Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Dec 2.
Published in final edited form as: J Biomech. 2021 Sep 21;129:110756. doi: 10.1016/j.jbiomech.2021.110756

Does the neutral zone quantification method matter? Efficacy of evaluating neutral zone during destabilization and restabilization in human spine implant testing

T Di Pauli von Treuheim 1, L Zengerle 2, AC Hecht 1, HJ Wilke 2, JC Iatridis 1
PMCID: PMC8671349  NIHMSID: NIHMS1745839  PMID: 34619461

Abstract

Neutral zone (NZ) is an important biomechanical parameter when evaluating spinal instability following destabilizing and restabilizing events, with particular relevance for implant efficacy testing. It remains unclear what NZ calculation methods are most sensitive at capturing NZ changes across treatment conditions and a direct comparison is needed. The purpose of this study was to determine the most sensitive method at quantifying instability in human spines. Six cadaveric lumbar motion segments were subjected to a repeated measures implant testing schema of four sequential conditions: (1) Intact, (2) injury by herniation, (3) device implantation, (4) long-term cyclic fatigue loading. NZ was expected to increase after destabilization (steps 2 & 4) and decrease after restabilization (step 3). NZ methods compared in this study were: trilinear (TL), double sigmoid (DS), zero load (ZL), stiffness threshold (ST), and extrapolated elastic zone (EEZ). TL, ZL, and EEZ identified statistically significant NZ differences after each condition in flexion/extension and lateral bending. The ZL method also captured differences in axial rotation. All methods identified expected NZ trends after destabilization and restabilization, except DS in axial rotation. The TL, ZL, and EEZ methods were the most sensitive methods with this human cadaveric dataset. Future investigations comparing methods with additional datasets will clarify outcome generalizability and determine what curve profiles are most suitable for DS and ST methods. Understanding the applicability of NZ methods can enhance rigor and reliability of spinal instability measurements when quantifying the efficacy of novel implants and permits insight into clinically relevant biomechanical changes.

Keywords: Neutral zone, spine biomechanics, intervertebral disc, stability, spine implant

1. Introduction

Low back pain is a leading cause of global disability and is highly associated with motion segment instability (Panjabi, 1992; Stokes et al., 2011). Spinal instability results from laxity of its structural components (Panjabi, 2003) commonly occurring in the setting of spinal pathology and disease (Aebi, 2005; Voinier et al., 2020). Treatments aim to restore physiological stability through neuromuscular strengthening and device implantation (Mitchell and Hurrell, 2019). Diagnosing instability in vivo has been attempted with dynamic fluoroscopy to evaluate laxity-associated low back pain (Breen et al., 2015; Zhao et al., 2016) and with inertial motion sensors to investigate whole trunk stability in scoliosis patients (Voinier et al., 2020).

A critical purpose for quantifying spinal instability is to evaluate the efficacy of novel implants. This is done in accordance with standardized criteria, by subjecting specimen to an implant testing schema and tracking the implant’s ability to restore stability across destabilizing and restabilizing events (Wilke et al., 1998b). Destabilizing events refer to injury of the nucleus pulposus, annulus fibrosus, facet joints, ligaments, and/or surrounding boney structures. Contrarily, restabilizing events depend on spinal instrumentation, often through vertebral body replacements, fusion cages, dynamic implants, screw-rod-systems, gels etc.

Implants are designed and marketed as a treatment for clinical spinal instability and their efficacy is assessed during in vitro biomechanical testing, where parameters such as neutral zone length (NZ) and range of motion are extracted from the load-deflection curve. NZ represents the region over which the spinal motion segment moves essentially free of applied loading requiring minimal postural muscular effort (Panjabi, 1992; Voinier et al., 2020; Wilke et al., 1998b). It is often preferred over range of motion at quantifying instability, citing its increased sensitivity and independence of arbitrarily chosen maximum loading bounds or load-deflection curve asymmetry (Costi, 2021; Panjabi, 1992, 2003; Voinier et al., 2020). Despite these advantages, there exists no clear consensus on how to reliably and objectively measure the NZ (Panjabi, 1992; Wilke et al., 1998b). Many NZ quantification methods have been proposed; however, recent investigations have shown that differences exist between methods and that differences depended on the profile of the load-deflection curve (Di Pauli von Treuheim et al., 2020). If NZ measurements can vary by method of calculation, then it follows that an implant’s “apparent” efficacy at restoring motion segment stability may also depend on the NZ method and therefore adds variability to conclusions made on biomechanical efficacy of the implant.

