Authors: Yuko Nakahira, Masami Iwamoto, Tatsuya Igawa, Ken Ishii
Categories: Article, Biomedical engineering, Muscle, Orthopaedics, Computational science
Source: Scientific Reports
Authors: Yuko Nakahira, Masami Iwamoto, Tatsuya Igawa, Ken Ishii
The occurrence of diseases characterized by irregular spinal alignment, such as kyphosis, lordosis, scoliosis, and dropped head syndrome (DHS) is increasing, particularly among older adults. DHS is characterized by an excessive forward tilt of the head and neck, causing the head to droop. Although it is believed that muscle activity plays a role in both the onset and treatment of DHS, the underlying mechanisms remain unclear. To elucidate the mechanism, we used a human body finite element model, which included the erector spinae muscle group, and a muscle controller with fixed legs for spinal posture stabilization. The model replicated muscle activation levels during the maintenance of an upright posture under gravity, similar to those obtained from experimental data. Parametric simulations to investigate the effect of each spinal muscle impairment on upright posture with and without compensatory activities of the other muscles suggest that trunk extensors; the multifidus L1-S and longissimus thoracis muscles, and hip flexors; psoas major and iliacus muscles play an integral role in maintaining an upright posture. These findings support the results of a rehabilitation study that reported that exercises targeting the trunk, psoas muscles, and cervical extensors could improve global spinal alignment and clinical outcomes in DHS.
Diseases characterized by abnormal spinal alignment such as kyphosis, lordosis, scoliosis, and dropped head syndrome (DHS), have become increasingly prevalent, particularly among aging populations. DHS is characterized by an excessive forward tilt of the head and neck, causing the head to droop. Although relatively few large-scale studies have examined the epidemiologic trends of DHS, some research on its etiology and management, including case reports and case series, exists^1,2^. Based on combined data from all published cases of DHS, the female-to-male ratio is 2, with a mean age at presentation of 74.5 years^3^. Although DHS is relatively rare, its prevalence is likely to increase as life expectancy rises. Affected individuals experience difficulty in lifting the head independently. DHS not only manifests as a drooping head, but also results in a posture in which the chin touches the chest. The primary symptoms of DHS include neck pain and anterior flexion impairment. Furthermore, excessive flexion of the head and neck can lead to difficulties in walking and eating, as well as trismus, dysphagia, and skin lesions (ulcerations) in the anterior neck and chest. The causative factors for DHS are diverse and include neurogenic factors (arising from conditions such as various neurological disorders), myogenesis (resulting from various muscle diseases), and cervical spondylosis-related causes (arising from intervertebral disc bulging and bone-spur formation). However, the exact mechanism underlying its occurrence remains unclear^4^.
The treatment for DHS includes surgical intervention and conservative therapies such as orthotic treatment, pharmacotherapy, and physical therapy. The effectiveness of physical therapy, including cervical extension exercises, has begun to be demonstrated^5^, but the precise mechanisms underlying its effectiveness remain unclear. Elucidating the mechanism of action can aid in the development of more effective treatment approaches. The onset and treatment of DHS are believed to be associated with muscle activity^6–8^. Although muscle activity can be measured using electromyography (EMG), this method is limited to a small number of muscles on the body surface^9^. Muscles involved in spinal movements, such as the flexion and extension of the head and neck are located deep within the body^10^, making it challenging to measure their activity and placing a significant burden on study participants during experiments.
Conversely, predicting muscle activity through simulation using a human body model is useful for understanding the mechanisms underlying the development of DHS and the effects of rehabilitation. Recent advances in the construction of human body models using computer simulations and finite element (FE) analysis have enabled a detailed analysis of the complex dynamics of the musculoskeletal system^11–16^. This study aimed to determine the effect of individual muscle activities on the maintenance of an upright posture using a human body FE model to elucidate the mechanism underlying the onset of symptoms related to spinal asymmetry.
