Background: Segmental motion is a fundamental characteristic of the thoracic spine; however, studies of segmental ranges of motion have not been summarized or analyzed. The purpose of the present study was to present a summary of the literature on intact cadaveric thoracic spine segmental range of motion in each anatomical plane.
Methods: A systematic MEDLINE search was performed with use of the terms “thoracic spine,” “motion,” and “cadaver.” Reports that included data on the range of motion of intact thoracic human cadaveric spines were included. Independent variables included experimental details (e.g., specimen age), type of loading (e.g., pure moments), and applied moment. Dependent variables included the ranges of motion in flexion-extension, lateral bending, and axial rotation.
Results: Thirty-three unique articles were identified and included. Twenty-three applied pure moments to thoracic spine specimens, with applied moments ranging from 1.5 to 8 Nm. Estimated segmental range of motion pooled means ranged from 1.9° to 3.8° in flexion-extension, from 2.1° to 4.4° in lateral bending, and from 2.4° to 5.2° in axial rotation. The sums of the range of motion pooled means (T1 to T12) were 28° in flexion-extension, 36° in lateral bending, and 45° in axial rotation.
Conclusions: The pooled ranges of motion were similar to reported in vivo motions but were considerably smaller in magnitude than the frequently referenced values reported prior to the widespread use of biomechanical testing standards. Improved reporting of biomechanical testing methods, as well as specimen health, may be beneficial for improving on these estimations of segmental cadaveric thoracic spine range of motion.
Every aspect of spine surgery or treatment requires knowledge of spine kinematics. Importantly, expectations about deformity correction in the spine are closely related to information on normal spine motion. The range of motion of each spine segment, or functional spine unit (FSU), is the most basic and fundamentally important parameter for establishing spine function. Unlike the range of motion of many joints, such as the knee and elbow, the range of motion of spine segments, particularly thoracic spine segments, is difficult to measure in vivo without invasive procedures. Moreover, noninvasive in vivo measurements1-7, such as skin marker-based measurements, are affected by many uncontrollable variables. Consequently, knowledge about the range of motion of the thoracic spine is based primarily on studies that have involved the use of in vitro cadaveric models8-12.
Nearly four decades ago, White and Panjabi reported typical range-of-motion values for each level of the human spine13,14. The values for the thoracic spine were largely based on both the experience of the authors and the dissertation work by White, who applied off-axis compression and torsional loading to produce bending and axial rotation, respectively, in cadaveric spines8. However, that publication preceded the widespread usage of pure moments in numerous biomechanical studies that were published over the next decades. Notwithstanding succeeding publications, the authors of both clinical3,7 and in vitro15-21 studies have frequently referenced the range-of-motion values reported in White and Panjabi’s spine biomechanics textbook (which has been cited >3700 times, according to Google Scholar) to evaluate or contrast against their own results. That is, many authors have ignored numerous more recent studies that have included the range of motion of cadaveric thoracic spines, even though the majority of those studies involved the use of methods that are well recognized in the community as testing standards22,23. To our knowledge, no study has presented a review of the thoracic range of motion as reported in the literature.
The purpose of the present study was to establish and present an overview of the ranges of motion that have been estimated for each level of the thoracic spine.
Materials and Methods
Systematic MEDLINE Search
A systematic MEDLINE search was performed to identify in vitro biomechanical studies investigating the range of motion of the thoracic spine. Specifically, a MEDLINE search was performed using the search terms “thoracic spine,” “motion,” and “cadaver.” All of the abstracts obtained from the above search terms were reviewed to identify articles of interest. Once studies involving biomechanical testing on thoracic spine specimens had been identified, the full manuscripts were reviewed. In order to ensure that no articles were missed with use of the above search terms, the references cited in all of the identified articles were also reviewed. Finally, once an article was identified and chosen for inclusion in the present study, the “Related Citations” option of PubMed was also used to identify any additional articles missed by the initial search criteria and the references of the included articles.