While some NZ comparisons have been reported with proposals of a new method, no comprehensive work has directly compared the influence of the most popular NZ methods. The aim of this study is to determine the sensitivity of NZ methods for quantifying NZ changes in human motion segments during destabilizing and restabilizing conditions. We hypothesize that NZ methods will differ in their ability to track changes in NZ.

2. Materials and Methods

2.1. Specimen preparation

Specimen preparation and testing was performed in accordance with ethics committee approval (Zengerle et al., 2020). Six lumbar motion segments (L2–3, L3–4, L3–4, L4–5, L5–S1, L5–S1) with median age 64.5 years (range 57–72) and Pfirrmann grade 2.5 (2–3) were obtained from four fresh-frozen human cadaveric spines. Specimens were excluded if donor history included fractures, tumors, osteoporosis (overall BMD <80.0 mgCaHA/cm3), low disc height (<6.5 mm). All soft tissues were removed preserving the intervertebral disc, facet joints, and spinal ligaments. Specimen were stored in triple sealed polyethylene bags at −20 °C and thawed overnight at +4 °C prior to testing.

2.2. Test protocol

NZ was measured from load-deflection curves, which were obtained through a quasistatic flexibility test (Wilke et al., 1998b) which applies ±7.5 Nm at a loading rate of 1°/sec in the desired axis without preload, while uncoupling the remaining axes and simultaneously tracking displacement with an optical motion tracking system (VICON Motion Systems Ltd., Oxford, UK). Tests were performed in flexion/extension (FE), lateral bending (LB), and axial rotation (AR) for three loading cycles, where the first two cycles were used for preconditioning and the last for analysis. The implant testing schema used in this study included four conditions and has been described previously (Wilke et al., 2006; Wilke et al., 2013; Zengerle et al., 2020): (1) intact (labelled Int.), (2) the destabilizing injury (disc herniation with subsequent discectomy, labelled Destab.), (3) the restabilizing repair (nucleus implant and annulus sealant, labelled Restab.), and (4) long-term fatigue test (labelled Re-destab.) (Fig. 1). A combined nucleus replacement with annulus closure was chosen, as it was previously shown to restore biomechanical properties to the intact motion segment, thus satisfying the assumption that a restabilizing event had occurred (Zengerle et al., 2020).

Figure 1.

Figure 1.

Open in a new tab

The implant testing schema illustrates the timepoints at which NZ measurements were taken in relation to the treatment condition. The destabilizing injury was simulated by a disc herniation with subsequent discectomy and the re-destabilizing condition was simulated by a long-term fatigue test. Both the destabilizing herniation and the fatigue tests were performed using a “Hula-hoop test”, which was previously protocolled by Wilke et al. in 2013. The conditions differ in that the herniation required 5,000 cycles, whereas the long-term fatigue test required 100,000 cycles (~5.5 hours). In short, the “Hula-hoop” test consists of a sinusoidal (5 Hz), eccentric 100–600 N load, which is applied at a lever arm of 30 mm to the longitudinal axis of the specimen while it rotates at 360°/ min. The restabilizing condition was simulated by device implantation. Repair of the defect was achieved using a collagen-based nucleus implant (RESTORE, 3D Spine Matrix Biotechnologie GmbH, Krems, Austria) with annulus closure device (Barricaid, size 10, Intrinsic Therapeutics Inc., Woburn, MA, USA). The expected change in absolute NZ values and percent change are shown with arrows indicating the directionality of the expected trend.