The human muscle FE model used in this study (Fig. 1a, b) represents a typical adult male 175 cm tall, weighing 77 kg aged in the 30 s to 40 s, incorporating approximately 320 skeletal muscles modeled with solid elements. This model, based on The Total Human Model for Safety (THUMS) version 3^17^ for crash injury analysis, was tailored to the size of a standard adult male following the procedure outlined below. First, we adjusted the masses of 15 body parts (head, neck, thorax, abdomen, buttocks, and left and right upper arms, lower arms, thighs, lower legs, and feet) to match the standard male size, using measured data from previous studies^18,19^. This adjustment was made by altering mass densities and the thicknesses of bones, ligaments, internal organs, and muscles within their experimental ranges^20,21^ as shown in Supplementary Fig. S1. Second, we modified the model’s posture from a standing position to a sitting position and confirmed that the center of gravity of each body part remained consistent with that of the measured data^18,19^. We then integrated the model with a muscle model that reproduces the three-dimensional shapes of each skeletal muscle based on data from the Visible Human Project (National Institutes of Health, USA).Fig. 1External appearance of the human body FE model (a, b) and individual modeling of the erector spinae (ES) (c) multifidus (MF) (d), longissimus thoracis (LT) (e), iliocostalis thoracis (IT) (f), and iliocostalis lumborum (IL) (g).
A key feature of the human body FE model is its spinal flexibility, comprising 24 vertebral bodies, which mimic the human spine with seven cervical, 12 thoracic, and five lumbar vertebrae. By adjusting the intervertebral disc stiffness between these vertebrae to resemble that of a human, the model replicates the flexibility of the spine^22–24^. The vertebrae in the model were validated against experimental data from previous studies^22,23^. In the simulations, L3 was gradually loaded to 15 Nm for flexion and extension, whereas L4 was fully fixed to replicate the experimental test setups (Supplementary Fig. S2(a)). Moment–angle curves for flexion and extension predicted by the model were compared with experimental data (Supplementary Fig. S2(b)).
Another important characteristic is the three-dimensional modeling of the muscles^25^. By inputting the muscle activity levels, the muscles display a contractile force proportional to their activity, enabling the representation of movements, with increased muscle stiffness, and capturing changes in muscle paths during motion^11^. In addition, the model can represent the muscle–tendon strain distribution associated with muscle soreness. The human body FE model has been useful in the field of crash safety^11,25,26^, in assessing physical strain during daily activities^27,28^, as well as in medical research^12^. To investigate the effect of muscle activity on spinal movement (Supplementary Figs. S3, S4, S5) and on posture, we conducted a detailed modeling of the ES muscles on a previous human body FE model^11,28^ (Fig. 1c–g). We decomposed the previously represented unified ES muscle groups (Fig. 1c) into individually modeled 17 multifidus (MF) on each side (Fig. 1d), longissimus thoracis (LT) (Fig. 1e), iliocostalis thoracis (IT) (Fig. 1f), and iliocostalis lumborum (IL) (Fig. 1g) muscles.
Although the deformation characteristics of the skin are more rigid at higher strain rates and more deformable at lower strain rates, no material model to reproduce these characteristics exists in LS-DYNA (ANSYS, Inc., USA), the solver for FE analysis used in this study. The skin in the original model has been modeled as rigid for high-speed deformations, which is commonly employed in collision analyses. This study focused on spinal posture changes under gravity without considering any external forces. Although the skin should be modeled as more deformable, using such a soft material model often causes computational instability. The predominant movements influencing spinal posture changes are governed by the bones and tendons, with the skin and fat layers primarily following these actions. Therefore, excluding the skin from the model was considered appropriate for this study. By excluding the skin, the total number of elements in this model was approximately 210,000 with approximately 120,000 nodes.
When using a human body muscle FE model, it is necessary to input the muscle activation levels for each of the 320 skeletal muscles. The muscle controller^29–32^ employed in this study adjusts the muscle activation levels for the muscles acting on each joint using the proportional-integral-derivative (PID) control method to minimize the difference between the target joint angles and the current joint angles for the desired posture. Given the input of the target joint angles, the controller outputs the muscle activation levels using the percent maximal voluntary contraction (%MVC), defined as the percentage ratio of the applied force to the MVC for each muscle. The output muscle activation was automatically fed into the muscle FE model, generating posture changes based on the muscle contraction exerted with forces proportional to the muscle activation levels. The muscle controller for the human body FE model was extended for further use for the ES muscles (Fig. 2). Specifically, for the MF, 17 vertebral joint angles of the origin and insertion of each MF were added for both cervical and trunk angles. Two joint angles from T1 to T12 (T1 angle with respect to the T12 coordinate system) and T6 to S (T6 angle with respect to the sacrum coordinate system) was added to the LT and IL muscles, respectively (Fig. 2, top row).Fig. 2Overview of the muscle controller extended for the erector spinae (ES) muscles.