Only studies in which the range of motion of intact human cadaveric thoracic spines was reported were included. Studies in which the thoracic range of motion was evaluated only after simulated injury or treatment were excluded. Studies involving only cervical or lumbar motion segments were excluded. In addition, studies involving biomechanical testing on nonhuman specimens, e.g., canine specimens, were excluded. Clinical studies and computational studies were also excluded.
There were several instances in which a group of authors had written multiple articles that included the same intact motion data for the same groups of specimens. In those cases, only the article with the most comprehensive reporting of the intact range of motion data was included.
Once the relevant articles had been identified, each was reviewed, and the relevant independent and dependent variables were extracted. Independent variables included the tested motion segments in each multi-segment unit (e.g., T1-T2 or T4-T7), the number of specimens, specimen age, the presence of the rib cage (yes or no), the type of loading (pure moment, cantilever bending, or four-point bending), the preload, the type of loading control (displacement or moment), the loading rate, and the maximum bending moment (or load) in each of the three planes (flexion-extension, lateral bending, and axial rotation). Dependent variables were the ranges of motion in each reported anatomical plane: flexion-extension, lateral bending, and/or axial rotation. For studies that only investigated range of motion in unilateral/unidirectional directions (e.g., mean flexion range of motion and mean extension range of motion), the means were summed to provide average values for bilateral/bidirectional motions (flexion-extension range of motion). Segmental motions were estimated separately for studies that applied loading to thoracic spines without the rib cage and those with the rib cage intact.
Means and standard deviations of range of motion values were extracted. In a few cases, only medians were reported, and these data were recorded separately. For studies in which raw motion measurements were reported for each individual specimen, means and standard deviations were calculated. Only data that were clearly reported were used. Finally, in studies in which the motion values were not reported in the text or in tables, the values were extracted from the corresponding graphs when available.
Simplified Estimation of Range of Motion at Each Motion Segment
In several studies, range of motion was reported individually for each level of a multi-segment spine specimen (e.g., T1-T2, T2-T3, or T11-T12), whereas in others only the total multi-segment range of motion (e.g., T1-T12) was reported, rather than the motion at each segment. In the latter group of studies, the range of motion for each single level was approximated by dividing the total range of motion of the multi-segment spine specimen by the number of motion levels included in the specimen. For example, if the reported T1-T12 range of motion was 44°, this total range of motion was divided by eleven motion segments, providing an approximation of 4° of motion per level. That is, an entry would be made for each of the eleven motion segments (i.e., T1-T2, T2-T3,… T11-T12), each with a motion of 4°. Although this approximation introduced some error, it was necessary in order to compare range of motion across the studies and to maximize the number of studies that were included.
All statistical analyses were performed with use of SPSS 19.0 statistical software (IBM). The median, minimum, and maximum values of the estimated segmental motions among the studies were calculated. For studies that applied standard pure moment testing to cadaveric thoracic specimens without the rib cage and reported range of motion, the means, standard deviations, pooled means, and pooled standard deviations were calculated on the basis of the range of motion for each of the estimated motion segments (T1-T2, T2-T3,…T11-T12) in each of the three loading directions: flexion-extension, lateral bending, and axial rotation. High-low plots were used to represent the spreads in the estimated segmental motions.
The original MEDLINE search yielded 161 abstracts. A review of the 161 abstracts identified twenty-seven articles that included cadaveric thoracic spine range-of-motion measurements in degrees10-12,20,21,24-45. After reviewing the references cited in each of the identified articles as well as reviewing the abstracts of 9223 “Related Citations,” including repeats, fourteen additional articles were identified8,9,46-57, resulting in a total of forty-one identified articles. Five articles included redundant data (i.e., data from the same group of specimens and same author group) and thus were excluded24,25,34,47,53. Two studies that included only average motion of the lower thoracic spine and upper lumbar spine were excluded30,43. Finally, one of the identified articles was the original thesis by Dr. Augustus A. White III8. As the original table of typical ranges of motion reported by White and Panjabi was largely based on that work, the motions were not included in the present analysis. Instead, the results of the present analysis were compared with the previously reported motions, discussed later. Following the exclusion of those eight articles, thirty-three unique articles remained for the present analysis9-12,20,21,26-29,31-33,35-42,44-46,48-52,54-57.
