), and cutoff thresholds or diagnostic criteria used (if applicable); and (5) primary outcomes evaluated: imaging modality, timing, and diagnostic criteria.Mechanical complications reported by at least 2 studies, with relevant statistics, were considered for analysis ( Table 2 ).Missing data were sought out through contact with corresponding authors.
For each primary outcome we presented studies' findings of prognostic effect, including estimated odds ratio (OR) for binary outcomes and mean differences or unit odds ratio (UOR) for continuous outcomes.Event rates for dichotomous data were used to generate forest plots for each outcome.Overall, data were summarized but not pooled due to substantial variability in study methodologies.

Methodological quality assessment
Evidence quality was evaluated using the system proposed by the Grading of Recommendations Assessment, Development and Evaluation (GRADE) working group [ 21 ].Using major GRADE criteria, articles were evaluated in the context of methodological domains thought to be highly important for studies of prognostic factors, including patient selection and comparability of subjects, prognostic factor assessment, appropriateness of clinical endpoints, data collection and analysis practices, and disclosure of funding [22][23][24].Assessments were performed independently by 2 reviewers (A.F.+ A.R.), with discrepancies reconciled in discussion.

Study selection
The article selection process is detailed in Figure 1 .An initial database search returned 2,112 citations, 1,500 after removal of duplicates.An additional 16 articles were identified manually.Screening by title and abstract left 182 references for full-text review.Ultimately, 71 studies satisfied our inclusion and exclusion criteria.Notably, 48 of these (67.6%) were published since the year 2020 and nearly one-third since 2022 ( Figure 2 ).

Study characteristics and quality assessment
Main characteristics of the included studies are presented in Table 3 .In total, 12,278 patients (63% female) were included, with mean ages ranging between 44.9 and 72.5 years.All patients underwent primary or revision lumbar fusion for degenerative disease or deformity.
In terms of prognostic factor assessment, 38 studies used BMD measured by dual-energy X-ray absorptiometry (DXA), the current goldstandard for diagnosing osteoporosis [ 25 ].Only 22 studies specified the anatomic site(s) of DXA scans, typically either the proximal femur or lumbar spine for all participants.Five studies used each patient's lowest T-score, noting severe degeneration, scoliosis, or instrumentation to variably preclude certain measurements.Citing concerns regarding the availability or accuracy of DXA, 42 studies investigated alternative techniques, most commonly opportunistic measurement of Hounsfield Units (HU) from preoperative CT scans ( Table 4 ).
Evidence quality was designated as high, moderate, low, or very low based on GRADE criteria (Supplementary File, Table S2).Studies were frequently downgraded for risk of bias or indirectness in prognostic factor assessment, either by employing imaging-based selection

Table 1 Details of article selection criteria
Inclusion criteria 1) Original research articles with a minimum study size of 10 patients 2) Study population must be adult patients undergoing elective instrumented lumbar fusion surgery (includes standalone interbody fusions) 3) Measured primary outcome(s) of the incidence of specific radiographic surgical complication(s) or need for revision surgery 4) Comparison between osteoporosis or low bone mass group and normal bone density group * or risk stratification based on bone density (or surrogate measure) * Studies must clearly specify the diagnostic criteria for osteoporosis (ex., T-score ≤ -2.5 or history of fragility fracture) 5) Publication date from 2002 onward Exclusion criteria 1) Reviews, case reports, biomechanical studies, cadaveric research 2) Surgical interventions including non-instrumented procedures (ex., decompression alone, vertebroplasty) or those performed for indications other than degenerative disease (ex., infection, trauma, malignancy) 3) Investigations of only osteoporotic patients (no internal control group) 4) Failure to specify the incidence of specific complications (i.e., reporting generalized results for "all complications ") 5) Studies reporting on mixed populations without stratifying results based on osteoporosis assessment 6) Studies that did not perform a baseline evaluation of bone health for all eligible participants

Primary outcomes
An overview of findings for each primary outcome is shown in Table 5 .Ultimately, a meta-analysis was not possible due to different noncomparable methods of prognostic factor assessment, variable definitions and timing of clinical endpoints, and the use of different cut point values for statistical analysis.