2.3. NZ quantification methods

NZ was measured with the five most commonly used algorithms from the spine biomechanics literature using a custom-made MATLAB Graphical User Interface program (Release 2020b Mathworks Natick, MA, USA). Algorithms of all NZ methods were described in detail previously (Di Pauli von Treuheim et al., 2020) (Fig. 2). In brief, the trilinear method (TL) fits a 3rd degree polynomial to the bidirectional loading data and delineates NZ as the intersection of the elastic zone line and the minimum slope tangent of the polynomial (Boxberger et al., 2006; Di Pauli von Treuheim et al., 2020; Sarver and Elliott, 2005). This method generates 3 lines (1 tangent line to the polynomial’s minimum slope and 2 elastic zone lines) and we therefore retained the “trilinear” naming even though it uses a polynomial fit. The double sigmoid method (DS) fits a ten-parameter double sigmoid function to the data and subsequently derives two inflection points, which delineate NZ boundaries (Smit et al., 2011). The zero load method (ZL) attributes a NZ value to the hysteresis of the disc, measured as the deflection between the loading and unloading limbs at 0 load (Wilke et al., 1998b). To mitigate against data heteroscedasticity around the axis, a line was fit to the data of each limb immediately surrounding 0 load and its intersection point with the load axis (y axis) was used in the NZ value calculation. The stiffness threshold method (ST) scans the load-deflection curve for a minimal slope (stiffness) segment and then extends the segment from both poles until its slope increases by 5%, at which point NZ boundaries are set (Di Pauli von Treuheim et al., 2020; Thompson et al., 2003; Torre et al., 2019). The extrapolated elastic zone method (EEZ) extends the elastic zone line to its intersection with the 0 load axes, where it delineates NZ as the deflection between these intersections (Gay et al., 2006). NZ values measured by each method were compared as absolute values and as percent change from the preceding condition. The preceding condition was used, as opposed to the intact condition, in order to capture the relative destabilization/restabilization event, thus yielding positive/negative changes in NZ and providing clear targets for comparison.

Figure 2.

Figure 2.

Open in a new tab

NZ quantification methods are illustrated with their underlying algorithm and description. The NZ boundaries are shown by black/white circles and NZ length by brackets. Axis units are color coded (load=red, deflection=black) but not centered on data and are meant to easily convey inversion of axes and interpretation of methods (TL, ST, and EEZ use the same orientation). TL: Red—3rd degree polynomial fitted to loading portions from both limbs. Solid blue—tangent at minimum slope of polynomial fit. Dashed blue—elastic zone line extended from the linear fit between the blue markers (80–100% of max load). NZ boundary at intersection of dashed and solid blue lines. DS: Red—double sigmoid fit to a limb at a time. Light blue—1st derivative of red line. Dark blue—2nd derivative of red line. NZ boundary at minima and maxima of dark blue line. ZL: Red—line at zero load. NZ boundary at intersection of both limbs with red line. ST: Light blue—baseline segment of lowest slope. Dark blue—baseline segment data range was extended until its slope increased by 5%. NZ boundary at end of each dark blue line. EEZ: Blue—elastic zone fit line extended from the linear fit between the blue markers (80–100% of max load). Red—line at zero load. NZ boundary at intersection of blue and red lines.

2.4. Statistical analysis

Statistical analyses were performed in SPSS (IBM SPSS Statistics version 24; IBM Corp., Armork, NY). Shapiro-Wilk test showed non-normal distribution for some groups of absolute NZ values; accordingly, nonparametric statistics using a paired Wilcoxon signed-rank test with Bonferroni correction was used to assess differences in absolute NZ values only between adjacent testing conditions (i.e. int. vs. destab., destab. vs. restab., restab. vs. re-destab.) in order to limit type 1 error. Differences between methods in tracking NZ percent change across conditions was computed through an interaction effect. Normality was satisfied when pooling percent change across the entire outcome variable. A mixed model ANOVA was used to measure this interaction effect with post hoc Tukey HSD multiple comparisons. Agreement of absolute NZ values between methods was assessed by a two-way, single measures intraclass correlation analysis for agreement (a measure of fixed and proportional differences, ICCA-1) and consistency (a measure of proportional differences, ICCC-1) (Mehta et al., 2018). A priori power analysis showed that six specimens are sufficient to test implant intervention effects (Wilke et al., 1998b) and a subsequent sensitivity power analysis showed that this study design was sensitive to detect NZ percent change differences between methods with an effect size of Cohen’s f=0.5 (80% power, alpha=0.05) (Cohen, 2013).