In the muscle controller, when the current cervical joint angle is in flexion relative to the target cervical joint angle, the muscle activation level of the cervical flexor decreases and that of the cervical extensor increases (Supplementary Fig. S6). Such adjustments in muscle activation levels are performed at every time step of 1 × 10^−6^ s. Furthermore, the proportional (P) gain, integral (I) gain, and derivative (D) gain must be set to the parameters for the PID control for each joint. Based on the results of a parametric study performed to maintain an upright posture under gravity for 2 s, the P gains for the cervical, trunk, hip, and intervertebral joints were set to 5 rad^−1^, whereas the P gain for the knee joint was set to 1 rad^−1^. The I gain for all the joints was set to 0 rad^−1^ s^−1^, and the D gain was set to 0 s rad^−1^. The MVC represents the maximal force-generating capacity of a muscle or group of muscles in humans. The MVC is believed to be limited by inhibiting mechanisms in the brain^33,34^. The activation level of MVC is typically less than that achieved by electrical stimulation and does not reach 100%^35^. In experimental studies, EMG data of each muscle during specific tasks (such as brake pedal use or exercises) were normalized against the MVC, with the corresponding %MVC typically being less than 80%^11,36,37^. Therefore, the minimum and the maximum muscle activity values were set to 0% and 80% MVC, respectively. Physiological cross-sectional area (PCSA) and contribution ratios to spinal movement for the individually modeled MF, LT, IT, and IL muscles are listed in Table 1. The PCSA of the MF attached to the T10-S was set with reference to the measurements by^38^. For the other muscles, the PCSA was set based on the overall PCSA of the ES muscles^39^. The contribution ratio of each muscle to various spinal movements is generally set based on anatomical texts^40^, assigning values of 0.7–0.9 for agonist muscles and 0.05–0.4 for accessory muscles. MF exhibited changes in muscle activity similar to those of the LT^41^, the contribution ratios of the MF, LT, IT, and IL muscles were determined by trial and error to represent not only the behavior within their attachment ranges, but also the overall behavior of the spine under the restriction that the total contribution ratio of each muscle was equal to one. For the MF, the contribution ratios of the MF (T1–T4) was set not only for the T1 angle (extension) in the T4 coordinate system but also for trunk angles (extension and lateral bending) in the sacral coordinate system and a T1 angle (extension) in the T12 coordinate system (Table 1). The contribution ratios of the LT were set for extension and lateral bending of trunk angles in the pelvic coordinate system, whereas those for the IT were set at the trunk and T1 angle and those of the IL were set as the trunk angles and T6 angle (extension) in the sacral coordinate system (Table 1).Table 1Physiological cross-sectional areas and contribution ratios to various spinal movements for multifidus (MF), longissimus thoracis (LT), iliocostalis thoracis (IT), and iliocostalis lumborum (IL)
The model was used in a parametric study to elucidate the relationship between electrical spinae muscle activity and upright posture. Table 2 lists the simulation conditions of the parametric study conducted by altering the activity of muscles on the spinal movements. No muscles were activated in Case A. In Case B, a PID controller was used for all muscles, where current joint angles were fed back to the PID controller, and muscle activity are determined to reproduce the target posture at each time step. This type of controller is intended to simulate spinal reflexes. We assumed Case B as an intact condition and Cases C–Z as conditions where one or more muscles were impaired. When any muscle in the human musculoskeletal system is damaged, other muscles and ligaments often compensate by exerting extra effort. Understanding how the roles of the damaged muscles are compensated by other parts of the musculoskeletal system is valuable from a clinical and rehabilitation perspective^42^. Therefore, we conducted the parametric study of Cases C–Z to investigate the compensatory mechanisms of muscles in spinal movements. In Cases C–N with compensation, the activity of target muscles was consistently set to the percent maximum voluntary contraction (%MVC) of zero (0%MVC), while the activities of all other muscles were determined by using PID controller. In Cases O–Z without compensation, a constant 0%MVC muscle activity was inputted for the target muscles, while the time history of the muscle activity obtained from Case B was input for all other muscles.Table 2Presence or absence of muscle activity as conditions for the parameter study–: Inactive (Input 0%MVC constant muscle activation level)◯: Automatic adjustment by muscle controller▲: Input the muscle activation