Cage subsidence
Cage subsidence was investigated by 24 studies in relation to BMD.Complication rates varied significantly (between 8.25% and 59% of levels), reflecting differences in surgical procedures and diagnostic criteria used ( Table 6 ).In total, 21 studies found subsidence to be associated with poor bone health as assessed by DXA (6 studies), vBMD (4 studies), lumbar HU (10 studies), and VBQ scores (2 studies).A forest plot of all reporting relevant statistics is shown in Figure 3 .Four [26][27][28][29] of 5 studies with comparative data found HU to be more predictive of complications than traditional T-scores.Amorim-Barbosa et al. [ 26 ] showed that patients with HU < 135 had a 6-fold increased risk of CS after 1-2 level PLIF or TLIF; while CS was not associated with worse functional outcomes, lower HU predicted worse disability scores and less return to work postoperatively.The remaining study, published by Pu et al. [ 30 ], found nondominant forearm T-scores to more accurately predict subsidence than mean lumbar HU (AUC 0.840 vs. 0.744), though both were independent risk factors.Notably, there were no significant differences in lumbar T-scores.Several studies investigated BMD of the fusion-level endplates as a potential predictor, reporting mixed results.
Three studies did not find a significant association between BMD and CS [31][32][33].Among these, Alan et al. [ 31 ] observed no difference in T-scores (p = .78)or fusion-level HU (p = .26)between patients with and without subsidence.However, the authors noted their study was likely underpowered given a relatively low complication rate (8 of 97 cages) and prevalence of osteoporosis (3 of 55 patients).Furthermore, all patients with low BMD were referred for preoperative endocrinol-ogy consultation with initiation of anti-osteoporosis therapy if recommended.Another potential confounder discussed was how osteoporosis status may have altered the surgical plan in favor of using supplemental pedicle screw fixation due to a presumed higher risk of subsidence.In this study, all instances of CS occurred in standalone fusions.

Screw loosening
All 14 studies reporting on screw loosening found an association between bone density and complications ( Table 7 ).A forest plot of all contributing relevant statistics is shown in Figure 4 .Three studies found higher SL rates in patients with DXA-diagnosed osteoporosis [34][35][36].SL was also associated with HU (9 studies), vBMD (1 study), and VBQ scores (1 study).All 5 studies with comparative data found alternative metrics to be more predictive than DXA T-scores [37][38][39][40][41]. Zou et al. [ 41 ] demonstrated that among patients with non-osteoporotic T-scores, SL rates were significantly higher for those with HU ≤ 110 (44.4% vs. 18.6%, p < .001),suggesting that HU may be a more sensitive metric for predicting SL.Four studies measured HU along the screw trajectory, which was found to be the best predictor of loosening at those levels [ 39 , 40 , 42 , 43 ].

Pseudarthrosis
Seven studies evaluated fusion failure in relation to bone density.Criteria used to identify pseudarthrosis varied and commonly included: (1) dynamic motion at the fusion site, (2) absence of bridging trabecular bone, and 3) evidence of implant loosening ( Table 8 ).Three studies found a significant relationship between BMD and fusion rates.A forest plot of all contributing relevant statistics is shown in Figure 5 .Choi et al. [ 44 ] uniquely used 2 different CT-based radiographic criteria to investigate time to fusion after single-level TLIF and showed that osteoporosis (HU < 90) was an independent predictor of slower fusion.

Vertebral fracture
Six studies evaluated the relationship between osteoporosis and new vertebral fractures.Complications were identified using X-ray, CT, MRI, and/or bone scans, however, no study detailed explicit diagnostic criteria ( Table 9 ).All 5 studies of proximal fractures found low BMD, defined by T-score [ 45 , 46 ] or junctional HU [47][48][49], to be an independent risk factor for new VCF.Yao et al. [ 49 ] showed not only that HU < 120 at the planned UIV strongly predicted bony PJF (OR 5.74, 95% CI 1.01-32.54,p = .04),but that there was a significant linear correlation between HU and PJK angles (r = -0.475).Due to lack of available raw data, it was not possible to create a forest plot for this outcome.