3. Results

3.1. Method fitting algorithms were successfully applied to all load-deflection curves

Disc herniations were visually confirmed for all specimens, affirming that a destabilizing event was generated. NZ values were calculated for all load-deflection curves in flexion/extension and lateral bending; however, in axial rotation, the TL, ST, and EEZ methods failed to calculate a NZ value for one load-deflection curve, which did not have clear breakpoints (Fig. 3). The ST method also failed to calculate a NZ value for two specimens due to local noisy data within the NZ region, which was subsequently bypassed by manually increasing the baseline segment. Load-deflection curves were successfully fitted by the DS method’s double sigmoid function (R2≥0.99±0.01), the TL method’s polynomial function (R2≥0.99±0.01), and the ST method’s linear regression (R2≥0.94±0.02). The ZL and EEZ methods do not require curve fitting algorithms.

Figure 3.

Figure 3.

Open in a new tab

a) Load-deflection characteristics that influence NZ method algorithms. b) A visual rendition of intact condition load-deflection curve profiles for all specimen separated by their axis of rotation, showing a predominance of curves with smooth transitions between the elastic and neutral zone. Horizontal scale bars represent a load of 5 Nm for all axes. Vertical scale bars represent 5 degrees for flexion/extension and lateral bending axes, whereas it represents 1 degree for the axial rotation axis. The TL, ST, and EEZ methods failed for specimen L3–L4 in axial rotation, likely due to its lack of breakpoints. All methods measured larger values for specimen L4–L5, which were marked as outliers in figure 4.

3.2. TL, ZL, and EEZ methods were most sensitive at quantifying differences between adjacent treatment conditions

Absolute NZ values differed between methods; the ZL method consistently measured the smallest NZ values while the TL and EEZ methods measured the largest (Fig. 4). For all methods, NZ values were largest in flexion/extension and lateral bending and lowest in axial rotation. In flexion/extension, all methods captured one large NZ value, which was flagged as an outlier across the entirety of the implant testing schema. This specimen also consistently produced the largest NZ value in lateral bending and axial rotation. The TL, ZL, and EEZ methods measured significant differences after each treatment condition in flexion/extension and lateral bending. The ZL method was alone in measuring a significant difference between each condition in axial rotation. The ST and DS methods only captured a significantly difference in NZ after one event.

Figure 4.

Figure 4.

Open in a new tab

A comparison of five NZ methods in flexion/extension, lateral bending, and axial rotation in the intact condition (Int), after a destabilizing event (Destab, D), after a restabilizing event (Restab, R), and after long-term fatigue loading (Re-destab, F). Methods were assessed for their ability to track NZ values according to the expected trend with paired Wilcoxon signed ranks test showing significant differences (p<0.05 with Bonferroni correction) between adjacent conditions (i.e. I vs. D, D vs. R, etc.). TL, ZL, and EEZ methods were most sensitive at capturing differences.

3.3. ZL and EEZ best captured the expected NZ percent change trend

NZ percent change calculated from the preceding condition was expected to be positive after the destabilizing event (herniation/discectomy injury), negative after the restabilizing event (i.e. device implantation), and positive after the re-destabilizing event (long-term fatigue loading) (Fig. 1). There was no difference in how methods tracked NZ percent change (i.e. no condition by method interaction effect) in flexion/extension, yielding no most sensitive method for this axis. In lateral bending, the ZL method tracked changes in NZ better than the TL, DS, and ST methods. In axial rotation, the DS method did not track the expected trend at all and was significantly inferior compared to the TL, ZL, ST, and EEZ methods (Fig. 5).

Figure 5.

Figure 5.

Open in a new tab

NZ percent change is shown with graphs separating axes of rotation. Percent change was calculated as the difference from the previous condition, yielding a positive change after destabilizing events (Destab and Re-destab) and a negative change after restabilizing events (Restab). A mixed-effects ANOVA was used to measure interactions between methods and conditions, which assesses whether methods differ in how changes in NZ are tracked across the testing schema. These results were tabulated. Additionally, a post hoc Tukey HSD test was executed to compare methods within a treatment condition, where significant differences (p<0.05) were illustrated in black bars. All methods tracked percent change according to the expected trend, except for DS in axial rotation, which was found to be significantly different to TL, ZL, ST, and EEZ when looking at the interaction effect. The ZL method most frequently generated a significantly greater percent change, according to pairwise comparisons.