Within the thirty-three included articles, a wide range of methods was applied, including differences in tested levels, intact specimen condition (i.e., with or without the rib cage), type of loading, loading rate, and loading magnitude as well as methods of measurement of motion and reporting of data. Five studies tested full-length human thoracic spines11,35,37,38,57, three studies tested individual functional spine units45,51,52, and the remaining studies tested multi-segment thoracic spine specimens. The mean donor age in the studies ranged from forty-three to eighty-three years.
The majority (twenty-three) of the included articles generally applied standard pure moments to thoracic spine specimens without the rib cage and sternum intact (Table I). There was wide variation in the reported mean range-of-motion values among these studies in each of the three motion planes. In these studies, moments ranging from 1.5 to 8 Nm were applied to a variety of thoracic regions and specimen lengths. Angular displacement-controlled rotation rates ranged from 0.5°/s to 2°/s, whereas torque-controlled moment rates ranged from 0.3 to 1 Nm/s; however, in many of the articles, the loading rate was not reported. In addition, as many of the studies involved the use of weight-pulley loading systems, the loads were applied incrementally, e.g., in steps of 2.25 Nm.
Five additional studies also assessed thoracic spine range of motion without the rib cage but with use of loading mechanisms other than pure moments. The loading mechanisms included four-point bending loads, cantilever bending, or unknown loading types (Table II). Similar variations existed among these studies in terms of the reported mean thoracic range of motion.
Large variations also were observed in the seven articles evaluating thoracic range of motion in specimens with the rib cage intact (Table III). Specifically, the full-length thoracic range of motion (T1-T12) ranged from 7.93° to 33.9° in flexion-extension, from 10.4° to 47.4° in lateral bending, and from 23.0° to 44.9° in axial rotation.
Estimated Range of Motion at Each Motion Segment (Intact Thoracic Spines without Rib Cage)
As described, on the basis of the twenty-three studies that assessed mean thoracic range of motion in cadaveric specimens without the rib cage, medians and ranges were calculated. In many of the articles, the range of motion of each individual segment was reported. In others, multi-segment segmental motions were approximated by dividing multi-segment range of motion by the number of segments for each specimen9,12,20,21,27,28,30,31,33,36,39,40,42,44-46,48-50,53,54,56,57. The estimated segmental motion ranged from 1.1° to 6.4° in flexion-extension, from 1.1° to 7° in lateral bending, and from 1.2° to 7.2° in axial rotation (Table IV). Among the studies in which means and standard deviations were reported, the pooled means ranged from 1.9° at T2-T3 and T3-T4 to 3.8° at T11-T12 in flexion-extension, from 2.1° at T1-T2 to 4.4° at T11-T12 in lateral bending, and from 2.4° at T1-T2 to 5.2° at T5-T6 in axial rotation. It should be noted that only one study each was used for the T1-T2 segment46, the T2-T3 segment9, and the T3-T4 segment9 as only those particular studies evaluated the mean ranges of motion of specimens including those segments.
The magnitudes of applied moment varied widely among the different studies; however, there was no systematic increase in range of motion with increased moment in any plane. Specifically, linear regression analysis of range of motion as a function of applied moment indicated slopes of 0.39°/Nm (p = 0.33), 0.04°/Nm (p = 0.95), and −0.14°/Nm (p = 0.88) in flexion-extension, lateral bending, and axial rotation, respectively.
Overall, estimated segmental motions were smallest in flexion-extension and largest in axial rotation. According to the pooled segmental range of motion means and standard deviations, large variability in range of motion existed at each thoracic motion segment in each of the three loading directions. In addition, pooled estimated segmental motion in flexion-extension was largest at the cranialmost and caudalmost motion segments, whereas pooled estimated segmental motion in lateral bending and in axial rotation were largest in the middle regions of the thoracic spine.
Given the estimated range of motion for each thoracic motion segment, an approximation was then also made for the total thoracic spine range of motion (T1-T12) based on the sum of the individual segments. This estimated total thoracic spine motion (T1-T12) was 27.9° in flexion-extension, 36.3° in lateral bending, and 44.9° in axial rotation.
Comparison of Segmental Range-of-Motion Estimates
As indicated, previous investigators have frequently referenced the spine segment ranges of motion reported by White and Panjabi to evaluate the results of their research studies3,7,16-21. However, in comparison with the motions reported by White and Panjabi13,14, the estimated segmental motions in all of the more recent studies that have been conducted over the last three decades were substantially smaller for nearly every thoracic motion segment in flexion-extension (Fig. 1), lateral bending (Fig. 2), and axial rotation (Fig. 3). The results are strikingly different and indicate a large variability among range of motion values at each thoracic level.
Intact thoracic spine range of motion in three planes was extracted from the literature to provide a summary of, and to establish the spread of, estimated segmental motions reported in previous studies. The disparities among studies may be attributed not only to inherent differences among thoracic spines but also to differences in the loading mechanisms as well as the motion measurement methods.
In the widely cited reports by White and Panjabi13,14, total thoracic range of motion (T1-T12) ranged from approximately 34° to 90° in flexion-extension. In contrast, the sum of the pooled means in the present study (T1-T12) was approximately 28° in flexion-extension, substantially smaller than the previously reported range. Similarly, White and Panjabi reported an estimated total range of motion of 94° to 164° in lateral bending, compared with the value of only 36° that was calculated in the present review, which also was much smaller than the previously reported range. Finally, White and Panjabi reported an estimated range of motion of 108° to 198° in axial rotation, whereas in the present review, the estimated total range of motion was only 45° in axial rotation.
As discussed, the pooled means reported in the present study were based on the results of more recent studies, which generally applied pure moment testing standards22,23. In contrast, White’s work was performed with use of off-axis compression loading modes for bending8, prior to the advent of pure moment testing standards. On the other hand, even studies involving the use of cantilever bending35,51 and four-point bending41 also have demonstrated motions substantially smaller than those reported by White and Panjabi13,14, suggesting that pure moment testing is not the only reason for the differences. Another potential issue may be the differences in the method of measurement of range of motion during mechanical testing. Specifically, more recent techniques of motion measurement, such as three-dimensional optical motion tracking, may be more accurate than the methods employed by White, which included measurements that were made with use of extensometers, displacement gauges, and radiographs8.
Biomechanical standards for torque-rotation testing of spines were in large part designed to allow for comparisons of results across studies51; however, variability among the results clearly exists. As the standard should theoretically produce comparable results, the variability in the results is likely attributed to the inherent variability in flexibility among cadaveric specimens. Additionally, disc degeneration has been shown to affect the motions of the lumbar spine58-63. The same is likely true for the thoracic spine; however, this topic has not been investigated, to our knowledge. Moreover, experimental factors may play a role in the resultant motions, including, for example, loading magnitude, loading rate, and measurement apparatus.
Unfortunately, it is difficult to adjust or normalize the range-of-motion data for comparison among studies that have involved the use of different methods and experimental setups. For example, different studies have applied different magnitudes of pure moments, but since the response of the spine to moments is hysteretic and highly nonlinear, it is impossible to normalize these measurements for comparison. Another important experimental factor is whether or not to include the rib cage. While the majority of the studies in the literature did not include the intact rib cage in their models, some investigators have argued that this is an important component for determining the range of motion of the thoracic spine. Therefore, in the present study, we reported the range-of-motion values obtained with and without the rib cage separately. It is hoped that, with improved reporting of the biomechanical methods as well as improved identification and reporting of specimen conditions, e.g., disc degeneration grade, presence of osteophytes, etc., the unwanted variance due to experimental conditions throughout the thoracic spine literature may be reduced.
Several studies have estimated and measured thoracic spine range of motion in vivo1-7. The reported full thoracic motion (T1-T12) has ranged from approximately 25.6° to 71° in flexion-extension3,4,6,7, from approximately 31.2° to 75° in lateral bending3-5, and from approximately 41.8° to 95.5° in axial rotation1-4. In comparison, the sums of the pooled motions in the present study (T1-T12) were 28°, 36°, and 45° in flexion-extension, lateral bending, and axial rotation, respectively. Each of these values fell within the lower ranges of reported in vivo motion, suggesting that pure moment testing may produce physiological ranges of motion. Compared with the ranges of motion reported by White and Panjabi13,14, the motions reported for flexion and extension in clinical studies3,4,6,7 have been comparable; however, the clinical findings for lateral bending and axial rotation1-5 have been substantially smaller and more comparable with the numbers reported in the present study.
The results of the present study are clearly limited by the simplified method that was used to estimate the range of motion of each individual thoracic segment in each plane. Specifically, in several studies, only the sum of motions of several segments was reported, rather than the motion at each individual level. In those cases, we estimated the motion at each level by dividing the total range of motion by the number of segments included. This simplification introduced an error as the motions are not uniformly distributed. However, including an estimated value, notwithstanding the simplification, was considered preferable to excluding those studies. With increased use of optical motion measurements and the increased capabilities of measuring motions at each spine segment, future studies should evaluate both global and segmental motions. Once those data become routinely available, more accurate estimations potentially may be developed. In addition, the analysis in the present study was limited to the details provided in the published literature. With more detailed explanations of experimental setups, such as including loading rates and specifics of loading mechanisms, more thorough analyses may be performed and a clearer picture of thoracic spine kinematics may emerge.
Many aspects of spine surgery stand to benefit from specific and accurate knowledge of the range of motion of spine segments. Preclinical biomechanical models involving cadaveric spines offer great potential for providing critical data on surgical approaches and implant performance9,64,65. Importantly, corrective surgery for spine deformity requires range-of-motion information as well as flexibility data, specifically for the thoracic spine. Biomechanical data are essential for preoperative planning and postoperative evaluation of the outcomes of such procedures. As cadaveric spines are typically old and degenerated and as most do not have untreated deformities, some have questioned the clinical relevance of biomechanical models for spine deformity procedures based on these specimens. However, the complex material and structural properties of the spine are virtually impossible to model without taking advantage of data from cadaveric spines as even the best computer-model simulations use material properties from cadaveric spines. Therefore, despite its shortcomings, cadaveric biomechanical modeling serves as an indispensable tool in advancing spine deformity surgery.
In summary, estimated segmental thoracic range of motion was established for each level of the thoracic spine in flexion-extension, lateral bending, and axial rotation. Given the estimated segmental motions at each thoracic motion segment, the total thoracic spine motion (T1-T12) was then approximated on the basis of the sum of the individual segments. The estimated total thoracic spine motion (T1-T12) was approximately 27.9° in flexion-extension, 36.3° in lateral bending, and 44.9° in axial rotation. The estimated motions fell within the lower ranges of reported in vivo motions but were substantially smaller than the frequently referenced values of cadaveric motions that have been established previously. The wide variation in the results suggests the need for improved characterization of experimental specimens, such as specimen health or intervertebral disc degeneration. Such improved specimen characterization, as well as improved reporting of experimental details and setups, may enable more accurate and comprehensive estimations to be developed in the future, ultimately establishing a set of normal motions at each level of the thoracic spine that could be beneficial for the diagnosis and treatment of thoracic spine disorders.
Source of Funding: The present study was supported by the Orthopaedic Institute for Children. No external funds were received for this study.
Investigation performed at the J. Vernon Luck, Sr., M.D. Orthopaedic Research Center, Orthopaedic Institute for Children, University of California, Los Angeles, Los Angeles, California
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. None of the authors, or their institution(s), have had any financial relationship, in the thirty-six months prior to submission of this work, with any entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. Also, no author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.
- Copyright © 2016 by The Journal of Bone and Joint Surgery, Incorporated