Adjacent segment disease
23 studies reported on adjacent-segment complications ( Table 10 ).A forest plot of all providing relevant data is shown in Figure 6 .Most    Data presented describe entire study population unless otherwise specified as cohort statistics.Surgery types include long (5 + levels) posterior, anterior, or combined (AP) fusion, short-segment lumbar fusion, and lumbar interbody fusion (LIF); all LIF performed with supplemental screw fixation unless indicated to be a standalone procedure.LIF types include: anterior lumbar interbody fusion (ALIF), lateral lumbar interbody fusion (LLIF), direct lateral interbody fusion (D-LIF), oblique lateral interbody fusion (OLIF), posterior lumbar interbody fusion (PLIF), transforaminal lumbar interbody fusion (TLIF), and minimally invasive TLIF (MI-TLIF).Abbreviations: not recorded (NR), dual-energy x-ray absorptiometry (DXA), computed tomography (CT) quantitative CT (QCT), Hounsfield Units (HU), bone mineral density (BMD), volumetric BMD (vBMD), magnetic resonance imaging (MRI), vertebral bone quality (VBQ), endplate bone quality (EBQ), endplate (EP), trabecular (Tb), femoral neck (FN), total hip (TH), upper instrumented vertebra (UIV), cage subsidence (CS), screw loosening (SL), vertebral compression fracture (VCF), proximal junctional kyphosis (PJK), proximal junctional failure (PJF), inter-quartile range (IQR)  According to International Society for Clinical Densitometry (ISCD) guidelines [ 146 ] routine BMD screening is indicated for all females over age 65 and males over age 70, as well as younger patients with risk factors for low bone mass.The current best-established standard for diagnosing osteoporosis or osteopenia relies on T-scores derived from areal BMD (g/cm 2 ) measured by DXA, ideally of the femoral neck or lumbar spine.These thresholds can be applied in postmenopausal women and men over age 50.Alternatively, volumetric BMD (mg/cm 3 ) can be directly measured by QCT.Of the 2 methods, central DXA is generally preferred for making therapeutic decisions and limiting radiation exposure, however, QCT may be considered superior to DXA in settings of severe degenerative disease or scoliosis [ 147 ].A number of studies have suggested that fracture risk can also be assessed with CT attenuation in Hounsfield units (HU), which can be measured from CT scans obtained for other purposes that include the lumbar spine (opportunistic bone density measurement).In the absence of established protocols, methodologies for HU measurement varied widely and included standardized (mean or segmental) and patient-specific (ex., junctional vertebrae, screw trajectory, fusion-level endplates) sites.MRI and micro-CT are other techniques used to assess bone quality; as purely research methods, both follow standardized protocols but do not have established clinical correlates or guidelines for identifying at-risk patients.Additionally, while MRI metrics can be obtained via opportunistic measurement, micro-CT is an ex-vivo study and therefore cannot be used for screening preoperatively.
Study acronyms are explained in the first footnote to Table 3  ‡ Includes one prospective observational study in which all patients over the age of 60 years (n = 598), as well as all younger patients with risk factors for osteoporosis (n = 63), underwent a preoperative lumbar DXA.In total, this amounted to 84.7% (661/784) of participants [ 81 ].
§ According to American College of Radiology (ACR) practice guidelines, vBMD (mean L1/2) thresholds for osteoporosis and low bone mass [ 147 ] ǁ Although there has been no established consensus regarding HU thresholds for diagnosing osteoporosis or osteopenia, several large-scale studies utilizing L1 HU have suggested values of 90 or 110 for osteoporosis and 135 or 120 for low bone mass [ 148 , 149 ].A recent systematic review of studies reporting on the correlation between lumbar HU and DXA T-scores identified 16 studies describing a cutoff for identifying osteoporosis (thresholds ranged 49.4-160), with a medium HU value of 114.8 (95% CI 90.9-138.7,p < .001).Notably, there was significant heterogeneity ( I 2 = 94.94%)among studies, including patient populations and location of HU measurement [ 150 ].Another meta-analysis of studies evaluating the accuracy of osteoporosis diagnosis using CT HU compared to DXA suggested a threshold of 135 to diagnose osteoporosis; the authors similarly noted that their conclusions were significantly limited by study heterogeneity [ 151 ]. (20 studies) discussed outcomes of PJK [ 50 ] or PJF after long-segment deformity correction.Index procedures varied in terms of primary versus revision, number of fused levels, location of end-instrumented vertebrae, osteotomy, interbody fusion, and fusion to the pelvis.Studies inconsistently commented on the use of cement augmentation, proximal hooks, or other modifications to improve fixation.All studies of junctional kyphosis found low BMD to be a risk factor.Yuan et al. [ 51 ] showed that osteoporotic patients had a 14-fold