3.4. TL and EEZ showed high agreement, but consistency was high between most methods

The highest level of agreement (i.e. lowest fixed differences between methods) was between TL and EEZ across all axes of rotation (Fig. 6a). The ZL method did not agree with any other method, measuring the smallest absolute NZ. Nevertheless, the ZL method showed low proportional differences (i.e. high consistency) compared to other methods. Proportional differences were minimal between most methods in flexion/extension and lateral bending (Fig. 6b).

Figure 6.

Figure 6.

Open in a new tab

a) Agreement (measure of fixed and proportional differences, ICCA-1) and b) consistency (measure of proportional differences, ICCC-1) of absolute NZ values were calculated using a two-way, single measures intraclass correlation analysis. Statistically significant (p<0.05) ICC values were marked by an asterisk, indicating that the correlation was not due to chance.

4. Discussion

The TL, ZL, and EEZ methods were most sensitive at measuring an increase in NZ after destabilizing events and a decrease in NZ after restabilizing events, while also capturing statistically significant differences between adjacent conditions (Fig. 4, Fig. 5). The DS and ST methods captured smaller changes after destabilizing and restabilizing events and rarely generated statistically significant differences after conditions. A commonality between the TL, ZL, and EEZ methods that may explain these differences is that these methods do not focus on calculating the load-deflection curve’s “most linear” neutral zone segment length. Instead, the TL and EEZ methods rely on the elastic zone, whereas the ZL method characterizes the hysteresis between loading and unloading curves when calculating NZ. Contrarily, the DS method delineates the linear neutral zone segment by 2nd derivative inflection points and ST method captures it via a threshold slope change. While the choice in method influenced NZ change detection sensitivity for load-deflection curves of this study, a more generalizing conclusion on each method’s sensitivity will only be possible after similar studies with varying load-deflection curve profiles are performed.

4.1. Performance of individual methods

TL and EEZ methods successfully measured NZ values that were sensitive to perturbations in motion segment stability for load-deflection curves lacking clear breakpoints (Fig. 3). This finding was surprising, since the TL method was specifically designed for tri-linear profiles—containing clear elastic and NZ regions separated by breakpoints—which are common for axial tension/compression biomechanical testing (Sarver and Elliott, 2005; Torre et al., 2019). The EEZ method is similar to TL in that it also relies on elastic zone to calculate NZ (Gay et al., 2006). This method is rarely used by the spine biomechanics community and its applicability and success at detecting changes in NZ for load-deflection curves with nonlinear elastic zones was therefore unexpected. Since load-deflection curves of this study did not have clear breakpoints, our findings suggest that the TL and EEZ methods should be considered efficacious options even when load-deflection curves lack breakpoints.

The DS method measured NZ values that were insensitive to changes in motion segment stability using this study’s load-deflection curves (Fig. 4, Fig. 5). This finding was surprising, since prior work showed promise for its algorithm due to its and high curve-fitting reliability (Di Pauli von Treuheim et al., 2020). Our MATLAB program fitted a template double sigmoid function to the data and systematically modified fitting parameters by minimizing the sum of squares error (R2≥0.99±0.01), as described by its authors (Smit et al., 2011). Usage of this method has gained traction, increasingly being deployed to measure NZ for in vitro and in vivo investigations (Fig. 7) (Voinier et al., 2020). Despite its excellent fit to our dataset, it was unable to reliably capture statistically significant changes in NZ. Given that this method did not track the expected NZ trend with predominantly linear load-deflection curves (i.e. axial rotation), we propose that this method should be avoided under these conditions (Fig. 5).

Figure 7.

Figure 7.

Open in a new tab

A review of the PubMed literature from 2011 up to 2020 illustrates a) the dominant usage of the ZL method, b) particularly with cadaveric specimens. Contrarily, TL and EEZ methods are rarely utilized. Lack of documentation continues to be a practice and the authors encourage documentation of the method.

ZL method measured NZ values that were sensitive to motion segment perturbations under most conditions. This method captured significant changes in NZ after each treatment condition in all axes of rotation and was the only method to do so in axial rotation. This method was previously shown to be sensitive to perturbations for axial rotation curves (Volkheimer et al., 2018). We also found utility in flexion/extension and lateral bending by measuring the entire NZ, as opposed to separating it into its positive and negative components (i.e. displacement toward the unloading and loading limb, respectively) (Wilke et al., 1998b; Zengerle et al., 2020). Separation of components may hold relevance for flexion/extension movements due to spinal column asymmetry; however, it may also dilute the signal and was therefore not done in this study. This method is most limited by load-deflection curves with small curve hysteresis (i.e. when loading and unloading curves have minimal offset) (Fig. 3).