level of the calculation result for Case B.Sca: Semispinalis capitis, SCe: Semispinalis Cervicis, IC: Iliocostalis Cervicis, OCI: Obliquus Capitis Inferior, RMa, Rectus Capitis, Posterior Major, RMi: Rectus Capitis, Posterior minor, LCa: Longissimus Capitis, LCe: Longissimus Cervicis, MF(T1-T4): Multifidus_T1-T4, MF(T2-T5): Multifidus_T2-T5, MF(T3-T6): Multifidus_T3-T6, MF(T4-T7): Multifidus_T4-T7, MF(T5-T8): Multifidus_T5-T8, MF(T6-T9): Multifidus_T6-T9, MF(T7-T10): Multifidus_T7-T10, MF(T8-T11): Multifidus_T8-T11, MF(T9-T12): Multifidus_T9-T12, MF(T10-L1): Multifidus_T10-L1, MF(T11-L2): Multifidus_T11-L2, MF(T12-L3): Multifidus_T12-L3, MF(L1-L4): Multifidus_L1-L4, MF(L2-L5): Multifidus_L2-L5, MF(L3-S): Multifidus_L3-S, MF(L4-S): Multifidus_L4-S, MF(L5-S): Multifidus_L5-S, LT: Longissimus Thoracis, IT: Iliocostalis thoracis, IL: Iliocostalis Lumborum, PM: Psoas major, Ili: Iliacus, HM: Hip muscle group
All calculations were performed to maintain the initial angles of all joints for 2 s under gravity, with gravitational acceleration applied to the field containing the human model in LS-DYNA. This study focuses on the motion of the head, neck, trunk, and pelvis. Postural changes in other parts had no or minimal effects on motions of the head, neck, trunk, and pelvis under these calculated conditions. Therefore, to reduce the computational time, the skull and skeletal parts from the shoulders to the fingertips in the human body FE model were set as non-deformable parts, and the bones from the femur to the foot and the muscles, tendons, and ligaments from the knee joint to the toes were constrained and maintained in a non-deformable and non-moving state (Supplementary Fig. S7). All the simulations in this study were performed using the explicit FE analysis solver LS-DYNA v971 R10.0.0.
When all muscles were inactive (Case A), the upright posture was not maintained (Fig. 3a), whereas when all muscles were active (Case B), the upright posture was maintained for 2 s (Fig. 3b). In Case A, an error occurred at approximately 1.8 s due to abdominal flexion causing a negative volume in the tendons of the internal oblique muscles, resulting in an element flipping error due to excessive deformation. This error occurred because excessive flexion of the abdomen caused slackening in the muscle portions of the rectus abdominis (RA), external oblique, and internal oblique muscles leading to generation of force in the direction of flexion of the portions of the tendon adjacent to the slackened muscles.Fig. 3Results of posture calculations (a) Case A (all muscles are inactive), (b) Case B (all muscles are active).
To verify the upright posture maintenance calculation, the computed muscle activity at 1.5 s, recovering from the postural change due to gravity in Case B was compared with the experimental results for four the RA, ES, rectus femoris (RF), and biceps femoris (BF) muscles^43^ (Fig. 4). The experimental results are presented as the average and standard deviation. The spinal erector muscle measured in the experimental tests was the LT^44^, the computed activity of the LT was compared to that of the spinal erector muscles, based on experimental data. The computed results almost fell within the range of experimental variability, indicating that the computed results were generally valid.Fig. 4Comparison between experimental results^43^ and computed results for muscle activity during upright posture. RA: rectus abdominis muscle, ES: erector spinae muscle, RF: rectus femoris muscle, BF: biceps femoris muscle. %MVC: the percent maximum voluntary contraction.
Figure 5 shows the average muscle activation levels extracted at 0.01-s intervals from 0 to 2 s for Cases B through N, whereas Supplementary Fig. S10 presents the activation for Cases O through Z. Red lines and symbols represent cases where muscle activation levels increased by more than 5%MVC compared to Case B, whereas blue lines and symbols represent cases where muscle activation levels decreased by more than 5%MVC from Case B. The activity of the neck extensors and hip flexors in Cases C through N was almost identical to that in Case B. In contrast, the activity in the trunk extensors was significantly higher in Cases I and J compared to Case B, whereas it was lower in Case L than in Case B. The activity of the hip extensors was higher in Case N and slightly lower in Case L than in Case B. In the absense of compensation, the activity of the hip flexors was higher in Case X than in Case B, whereas the activity of the hip extensors was higher in Case Z and lower in Case X than in Case B (Fig. S10).Fig. 5Calculated results of muscle activation levels for Case B and compensation cases (Cases C~N). Red lines and symbols represent cases where muscle activation levels increased by more than 5%MVC from Case B, whereas blue lines and symbols represent cases where muscle activation levels decreased by more than 5%MVC from Case B.