increased risk of PJK (p = .028);at final follow-up, those with PJK had significantly worse back pain and disability scores.Park et al. [ 52 ] found a combination of 3 factors to highly predict PJF after multilevel fusion to the sacrum: age ≥ 70, osteoporosis, and PJA ≥ 0°.PJF developed in 55.6%, 73.3%, and 100% of patients with 1, 2, and all 3 characteristics, compared to none of patients without any risk factors.Of 11 studies using HU, 10 showed that values from junctional levels best predicted complications [ 45 , 47 , 49 , 52-58 ] Kuo et al. [ 59 ] performed opportunistic screening using MRI.They ob- Age [ 126 ], BMI [ 143 ], paraspinal muscle atrophy [ 126 ], disc morphology [ 81 , 144 ], cage height [ 126 , 134 , 143 ] or shape [ 26 , 33 , 39 , 81 , 143 ], cage position disc overdistraction [ 144 ], intraoperative endplate injury [ 81 , 144 ], SA fusions [ 27 , 122 ], L5/S1 level [ 33 ], absence of endplate sclerosis [ 144 ] Screw Loosening Defined by peri-screw lucency on X-ray and/or CT Minimum follow up: range 3 to 24 months 2,454 patients (14 studies: 13 RR, 1 CC) 13%-54.6% of patients All studies found an association between bone health and SL.DXA-diagnosed osteoporosis was an independent risk factor in 2 studies.Bone density was most commonly assessed using CT or QCT (10 studies).Eight studies found lumbar HU to be an independent risk factor for SL; calculated cutoffs ranged 104-130, ‡ though results were frequently presented using a UOR.Four studies measured regional bone density from the pedicle or screw trajectory, which was highly predictive of SL.All studies with data for comparison found DXA to be less predictive than alternatives.
‡ One study calculated different cutoffs for female (153.5,AUC 0.88) and male (186.5, AUC 0.635) patients; another showed optimal HU thresholds varied based on number of levels fused (1-4) and the degree of postoperative residual deformity § One study showed that fusion took significantly longer in patients with osteoporosis (defined by HU < 90) compared to low (HU between 90 and 120) and normal (HU > 120) BMD; this study also demonstrated that fusion rates significantly varied based on the diagnostic criteria used ǁ One study found lower trabecular number and higher trabecular separation in spinous process specimens of patients with nonunion; bone quality was also shown to correlate with patient-reported postoperative outcomes of pain and disability.
¶ No study provided a clear description of radiographic criteria used for diagnosis (ex., % loss of vertebral body height) ♯ Adjacent segment degeneration after 1 to 2 level TLIF or PLIF.Two of 3 studies utilized flexion-extension X-rays, one of which reported specific diagnostic criteria.The third study specified including all symptomatic cases requiring revision.
* * One study presented T-scores of patients with and without adjacent segment degeneration (− 1.23 ± 0.23 vs. − 1.12 ± 0.19; p = .08),notably with very narrow SD and relatively better bone quality in both groups.The other study found higher rates of osteoporosis or severe osteoporosis in patients with progression of adjacent segment degeneration (30.7% vs. 17.5%;p = .069).Neither result reached statistical significance.
† † For outcomes of proximal junctional kyphosis (PJK), the most commonly utilized definition was that initially proposed by Glattes et al, [ 50 ] a proximal junctional angle (PJA), sagittal Cobb between the inferior endplate of the UIV and superior endplate of UIV + 2, that is both > 10°and at least 10°greater than the preoperative measurement.Proximal junctional failure (PJF) definitions varied more significantly and commonly included cases of PJK with additional signs of mechanical failure (fracture, spondylolisthesis, fixation failure) or any symptomatic PJK requiring revision.
‡ ‡ In 2 studies investigating both L3/4 and UIV/UIV + 1 HU measurements, junctional values were the only independent risk factor.§ § Surgical indications for revision were variably reported and included diagnoses related to hematoma, infection, pain or neurologic deficit, and construct failure.Follow-up timing was inconsistently reported.One study did not give a minimum follow-up time, instead providing a mean cohort follow-up of 18.5 ± 68.7 months and mean time to reoperation of 32.2 ± 64.1 months in osteoporotic patients and 24.2 ± 36.6 months in those without osteoporosis Three studies reported on adjacent segment degeneration after 1 to 2 level LIF [60][61][62].Ye et al. [ 62 ] observed a higher incidence of osteoporosis among symptomatic patients requiring revision (30.7% vs. 17.5%;p = .069).The other 2 studies compared T-scores between those with and without complications.While Chen et al. [ 60 ] found affected patients had slightly lower T-scores (− 1.23 ± 0.23 vs. − 1.12 ± 0.19; p = .08),Wang et al. [ 61 ] reported no differences (-1 ± 0.2 vs .-1.2 ± -0.3, p = .413).Notably, these studies were comprised of younger patients (mean ages 53.4 and 53.2, respectively) with relatively narrow BMD ranges.