The ST method was unable to reliably capture a change in NZ after destabilizing and restabilizing events. This method was designed as a computational surrogate for visual breakpoint inspection (Torre et al., 2019). Load-deflection curves of the present study lacked visually apparent breakpoints, making it difficult to choose a stiffness threshold that was applicable for all load-deflection curves; nevertheless, our threshold of 5% lead to excellent linear fits with low variability (R2≥0.94±0.02). This value was chosen ad hoc after satisfying a select subset of samples during preliminary testing and then not altered thereafter unless fitting failed or NZ boundaries were obviously wrong. We propose against modifying the stiffness threshold to satisfy curves on case-by-case basis as this will introduce bias. This method has previously been found efficacious in capturing a change in NZ after acute injury in rat specimen when applied to axial tension/compression load-deflection curves with visually obvious breakpoints (Torre et al., 2019). Incorporating the results of our study, it appears applicable in a subset of load-deflection curves with apparent breakpoints between the NZ and elastic zone and when hysteresis between unloading and loading limbs is small.

4.2. Method usage in the spine biomechanics literature

Several NZ calculation methods have been used in the literature with a most pronounced usage of the ZL method (Fig. 7). Our results underscore its efficacy and also argue favorably for the applicability of the TL and EEZ methods. A number of publications continue to use manual, visual, and undocumented techniques for obtaining NZ. We encourage the usage of established methods rather than visually inspecting the load-deflection curve in order to minimize bias. Prior work has shown that methods do not agree in their NZ measurement and should not be assumed to be interchangeable (Di Pauli von Treuheim et al., 2020). The current findings build on this point and indicate that methods that disagree in their measurement may still equally capture significant NZ changes (i.e. ZL and EEZ). We therefore conclude that NZ methods differ in their agreeability but may not necessarily impact conclusions, and strongly recommend the methodology be documented.

4.3. Clinical implication

This study is the first to show that “apparent” implant efficacy at capturing significant NZ changes depends on the NZ method. Quantifying stability is routine for in vitro spinal implant efficacy studies but remains an uncommon clinical practice. Few studies have used NZ to evaluate surgical candidacy (Esmailiejah et al., 2018; Voinier et al., 2020; Volkheimer et al., 2018) but quantitative fluoroscopy appears promising, cited for its diagnostic potential for lumbar instability (Breen et al., 2015; Esmailiejah et al., 2018; Zhao et al., 2016). Alternatively, Voinier et al. quantified asymmetrical instability in scoliotic patients using inertia motion sensors (Voinier et al., 2020). Objectively measuring instability in vivo without invasive procedures remains uncommon clinically but may grow in popularity with increased confidence during in vitro testing.

5. Limitations

This study was limited by our designation of destabilizing and restabilizing events. While a target numerical NZ value constituting these events was not predetermined, we established that destabilization and re-destabilization should increase NZ and that restabilization should decrease NZ. A second limitation relates to the use of only 6 specimens obtained from 4 different cadavers, where only lumbar levels with similar baseline characteristics were tested. Although this does not represent the variability across the entire healthy human spine, this sample does provide a range of different human motion segment behaviors by which to challenge our NZ methods. Another limitation surrounds our testing protocol, which applies uncoupled pure moments at a constant velocity and without preload. While alterations in loading rate between 0.5–5.0 % are not known to influence load-deflection curves (Wilke et al., 1998a), we acknowledge that the influence of preload on NZ shows mixed data. Prior studies using our spine simulator have shown that a follower load (the multisegmental equivalent of preload) minimally affects the load-deflection curve shape (Rohlmann et al., 2001), that in vitro forces were comparable to in vivo forces (Wilke et al., 2001), and that our pure moment protocol generates reproducible results necessary for implant testing (Wilke et al., 1998b). Permutations to testing conditions may nevertheless alter the load-deflection curve profile and consequently affect comparisons between NZ methods. We therefore caution against applying results from this study to dissimilar profiles but maintain that our protocol and findings provide the first objective comparison demonstrating the differing sensitivity of methods to detect NZ changes.