In Cases I and J, the activity of trunk extensors significantly increased. This suggest that when the trunk extensors, either MF(L1-S) or LT muscles was impaired, compensatory activity significantly increased in the trunk extensors. In particular, when the MF(L1-S) was not active, the increase in activity was most significant in trunk extensor muscles, increasing the activity of the LT and IL muscles to approximately 50%MVC, which is equivalent to the intensity of a single-leg full squat^37^. This result highlights the importance of MF(L1-S) and LT muscles in maintaining an upright posture. Meanwhile, in Case L, that is, when the hip flexors was impaired, the activities of trunk and hip extensors decreased. Furthermore, in Case X, the activity of hip extensor muscles significantly decreased, and the activity of the hip flexor muscles was markedly increased. This suggest that the hip flexors, namely, the psoas major (PM) and iliacus (Ili) muscles, are not active, it not only promotes compensatory activity in the hip flexor muscles but also reduces the activity of the trunk and hip extensors. This result suggests the importance of the PM and Ili muscles in maintaining an upright posture.
For the quantitative evaluation of posture, the alignment parameters used in spinal biomechanics were compared as chin-brow vertical angle (CBVA) and C2–C7 angle as indicators of horizontal gaze disorder in patients with spinal lordosis and sagittal vertical axis (SVA) (C2–C7) and SVA (C7-S) to assess the anterior/posterior displacement of the center of gravity in head and chest, thoracic kyphosis (TK) angle (T1–T5), and TK angle (T5–T12) to evaluate the severity of TK, and T1 slopes, which correlated with cervical lordosis and TK^45^. The sacral slope was also compared with an indicator of pelvic anteversion or retroversion. These spinal alignment parameters can be measured using spinal radiographs and are used to assess the severity of conditions, such as DHS and spinal lordosis.
Figure 6 illustrates the range of the minimum to maximum values for each parameter in simulations from 0 to 2 s across all cases. Although the range of changes in each parameter for Case B was the narrowest, consistent postural changes were observed. This is attributed to the sudden increase in gravity, which is not typically encountered in daily life, as the calculations began from zero gravity and a gravitational load was then applied. Figure 6 also includes the range of the minimum and maximum values for each alignment parameter obtained from experimental data of patients with DHS, as reported in previous studies^5,8,45–49^, shown in blue. The mean and standard deviation (SD) of each alignment parameter were plotted from these experimental data and the minimum and maximum values were extracted from the plotted data. Red rectangles indicate alignment parameters for which the intact Case B did not fall within the range of DHS patient data. This suggests that five alignment parameters—CBVA, C2–C7 angle, SVA (C2–C7), SVA (C7–S), and Sacral slope—are significant for the diagnosis of DHS. Supplementary Tables S1, S2 show the sagittal midsection of the spine for each case at 0.5-s intervals. Supplementary Fig. S8 presents the results for Case A as an example of the time history for each of the five alignment parameters.Fig. 6Range of minimum to maximum values of alignment parameters for each case.
In Fig. 6, the positive values of CBVA larger than that of Case B indicate a tendency for the face to tilt downward. In descending order they are as Cases A, N, I, and J with compensation while Cases U, Z, W, V, and Y without compensation. Conversely, the negative values of CBVA larger than in Case B indicate a tendency for the face to tilt upward. The order was as Cases X, O, R, and Q without compensation. The C2–C7 angle was negative in all cases, indicating maintenance of cervical lordosis. Among them, the cases with smaller negative values, indicating a tendency for a decrease in cervical lordosis, were in the order of Cases A and N with compensation while Cases Z, V, W, and U without compensation.
Conversely, the case with a larger negative C2–C7 angle, indicating an increase in cervical lordosis, was Case X. The positive values of SVA (C2–C7) larger than Case B indicate a tendency for forward inclination of the cervical spine in the order of Cases N, A, I, and J with compensation while Cases Z, V, U, W, and Y without compensation. Conversely, the negative values of SVA (C2–C7) indicate a tendency for backward inclination of the cervical spine in the following Cases X, O, Q, and R without compensation. The positive values of SVA (C7-S) larger than Case B indicate a tendency for forward inclination of the trunk in the following Cases A, N, I, and J with compensation while Z, V, W, U, and Y without compensation. Conversely, the negative values of SVA (C7-S) indicate a tendency for backward inclination of the trunk in the following Case L with compensation while Cases X, O, Q, R without compensation. The TK (T1–T5) was positive in all cases, indicating the maintenance of upper TK. Cases with smaller positive values than Case B, indicating a decrease in the degree of upper TK, were in the following Cases N and A with compensation while Z, V, U, W, and X without compensation. There were no cases in which the value was larger than Case B, indicating an increase in the degree of upper TK. The TK (T5–T12) was positive in all cases, indicating the maintenance of mid-TK. The cases with smaller positive values than Case B, indicating a decrease in the degree of mid-TK as Cases A, N, and I with compensation while X, W, and U without compensation. Conversely, cases with values larger than Case B indicate an increase in the degree of thoracic mid-kyphosis were Cases L with compensation, P and X without compensation; however, the changes were very small. The positive values of T1 slope larger than that of Case B indicate that T1 was more prone to forward tilting. The order of cases with T1 slopes larger than Case B in the positive values were as Cases A, N, I, and J with compensation while Z, V, W, U, and Y without compensation. The case with a negative T1 slope, indicating that the T1 was more prone to backward tilting (Case X). The sacral slope was increased, indicating that the pelvis is more prone to forward tilting. The order were as Cases A, N, I, and J with compensation, while Z, V, W, U, and Y without compensation. Conversely, the cases with sacral slopes smaller than in Case B, indicating that the pelvis was more prone to backward tilting, were Cases A, L, N, and I with compensation while X, Z, and U without compensation.
From the alignment results (Fig. 6), under the conditions with compensation, only Case N showed a significant forward tilt of the neck, trunk, and pelvis, whereas the other cases (Cases C-M) generally maintained the upright posture for the entire 2 s. The instability in the upright posture observed in Case N was attributed to the absence of muscles compensating for trunk extension. Meanwhile, under the condition without Cases U, V, W, and Z showed a significant forward tilt of the neck, trunk, and pelvis and were unable to maintain an upright posture. Furthermore, Case X showed a significant backward tilt of the neck, trunk, and pelvis, with the inability to maintain an upright posture. These results indicate that MF(L1-S), LT, IL, PM, and Ili muscles are important muscles for maintaining the upright posture.
Figure 7 compares the spinal midline cross-sections at 0 and 0.18 s, as well as the comparison between 0.18 and 0.4 s, for Case A (no muscle activity). When comparing the posture at 0 s (gray) and 0.18 s (red), the pelvis was significantly tilted backward (Supplementary Fig. S8), accompanied by a decrease in TK and posterior movement of the ribcage (Fig. 7a). When comparing the posture at 0.18 s (red) and 0.4 s (light blue), the ribcage further moved backward, and the face tilted downward between 0.18 s and 0.4 s (Fig. 7b). Considering that 0.18 s immediately follows the onset of gravity loading (Fig. 7a), the initial pelvic tilting, TK reduction, posterior movement of the ribcage, and downward tilting of the face can be attributed to the influence of gravity. Similar spinal forward inclinations were observed in Cases N, I, and J with compensation, whereas Cases Z, U, V, and W showed these inclinations without compensation. Similar pelvic backward tilting was seen in Cases L, N, and I with compensation, whereas Case X, Z, and U showed pelvic tilting without compensation. The sacral slope tends to increase in patients with DHS^47–50^. Our findings, which show backward pelvic tilting in Case A most similar to spinal posture of patients with DHS, align with experimental results on the sacral slope in patients with DHS.Fig. 7Change in sagittal planes at 0 s, 0.18 s, and 0.4 s for Case A.
In addition, we performed a simulation using a fat adult male model with a body mass index (BMI) of 30 kg/m^2^ to investigate the impact of increased upper body weight on spinal posture. The fat adult male model was developed by increasing the density of abdominal internal organs by 20 times, adding a fat weight of 26 kg to the abdominal internal organs of the original model. This modification was based on a previous report indicating that an adult male with a body weight of 90 kg has approximately 26 kg of fat weight^51^. We assumed that the original model had a BMI of 25 kg/m^2^, whereas the fat adult male model had a BMI of 30 kg/m^2^. Without muscle activity (Case A), SVA (C7-S) increased more rapidly in the fat model with a BMI of 30 kg/m^2^ than in the original model with a BMI of 25 kg/m^2^ (Supplementary Fig. S9a), indicating that the trunk leans forward due to the forward displacement of abdominal center of gravity in the BMI 30 kg/m^2^ model. With muscle activity (Case B), TK in the T1–T5 and T5–T12 regions was greater in the BMI 30 kg/m^2^ model than in the BMI 25 kg/m^2^ model between 0.1 and 0.5 s after gravity was applied at the simulation onset (Supplementary Fig. S9b, S9c). This suggests that the curvature of the spine increases in the BMI 30 kg/m^2^ model.
This study aimed to elucidate the mechanism underlying the occurrence of symptoms related to spinal alignment and, initially, to clarify the relationship between muscle activity and spinal behavior. Using the human body FE model and muscle controllers that individually predict the activity of various spinal erector muscles including the MF and LT muscles, the FE analysis was conducted to maintain an upright posture under gravity. First, we identify the muscles that are important for maintaining an upright posture. The results of muscle activity (Fig. 5) suggested that the MF (L1-S) and LT muscles are crucial for maintaining the upright postures. When the MF (L1-S) or the LT muscles was inactive, a pronounced forward tilt of the neck, trunk, and pelvis was observed (Cases I, J, and N in Fig. 6). The MF (L1-S) is an extensor muscle in the lumbar region, whereas the LT is an extensor muscle in the trunk, including the lumbar region. Therefore, a decrease in the strength of the lumbar extensor muscles may lead to a forward tilt of the trunk accompanied by a forward tilt of the neck and pelvis, suggesting difficulties in maintaining an upright posture. Furthermore, the results of muscle activity (Fig. 5) suggest that the PM and Ili muscles are also important for maintaining upright posture. When the PM and Ili muscles were inactive, a pronounced backward tilt of the neck, trunk, and pelvis was observed (Case X in Fig. 6). The PM and Ili muscles are hip flexors. Therefore, in this condition, where the area below the femur is fixed, a decrease in the strength of the hip flexor muscles may lead to easier hip extension, causing a posterior tilt of the pelvis and subsequent posterior tilt of the neck, trunk, and pelvis. The results of both muscle activity and alignment analysis suggest that the MF (L1-S), LT, PM, and Ili muscles are crucial for maintaining an upright posture.
Second, we discuss the compensatory mechanism of muscles for spinal movements. Stutzig and Siebert reported that, following fatigue, single-twitch torque decreased by approximately 20% when a single muscle in triceps surae was fatigued. However, the fatigue of the single muscle can be compensated for by increasing activation levels of the synergistic muscles resulting in unchanged MVC torque^52^. They suggest that the synergistic muscles may be controlled by descending command signals at the supraspinal and spinal levels. In our simulation results with compensation, muscle activation levels of the MF from T1 to T12, as well as those of the LT, IT, and IL, were significantly increased in Case I (without MF from L1 to S), whereas the activation levels of all other trunk extensors, except for LT, were significantly increased in Case J (without LT) (see Fig. 5). As a result, the CBVA were maintained similarly to Case B (Fig. 6). In contrast, Case N (without trunk extensors), the muscle activation levels of the remaining muscles did not increase. This suggests that trunk extensors have the ability to compensate via spinal reflexes, similarly to the triceps surae. CBVA, C2–C7 angle, and SVA (C2–C7) of Cases N and Z (with activity of neck extensors but without trunk extensor activity) were similar to those of Case A (Fig. 6). The neck extensors did not compensate for the activity of the trunk extensors in Case N (Fig. 5). This finding suggests that the neck extensors do not contribute to DHS, replicating previous findings that horizontal gaze disorders observed in patinents with DHS are not associated with cervical muscle strength^46^.
Third, we discuss the relationship between the spinal alignment and muscle activity. Posture of the patients with lumbar vertebrae lordosis is characterized by a large SVA (C2–C7), SVA (C7-S), TK (T1–T5), and TK (T5–T12) and a small sacral slope^53^. In our computation results, TK (T1–T5) and TK (T5–T12) were not significantly large; therefore, the posture resembling lumbar lordosis was not reproduced. However, the result indicated that the sacral slope decreased when the PM and Ili muscles were not active, suggesting the potential for pelvic retroversion due to weakening of the hip flexor muscles such as the PM and Ili muscles. Notably, the thoracic spine showed only a minimal change of approximately 1° at the kyphosis direction. The thoracic region is relatively rigid and stable because of the presence of the ribs and thoracic vertebrae, resulting in minimal postural changes. However, postural changes can become more significant with age. Meanwhile, experimental data on the material properties of the intervertebral discs and ligaments in the thoracic spine are limited. Therefore, further studies are required to investigate the material properties, particularly with regard to aging. The posture of patients with DHS is classified into cervical and thoracic types^4^. For both types, CBVA and SVA (C2–C7) and T1 slope increase. The cervical type has a smaller C2–C7 angle (larger kyphosis) and a negative SVA (C7-S) value. The thoracic type had a larger C2–C7 angle (from mild kyphosis to lordosis) and a larger positive SVA (C7-S) value. In our computation results for cases with increased CBVA, SVA (C2–C7), and T1 slope included Cases A, I, J, and N with compensation, while U, V, W, Y, and Z without compensation. Although all C2–C7 angle indicated cervical lordosis, cervical kyphosis equivalent to DHS was not observed in these calculated results. Therefore, cervical kyphosis equivalent to DHS was not predicted in the human body FE model used in this study because the cervical spine did not flex forward relative to the trunk and cervical spine was inclined forward together with the trunk (Supplementary Fig. S8). This discrepancy is because of the following limitations of the model; First, the tendon model is stiffer than real soft tissues due to the lack of a material model that describes strain rate dependency. Additionally, the tendons of the trapezius, which are extensively attached to the cervical spine, were partially modeled using solid elements. This resulted in a lower CBVA than that observed in patients with DHS. Further studies are needed to model the tendons without using solid elements. Second, the model did not account for muscle weakness or fractured vertebrae, which are common in older patients. To improve the accuracy of the model, further analyses are necessary to incorporate the PCSA of each muscle for older individuals and to represent bone fractures, allowing for more accurate predictions of TK (T5–T12) in patients with DHS.
Fourth, the mass of the upper body, especially the internal organs inside the body cavity, is not linked to all parts of the spine differently from the other human body FE models^54,55^. However, the internal organs are in contact with the spine through the serous membrane including the serous fluid. Therefore, when standing, the mass of the internal organs is supported by the pelvis, which is in the direction of gravity, rather than the adjacent spine. Our model has such weight-bearing structure. Although we investigated the impact of different upper body masses on the posture stability (see Fig.S9), further studies are needed to investigate the impact of the connection between the internal organ and the spine on the curvature of the spine when standing.
The model could not replicate all the characteristic postures of patients with kyphosis or DHS. Nevertheless, the crucial muscles, MF (L1–S1), LT, the PM, and Ili muscles, for maintaining an upright posture were identified. Weakening of these muscles leads to postures similar to those of patients with DHS, suggesting that these muscles are crucial targets for rehabilitation^5,49^.
We used a human body FE model that included the ES muscle group, along with a muscle controller and fixed legs for spinal posture stabilization, to investigate effect of individual spinal muscle activities on upright posture. A comparison of muscle activation states and postural changes revealed that the MF (L1-S) and LT muscles are crucial for suppressing forward tilting of the neck, trunk, and pelvis during the maintenance of an upright posture. Additionally, the PM and Ili muscles were identified as key muscles in inhibiting backward tilting of the neck, trunk, and pelvis. Furthermore, weakening of the hip flexors may contribute to the pelvic posterior tilting observed in kyphosis. This model has demonstrated its utility in the study of postural maintenance strategies. In the rehabilitation of DHS, exercises targeting cervical extensors, as well as posture improvement exercises for the thoracolumbar region and localized rehabilitation of muscles like the PM, are essential^5^. Postural improvement exercises for the thoracolumbar region specifically focus on strengthening the MF (L1-S) and LT muscles. Therefore, our study supports the notion that the MF (L1-S), LT, PM, and Ili muscles should be effectively stimulated through rehabilitation for DHS. Although this model has proven useful in identifying the individual muscles necessary for upright posture stabilization, further investigations are needed to model tendons and account for muscle weakness and fractured vertebrae in older patients. These efforts are crucial to elucidating the mechanisms underlying conditions such as kyphosis and DHS.
Supplementary Information.