Reoperation
Five studies evaluated bone health as a predictor of reoperation, the timing and indications for which varied ( Table 11 ).A forest plot of all contributing relevant data is shown in Figure 7 .Wang et al. [ 63 ] showed osteoporotic patients had a 3.6-fold increased risk of reoperation within 3 months, most commonly for surgical site infection (32.3%), hematoma (23.5%), or hardware failure (20.6%).Mugge et al. [ 64 ] also observed higher revision rates with osteoporosis (33.3% vs .16.2%; OR 2.93, 95%     Study acronyms are explained in the first footnote to Table 3 .Abbreviations: AF, anterior fixation; BPS, bilateral pedicle screw; UPS, unilateral pedicle screw; OR, odds ratio; UOR, unit odds ratio; CI, 95% confidence interval; VAS, visual analog scale; ODI, Oswestry disability index; JOA, Japanese Orthopaedic Association score; AUC, area under curve; NRS, numerical rating scale. * Given as number of patients (rather than number of cages) only.† Unless otherwise indicated, complication rates defined as number of cages with any amount of subsidence, with severe CS referring to Grade II or Grade III CS as defined by Marchi et al.

Discussion
There has been a dramatic rise in the number of elective lumbar fusions performed over the past few decades, with the most significant increases occurring in patients over 65 [ 67 ].As these procedures are associated with relatively high complication rates, particularly in elderly patients [ 68 , 69 ], surgeons must be aware of modifiable risk factors to allow for identification of those who may benefit from medical or surgical optimization.Biomechanically, osteoporotic bone offers poor support for instrumentation, which may predispose to failures at the implant-bone interface.In the present study, we reviewed the literature on osteoporosis as a risk factor for different mechanical complications of lumbar fusion.
An increasing number of studies have reported osteoporosis as a risk factor for cage subsidence, a finding supported by our review and others [ 70 , 71 ].In osteoporotic patients, compromised vertebral strength, decreased endplate failure loads, and increased stress concentration within the surgical segment all contribute to failure at the cage-endplate interface.Surgical variables like implant design (size, shape, material properties), cage positioning on the endplate, and use of supplemental fixation are other important predictors of endplate stress and subsidence [ 72 , 73 ].Selecting implants with greater endplate contact, positioning over stronger regions of the endplate, and using supplemental fixation can all help prevent subsidence [ 74 ].Study acronyms are explained in the first footnote to Table 3 .Abbreviations: OR, odds ratio; HR, hazard ratio; PJA, proximal junctional angle.
Screw loosening is another common complication of lumbar fusion often associated with poor bone stock [ 75 ].Concerns for adequate fixation in osteoporosis has prompted investigation of a number of technique modifications including cement augmentation of high-risk levels to enhance screw purchase and prevent complications [ 76 , 77 ].Use of alternative screw trajectories that take advantage of stronger regions of the vertebra is also an option.BMD has a well-established association with regional vertebral strength and pedicle screw stability in vitro [ 78 ].More specifically, BMD measurements made along a screw's trajectory can provide a particularly accurate prediction of mechanical performance and are commonly used in biomechanical investigations of fixation strength and stability [ 79 , 80 ].Similar metrics are increasingly being investigated for predicting complications clinically.
Pseudarthrosis is a common consequence of implant malfunction and may be more likely in patients with osteoporosis.Among studies in this review, pseudarthrosis was most frequently reported as a secondary outcome in relation to cage subsidence or screw loosening.Park et al. [ 81 ] reported coexistence of all 3 complications: pseudarthrosis occurred in 2.9% of cages without migration compared to 45%, 58.3%, and 82.4% with migration, subsidence, and retropulsion, respectively; concomitant SL rates were 4.7%, 10%, 61.1%, and 70.6%, respectively.Unfortunately, we found relatively limited and inconclusive data regarding a direct association between bone density and fusion.One study observed that fusion took significantly longer in osteoporotic patients [ 44 ], which is consistent with findings reported by meta-analyses of randomized controlled trial data showing osteoporosis treatment improves fusion rates after lumbar instrumentation [ 82 , 83 ].Liu et al. [ 84 ] published the only study in our review using micro-CT.Ex-vivo analysis of spinous process specimens obtained during index surgery revealed higher trabecular number and lower trabecular separation with greater bone surface/total volume (BS/TV) among patients ultimately achieving solid fusion.Low BS/TV was the only independent predictor of pseudarthrosis and was strongly associated with worse low back pain and disability outcomes.Although micro-CT cannot be used for preoperative risk stratification, these results can help strengthen the evidence associating bony structural deficiencies with complications.* PJK defined as proximal junctional angle (PJA), measured as the sagittal Cobb between the inferior endplate of the UIV and superior endplate of UIV + 2, that is both > 10°and at least 10°greater than the preoperative measurement [ 50 ].Study acronyms are explained in the first footnote to Table 3 .Additional abbreviations: odds ratio (OR), unit odds ratio (UOR), visual analog scale (VAS), cerebrospinal fluid (CSF) * Indications for revision noted to be not completely recorded, but included diagnoses related to CS, pseudarthrosis, and adjacent segment disease Thoracolumbar fragility fractures are a hallmark complication of osteoporosis, usually occurring after a fall or low-energy trauma [ 85 ].Patients who undergo long-segment fusions may be particularly susceptible to new junctional fractures under the increased stress of instrumentation.These classically occur as either (1) simple, usually chronic, compression of the first uninstrumented vertebra (UIV + 1), or (2) acute UIV collapse, followed by ligamentous failure and adjacent vertebral subluxation [ 45 , 86 ].The latter case is thought to be directly precipitated by significant mechanical stress from substantial alignment correction and tends to result in a more severe kyphotic deformity associated with higher rates of neurologic deficits, PJF, and revision surgery.Our review confirms osteoporosis to be a significant risk factor for PJK and PJF, particularly secondary to fracture [ 45 , 49 ].In addition to considering alternative alignment targets for correction in osteoporotic patients, choosing an appropriate UIV can help minimize junctional stresses.For patients still thought to be high-risk for bony failures, prophylactic augmentation of the UIV and UIV + 1 may be performed during the index procedure.There is also an abundance of evidence that perioperative anti-osteoporosis therapy can help prevent PJK.[ 87 ] Yagi et al. [ 88 ] showed that 6 months of teriparatide following thoracolumbar fusion significantly increased BMD at the UIV + 1 and was associated with lower rates of bony PJK at 2 years compared to untreated controls (4.6% vs. 15.2%;p = .02).
In terms of functional outcomes, mechanical complications in this review ranged from asymptomatic radiographic findings to disabling events requiring additional surgery.In addition to construct revision for mechanical failures, several studies also found osteoporotic patients were more likely to undergo unplanned reoperations for other indications as well [ 64 , 63 ].Quality of life metrics were inconsistently evaluated and represent an area of future research needed.

Implications for practice
Collectively, our findings support the utility of osteoporosis screening prior to elective lumbar fusion.The International Society for Clinical Densitometry (ISCD) recommends preoperative bone health assessment for all females aged ≥ 65 years and males ≥ 70 years, as well as those with prior fragility fracture or other risk factors for osteoporosis [ 89 ].However, relatively high rates of poor bone quality have been reported in younger surgical spine patients [90][91][92].Williamson et al. [ 93 ] further demonstrated the clinical significance of this, showing that in patients under 65 with minimal deformity, osteoporosis was the most significant risk factor for major mechanical (33% vs. 7% without osteoporosis, p = .025;OR 5.9, p = .048)and major radiological (29% vs. 6%, p = .001;OR 7, p = .003)complications, trends not observed in their overall cohort.Moreover, our review suggests that patients with poor bony strength despite a non-osteoporotic BMD may also be at increased risk for mechanical complications.Thus, it may be necessary to consider alternative screening recommendations in adults presenting for elective spinal fusion [ 91 , 94 , 95 ].
Additionally, there are no current guidelines for how BMD values should be used to inform treatment in elective lumbar fusions.In particular, there has been growing interest in using site-specific HU obtained in opportunistic screening to guide risk assessment, for example using values from the UIV and adjacent levels to predict PJK.[ 96 ] However, although individual surgeons may currently employ these techniques to inform their own decision-making processes, methodologies have not been standardized and optimal thresholds for predicting complications remain unknown.Ultimately, clinical care could benefit from standardized evidence-based recommendations that reference specific BMD cutoff values and related implications for treatment.

Limitations and future directions
The findings of this review should be interpreted in the context of its limitations.Current literature on osteoporosis and surgical outcomes is largely retrospective, which introduces a number of concerns for bias [ 97 ] Unfortunately, due to insufficient practices of osteoporosis screening, there is a relative lack of available DXA data for analysis [ 98 , 99 ].As a result, many retrospective cohort studies will use International Classification of Diseases (ICD) codes to obtain information about osteoporosis status from medical records or other healthcare databases.However, it has been well-established that even among patients with documented fragility fractures, osteoporosis is profoundly underdiagnosed in both electronic medical records [100][101][102] and administrative databases [ 103 , 104 ].Evaluations of osteoporosis reporting patterns in claims data have revealed the magnitude of these deficiencies, leading to recommendations against using this data in place of BMD-based reference standards [ 103 ].In light of these limitations, the authors therefore felt it necessary to employ a more restrictive selection criteria with respect to study methodologies, focusing on comparative evaluations that explicitly investigated BMD as a predictor of mechanical complications.This excluded studies using age as a proxy for poor bone health, including those performed in elderly populations where many were likely osteoporotic but lacked attention to this diagnosis [ 105 , 106 ].We also eliminated all studies in which osteoporosis status was assigned solely based on ICD code, as the absence of an osteoporosis diagnosis alone cannot be considered a reliable indicator of good bone health.Nevertheless, even among included studies, many still had potential for selection bias due to the use of imaging-based patient selection criteria, which did lower the strength of this evidence.Consideration of these limitations highlights the need for greater attention to the screening, diagnosis, and documentation of conditions like osteoporosis.
Another limitation of included studies is the potential that patients' presumed osteoporosis status influenced the treatment they received, which could have dramatically reduced the apparent impact of osteoporosis on patient outcomes.Unfortunately, confounding variables related to the initiation, type, and duration of pharmacologic therapy as well as the use of surgical technique modifications were infrequently addressed.
Finally, studies varied considerably in patient selection criteria, surgical indications and procedures, imaging modality and anatomical site(s) of BMD assessment, and the diagnostic study, criteria, and timing used to define clinical endpoints.Together, these factors likely resulted in significant differences in both the number of complications identified and proportion attributed to osteoporosis.Use of variable and nonstandardized thresholds for statistical analyses further limited any ability for direct comparison and precluded meta-analyses.Recognition of this heterogeneity is critical as it reflects the lack of consensus among surgeons and researchers regarding how to best screen for spinal osteoporosis and predict related surgical complications, concepts that are not necessarily synonymous.
The current gold-standard for assessing bony strength, skeletal fragility, and fracture risk relies on BMD, measured at the spine or hip, using DXA [ 25 ].The anatomical site of BMD measurement is also important, as T-scores obtained from different locations are not necessarily interchangeable [ 107 ].In a population defined by the coexistence of spinal osteoporosis and surgical degenerative pathology, measurement of BMD in the lumbar spine would theoretically be ideal for predicting focal osteoporosis-related mechanical failures [ 91 ].However, these degenerative changes also can falsely elevate lumbar T-scores, making them paradoxically less accurate for predicting regional bony strength [ 108 , 109 ].
Given these shortcomings, there has been recent interest in opportunistic screening using CT or MRI, which may improve access to clinically relevant information on bone health for surgical spine patients [ 91 , 110 ] CT-based methods have been shown to generate reliable measures of volumetric BMD that correlate well with vertebral biomechanical properties, fracture risk, and outcomes of lumbar fusion [111][112][113].A notable advantage of CT lies in the ability to obtain measurements from customizable regions of interest, usually located in trabecular bone and excluding osteophytes or degenerated facet joints, making them less susceptible to error from degenerative changes [ 114 ].These measurements can also be isolated to surgically-relevant areas like pedicles, endplates, and vertebral bodies of planned instrumented levels, which may allow for personalized risk stratification and surgical decision-making.Less extensively investigated, MRI-determined VBQ scores have also shown promise for identifying osteoporosis and predicting postoperative complications [ 37 , 66 , 115 ].As overall bony strength is determined by both bone density and quality, these investigations may provide important supplemental information to inform risk assessment [ 116 ].
Large-scale prospective evaluations will be necessary to evaluate the utility of these different methodologies for bone health assessment and determine which imaging study and threshold value(s) best predict mechanical complications and patient outcomes.In the meantime, it is important to consider the mounting evidence that osteoporosis is a significant risk factor for complications after lumbar fusion, and the crucial role that preoperative bone health assessment can play in mitigating these risks.Expanding access to tools like DXA, which has been widely validated and remains the current gold-standard for diagnosing osteoporosis and initiating pharmacologic therapy, will be important for identifying and treating at-risk patients [ 25 , 107 , 117 ].