6. Conclusions

This study shows that using NZ to quantify instability following destabilizing and restabilizing events depends on the NZ method. We found the TL, ZL, and EEZ methods to be most sensitive at capturing changes in motion segment stability. Given its minimal usage in the literature, our study recommends the EEZ method as a viable alternative. Contrarily, the DS and ST methods were inferior to the others, and we found the DS method to be unsuitable when testing in axial rotation. Nevertheless, these methods may prove to be valuable when applied to load-deflection curves with different profiles. All NZ methods were introduced previously, and this investigation does not change conclusions from those prior studies but rather highlights that NZ method influences NZ value and sensitivity to motion segment changes when applied to a single, broadly relevant dataset. Further investigation with heterogenous datasets is warranted to generate robust, comprehensive recommendations. Identifying a singularly superior method may not be possible given the heterogeneity load-deflection curves.

Acknowledgements

We thank Anne Köhler and Elisabeth Debout who contributed with analysis and Carsten Hackenbroch, the radiologist who scanned specimen. This work was funded by R01AR057397 (JCI) from NIH/NIAMS and by the German Research Foundation (DFG, Project No. WI1352/14-3).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of Competing Interest

The authors declare no known competing financial interests or personal relationships that could have appeared to influence the work. Dr. Andrew C. Hecht, MD holds royalties for Zimmer spine $16,000 and Atlas spine $8,000.