Conclusion
This systematic review provides a comprehensive summary of osteoporosis and mechanical complications of lumbar fusion.Our results demonstrate that poor bone health is an important risk factor for implant-related failure, pseudarthrosis, VCF, junctional deformity, and reoperation after elective lumbar fusion.These findings strongly support the role of preoperative screening to identify high-risk patients and allow implementation of low-risk management strategies.Our review also highlights current challenges in the evaluation and management of osteoporotic patients undergoing lumbar fusion, including a paucity of relevant and complete clinical data, variability in methods of bone health assessment and reporting of complications, and the use of heterogeneous definitions that limit the interpretation, generalizability, and meta-analysis of available evidence.As we move towards addressing these gaps, it is important to consider the mounting evidence that osteoporotic patients with degenerative spinal disease may represent a unique population in which bone health is of utmost importance, but current practices of identifying high-risk patients are inadequate.The authors therefore suggest a collaborative, multidisciplinary Academic Consortium specifically dedicated to addressing the unique challenges of treating spinal disease in patients with osteoporosis.The goals of such a consortium would begin with development of consensus criteria for best practices of bone health assessment and uniform definitions for clinical endpoint evaluation.Establishment of standardized metrics will facilitate the consistency of data collection and minimize ambiguity in a way that enables direct comparison and meta-analysis, which will be essential for adequately investigating these relationships.Ongoing integration of evolving evidence will be necessary to identify unmet needs, advance targeted research, and guide clinical decision-making towards evidence-based practice, ultimately leading to better patient outcomes.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
. Abbreviations: CSF, cerebrospinal fluid; BS/TV, bone surface / total volume; BS/BV, bone surface / bone volume; Tb.Th, trabecular thickness; Tb.N, trabecular number; Tb.Sp, trabecular separation.*Hip and spine measurements preferred; distal radius recommended only when hip and spine cannot be obtained.† According to the World Health Organization (WHO), gold-standard T-score thresholds used for the diagnosis of osteoporosis and osteopenia[ 25 ]

Table 2
Primary outcomes included Interbody cage subsidence Screw loosening PseudarthrosisNew vertebral fracture Junctional pathologies (adjacent segment disease, proximal junctional kyphosis or failure) Revision surgery criteria or lacking a gold-standard comparison.Insufficient accounting for confounders was another common reason for downgrading.

Table 3
Characteristics of included studies

Table 4
Overview of methods used for bone health assessment, and their proportions

Table 5
Summary of findings

Table 5 (
continued ) Study populations consisted of patients undergoing primary or revision instrumented lumbar fusion (specific indications and procedures varied) for degenerative disease.All studies assessed osteoporosis, or surrogate measure of bone health, as a risk factor for specific surgical complications including cage subsidence, screw loosening, pseudarthrosis, adjacent-level fractures, junctional disease, and revision surgery.Study acronyms are explained in the first footnote to Table3.Additional abbreviations: RR, retrospective review; PC, prospective cohort; CC, case-control; BMI, body mass index; PA, proximal junctional angle; TLK, thoracolumbar kyphosis; SS, sacral slope; LL, lumbar lordosis; SVA, sagittal vertical axis; OR, odds ratio; UOR, unit odds ratio; PEEK, Polyetheretherketone. Explanations:* Two studies provided complication rates in terms of number of patients rather than number of cages † One study did not find DXA T-scores or lumbar HU to predict CS and discussed possible confounding factors including: (1) relatively low incidence of complications, (

Table 6
Details and results of studies reporting on cage subsidence.
( continued on next page )

Table 6 (
continued ) ( continued on next page )

Table 7
Details and results of studies reporting on screw loosening.

Table 7 (
continued ) Study acronyms are explained in the first footnote to Table 3 .Abbreviations: OR, odds ratio; UOR, unit odds ratio; VAS, visual analog scale; ODI, Oswestry disability index; (SRS)-22 score, scoliosis research society; AUC, area under curve.

Table 8
Details and results of studies reporting on pseudarthrosis.

Table 9
Details and results of studies reporting on new vertebral fractures.

Table 10
Details and results of studies reporting on junctional complications.

Table 11
Details and results of studies reporting on reoperation