References

  1. Aebi M, 2005. The adult scoliosis. European spine journal : official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society 14, 925–948. [DOI] [PubMed] [Google Scholar]
  2. Boxberger JI, Sen S, Yerramalli CS, Elliott DM, 2006. Nucleus pulposus glycosaminoglycan content is correlated with axial mechanics in rat lumbar motion segments. Journal of orthopaedic research : official publication of the Orthopaedic Research Society 24, 1906–1915. [DOI] [PubMed] [Google Scholar]
  3. Breen AC, Dupac M, Osborne N, 2015. Attainment rate as a surrogate indicator of the intervertebral neutral zone length in lateral bending: an in vitro proof of concept study. Chiropractic & manual therapies 23, 28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cohen J, 2013. Statistical power analysis for the behavioral sciences. Academic press. [Google Scholar]
  5. Costi J, Ledet EH, O’Connell GD, 2021. Spine biomechanical testing methodologies: The controversy of consensus vs scientific evidence. JOR Spine e1138. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Di Pauli von Treuheim T, Torre OM, Mosley GE, Nasser P, Iatridis JC, 2020. Measuring the neutral zone of spinal motion segments: Comparison of multiple analysis methods to quantify spinal instability. JOR spine 3, e1088. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Esmailiejah AA, Abbasian M, Bidar R, Esmailiejah N, Safdari F, Amirjamshidi A, 2018. Diagnostic efficacy of clinical tests for lumbar spinal instability. Surg Neurol Int 9, 17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Gay RE, Ilharreborde B, Zhao K, Zhao C, An KN, 2006. Sagittal plane motion in the human lumbar spine: comparison of the in vitro quasistatic neutral zone and dynamic motion parameters. Clinical biomechanics (Bristol, Avon) 21, 914–919. [DOI] [PubMed] [Google Scholar]
  9. Mehta S, Bastero-Caballero RF, Sun Y, Zhu R, Murphy DK, Hardas B, Koch G, 2018. Performance of intraclass correlation coefficient (ICC) as a reliability index under various distributions in scale reliability studies. Stat Med 37, 2734–2752. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Mitchell UH, Hurrell J, 2019. Clinical spinal instability: 10 years since the derivation of a clinical prediction rule. A narrative literature review. J Back Musculoskelet Rehabil 32, 293–298. [DOI] [PubMed] [Google Scholar]
  11. Panjabi MM, 1992. The stabilizing system of the spine. Part II. Neutral zone and instability hypothesis. Journal of spinal disorders 5, 390–396; discussion 397. [DOI] [PubMed] [Google Scholar]
  12. Panjabi MM, 2003. Clinical spinal instability and low back pain. J Electromyogr Kinesiol 13, 371–379. [DOI] [PubMed] [Google Scholar]
  13. Rohlmann A, Neller S, Claes L, Bergmann G, Wilke HJ, 2001. Influence of a follower load on intradiscal pressure and intersegmental rotation of the lumbar spine. Spine 26, E557–561. [DOI] [PubMed] [Google Scholar]
  14. Sarver JJ, Elliott DM, 2005. Mechanical differences between lumbar and tail discs in the mouse. Journal of orthopaedic research : official publication of the Orthopaedic Research Society 23, 150–155. [DOI] [PubMed] [Google Scholar]
  15. Smit TH, van Tunen MS, van der Veen AJ, Kingma I, van Dieen JH, 2011. Quantifying intervertebral disc mechanics: a new definition of the neutral zone. BMC musculoskeletal disorders 12, 38. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Stokes IA, Gardner-Morse MG, Henry SM, 2011. Abdominal muscle activation increases lumbar spinal stability: analysis of contributions of different muscle groups. Clinical biomechanics (Bristol, Avon) 26, 797–803. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Thompson RE, Barker TM, Pearcy MJ, 2003. Defining the Neutral Zone of sheep intervertebral joints during dynamic motions: an in vitro study. Clinical biomechanics (Bristol, Avon) 18, 89–98. [DOI] [PubMed] [Google Scholar]
  18. Torre OM, Evashwick-Rogler TW, Nasser P, Iatridis JC, 2019. Biomechanical test protocols to detect minor injury effects in intervertebral discs. Journal of the mechanical behavior of biomedical materials 95, 13–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Voinier SD, Agnew MJ, Carmouche JJ, 2020. Passive stiffness characteristics and neutral zone quality of the scoliotic lumbar torso in the principle anatomical planes of motion. Clinical biomechanics (Bristol, Avon) 80, 105162. [DOI] [PubMed] [Google Scholar]
  20. Volkheimer D, Galbusera F, Liebsch C, Schlegel S, Rohlmann F, Kleiner S, Wilke HJ, 2018. Is intervertebral disc degeneration related to segmental instability? An evaluation with two different grading systems based on clinical imaging. Acta radiologica (Stockholm, Sweden : 1987) 59, 327–335. [DOI] [PubMed] [Google Scholar]
  21. Wilke HJ, Jungkunz B, Wenger K, Claes LE, 1998a. Spinal segment range of motion as a function of in vitro test conditions: effects of exposure period, accumulated cycles, angular-deformation rate, and moisture condition. Anat Rec 251, 15–19. [DOI] [PubMed] [Google Scholar]
  22. Wilke HJ, Mehnert U, Claes LE, Bierschneider MM, Jaksche H, Boszczyk BM, 2006. Biomechanical evaluation of vertebroplasty and kyphoplasty with polymethyl methacrylate or calcium phosphate cement under cyclic loading. Spine 31, 2934–2941. [DOI] [PubMed] [Google Scholar]
  23. Wilke HJ, Ressel L, Heuer F, Graf N, Rath S, 2013. Can prevention of a reherniation be investigated? Establishment of a herniation model and experiments with an anular closure device. Spine 38, E587–593. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Wilke HJ, Rohlmann A, Neller S, Schultheiss M, Bergmann G, Graichen F, Claes LE, 2001. Is it possible to simulate physiologic loading conditions by applying pure moments? A comparison of in vivo and in vitro load components in an internal fixator. Spine 26, 636–642. [DOI] [PubMed] [Google Scholar]
  25. Wilke HJ, Wenger K, Claes L, 1998b. Testing criteria for spinal implants: recommendations for the standardization of in vitro stability testing of spinal implants. European spine journal : official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society 7, 148–154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Zengerle L, Köhler A, Debout E, Hackenbroch C, Wilke HJ, 2020. Nucleus replacement could get a new chance with annulus closure. European spine journal : official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society 29, 1733–1741. [DOI] [PubMed] [Google Scholar]
  27. Zhao KD, Ben-Abraham EI, Magnuson DJ, Camp JJ, Berglund LJ, An KN, Bronfort G, Gay RE, 2016. Effect of Off-Axis Fluoroscopy Imaging on Two-Dimensional Kinematics in the Lumbar Spine: A Dynamic In Vitro Validation Study. Journal of biomechanical engineering 138, 054502. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES