|Year : 2021 | Volume
| Issue : 2 | Page : 188-197
Cervical and thoracolumbar radiological sagittal parameters in asymptomatic Indian population
Yogesh Kishorkant Pithwa1, Sanjeev Sankar Chandran1, Vishnu Vardhan Rudravaram2
1 HOSMAT Hospital, Sattvik Spine and Scoliosis Center, Bengaluru, Karnataka, India
2 Department of Statistics, Pondicherry University (A Central), Puducherry, India
|Date of Submission||13-Dec-2020|
|Date of Decision||14-Mar-2021|
|Date of Acceptance||22-May-2021|
|Date of Web Publication||16-Jul-2021|
Yogesh Kishorkant Pithwa
Flat No. D-209, USHAS Apartments, Site No. 26, 16th Main, Jayanagar 4th Block, Bengaluru 560011, Karnataka
Source of Support: None, Conflict of Interest: None
Introduction: There is a dearth of normative data for radiological sagittal parameters of asymptomatic Indians. The present study aimed to address this lacuna. Materials and Methods: Sagittal radiological parameters were studied in asymptomatic volunteers: seven lumbopelvic, i.e., pelvic index (PI), pelvic tilt (PT), sacral slope (SS), lumbar lordosis (LL), cranial LL (crLL), caudal LL (caLL), PI-LL; three thoracic and thoracolumbar, i.e., thoracolumbar alignment (TL), thoracic kyphosis (TK), T1 slope (TS0); five cervical, i.e., cervical sagittal vertical axis (cSVA), cervical lordosis (CL), TS-CL, C2 slope (CS), C2 T1 pelvic angle (CTPA); and lastly, five global parameters: SVA, T9 spinopelvic inclination (T9SPI), T1 spinopelvic inclination (T1SPI), T1 pelvic angle (TPA), C2 pelvic angle (CPA) were studied. Results: Volunteers (n = 125) aged 41.49±12.93 years were included. Mean PI, PT, SS, LL, crLL, caLL, PI-LL, TL, TK, TS, cSVA, CL, TS-CL, CS, CTPA, SVA, T9SPI, T1SPI, TPA, and CPA were 47.23±8.04°, 13.4±6.61°, 33.68±4.59°, −56.19±7.83°, −22.71±9.82°, −45.02±10.07°, −7.83±9.4°, 9.99±11.17°, 22±7.33°, 4.88±7.64°, 34.77±12.2 mm, −6.36±10.99°, −0.5±10.02°, 2.56±9.53°, 3.54±1.04°, −36.49±23.4 mm, −10.89±2.75°, −7.88±2.17°, 5.52±6.82°, and 10.72±6.69°, respectively. As per Roussouly’s classification, the distribution for types I, II, III, and IV was 32 (25.6%), 41 (32.8%), 45 (36%), and 7 (5.6%), respectively. LL correlated significantly with PI, SS, and TK. TS had significant correlation with CL and TS-CL. cSVA significantly correlated with CL. PI-LL significantly correlated with TS-CL. CS significantly correlated with cSVA and TS-CL. TS-CL significantly correlated with cSVA. TPA correlated significantly with PT, SVA, PI, and PI-LL. CTPA correlated significantly with CL, cSVA, TK, and TS-CL. Males had significantly different SVA (−35.3 mm), TK (22.4°), TS (6.2°), TPA (6.35°), cSVA (37.1 mm), and CTPA (3.95°) when compared with females (−58.4 mm, 17.2°, −0.15°, 0.3°, 24.95 mm, and 2.85°, respectively) (P = 0.008, 0.003, 0.002, 0.003, 0.002, and 0.0005, respectively). Conclusion: Normative data for sagittal profile in Indian volunteers, enunciated in this study, can be used to guide decisions in surgery.
Keywords: Asymptomatic, cervical, Indian, lumbar, sagittal, sagittal radiological, sagittal radiological parameters, spine, thoracolumbar
|How to cite this article:|
Pithwa YK, Chandran SS, Rudravaram VV. Cervical and thoracolumbar radiological sagittal parameters in asymptomatic Indian population. Indian Spine J 2021;4:188-97
|How to cite this URL:|
Pithwa YK, Chandran SS, Rudravaram VV. Cervical and thoracolumbar radiological sagittal parameters in asymptomatic Indian population. Indian Spine J [serial online] 2021 [cited 2021 Dec 4];4:188-97. Available from: https://www.isjonline.com/text.asp?2021/4/2/188/321587
| Introduction|| |
Nature has shaped the normal spine in a manner to allow for flexibility and, at the same time, to allow for minimum expenditure of energy in static postures. Despite this ubiquitous principle, there exist lots of variations in normal spinal curvatures. To further elucidate normal variations, Roussouly et al. classified thoracolumbar sagittal profiles of asymptomatic volunteers into four different types. Type I profile had sacral slope (SS) less than 35° with the apex of lumbar lordosis (LL) at L5. Type II profile too had SS less than 35°, but with the apex of LL at L4. Type III profile had SS between 35° and 45°, whereas type IV profile had SS greater than 45°.
In addition to these normal variations, it is reasonably certain that there exists a lot of racial variations too in these parameters. Mac-Thiong et al. conducted a study on 709 asymptomatic European Caucasian subjects and noted a mean pelvic incidence (PI) of 52.6°. However, in a study in Korean population, Lee et al. noted a mean PI of 47.8°, which is much lower than that in the Caucasians. Needless to say, surgical goals for these parameters would vary based on the normative data for that population.
Though there has been some literature to define some normative parameters in the Indian population, a comprehensive database including all the current radiological parameters is missing.,, The present study aims to address this lacuna.
| Materials and Methods|| |
This was a prospective study carried out after approval for the same from the Institutional Review Board. Asymptomatic adult volunteers consenting for the study were included. Patients with history of spinal surgery, asymptomatic spondylolisthesis or spinal deformities, transitional vertebrae, patients with unclear radiographs in which radiological landmarks were not noticeably clear, those with flexion deformities of hips and knees, and those with unsteadiness in stance were excluded.
Full-length standing radiographs were taken uniformly at a distance of 6 feet, with the right side of the patient approximated to the radiographic cassette, hips and knees extended, feet approximated, straightforward horizontal gaze, and with fists on opposite clavicles. Radiographs were taken on two separate 14″ × 17″ cassettes and “stitched” together using “scanogram” technique using a digital “stitching” software (Fujifilm Medical Systems, USA). All radiographs were supervised by a single author.
Twenty different sagittal radiological parameters including pelvic index (PI), pelvic tilt (PT), SS, LL, cranial lumbar lordosis (crLL) (from upper endplate of L1 to lower endplate of L4), caudal lumbar lordosis (caLL) (from upper endplate of L4 to upper endplate of S1), PI-LL, sagittal vertical axis (SVA), T9 spinopelvic inclination (T9SPI) (angle between a vertical line drawn from the center of femoral heads and another line connecting the center of femoral heads with center of T9 vertebral body) [Figure 1], thoracolumbar alignment (TL) (angle between cranial endplate of T10 with caudal endplate of L2), thoracic kyphosis (TK) (angle between cranial endplate of T4 with caudal endplate of T12), T1 slope (TS) (angle between horizontal and cranial endplate of T1), T1 spinopelvic inclination (T1SPI) (angle between a vertical line drawn from the center of femoral heads and another line connecting center of T1 vertebral body with the center of femoral heads) [Figure 2], T1 pelvic angle (TPA) (angle between line connecting center of T1 vertebral body with center of femoral heads and another line connecting center of femoral heads with center of sacral endplate) [Figure 3], cervical sagittal vertical axis (cSVA) (horizontal offset of a vertical plumb-line from center of C2 vertebral body to posterosuperior corner of C7), cervical lordosis (CL) (angle between caudal endplate of C2 with caudal endplate of C7), TS-CL, C2 slope (CS) (angle between horizontal and caudal endplate of C2), C2 pelvic angle (CPA) (angle between line connecting center of C2 vertebral body with center of femoral heads and another line connecting center of femoral heads with center of sacral endplate) [Figure 4], and C2 T1 pelvic angle (CTPA) (angle between a line connecting center of C2 vertebral body with center of femoral heads and another line connecting center of femoral heads with center of T1 vertebral body) [Figure 5] were studied. These were sub-classified into seven lumbopelvic, three thoracolumbar and thoracic, five cervical, and lastly five global parameters [Table 1]. Measurements were made using Surgimap version 2.2.15. All measurements were initially made by one author and were later verified by the senior author.
|Figure 1: T9SPI measured as an angle between vertical line drawn from the center of femoral heads and another line connecting the center of femoral heads with center of T9 vertebral body|
Click here to view
|Figure 2: T1SPI measured as an angle between vertical line drawn from the center of femoral heads and another line connecting center of T1 vertebral body with the center of femoral heads|
Click here to view
|Figure 3: TPA measured as an angle between line connecting center of T1 vertebral body with center of femoral heads and another line connecting center of femoral heads with center of sacral endplate|
Click here to view
|Figure 4: CPA measured as an angle between a line connecting center of C2 vertebral body with center of femoral heads and another line connecting center of femoral heads with center of sacral endplate|
Click here to view
|Figure 5: CTPA measured as an angle between a line connecting center of C2 vertebral body with center of femoral heads and another line connecting center of femoral heads with center of T1 vertebral body|
Click here to view
All the subjects were also classified based on Roussouly’s classification of thoracolumbar sagittal alignment.
Statistical analysis was done using statistical software “R,” version 3.4.1. Data following Gaussian distribution were presented as mean and standard deviation, whereas non-parametric data were presented as median with range. Statistical significance was considered if the two-tailed P-value was less than 0.05. Based on normality of data, appropriate parametric (unpaired “t” test) or non-parametric tests (Mann–Whitney) were conducted. Correlations were studied using Spearman’s rank correlation test. To overcome the problem of imbalanced data, statistical technique of SMOTE (synthetic minority oversampling technique) was used. Statistical analysis was carried out to detect ideal sample size considering a power of 0.8.
| Results|| |
One hundred and twenty-five volunteers with mean age of 41.49±12.93 years were included. There were 109 males and 16 females.
Mean values of the various radiological parameters were as elaborated in [Table 1].
As per Roussouly’s classification of the entire study population, the incidences of types I, II, III, and IV were 32 (25.6%), 41 (32.8%), 45 (36%), and 7 (5.6%), respectively.
[Table 2] lists all the correlation results using Spearman’s rank test. Increase in LL correlated significantly with increase in PI, SS, and TK (P < 0.0001 for all the three correlations) with the maximum correlation being demonstrated with SS (r=0.7485). Increase in TK had a significant correlation with increase in TS (P < 0.0001). In turn, increase in TS correlated with a significant increase in CL (P < 0.0001). Being similar in nature, correlation between TS and SS was assessed. However, it just missed being significant (P = 0.0533). Increase in TPA showed a significant correlation with increase in SVA, PT, PI, as well as PI-LL (P < 0.0001 for all the correlations), with the maximum correlation being demonstrated with PT (r=0.76220). Increase in cSVA significantly correlated with decrease in CL (P < 0.0001). Being similar indices, PI-LL significantly correlated with TS-CL (P = 0.0283). Increase in CS showed a significant correlation with increase in cSVA as well as TS-CL (P < 0.0001 for both the correlations) with greater correlation with cSVA (r=0.731). Increase in cSVA showed a significant correlation with increase in TS-CL (P < 0.0001). Increase in CTPA correlated significantly with decrease in CL and increase in cSVA, TS-CL, and TK (P < 0.0001, <0.0001, 0.0043, and 0.0005, respectively) with the greatest correlation being demonstrated with cSVA (r=0.916). Numerous other correlations were studied, as listed in [Table 2]. However, the other correlations did not reveal any statistical significance.
Though comparable in most parameters, males had a more anteriorly placed SVA, higher TK, TS, TPA, cSVA, and CTPA when compared with females [Table 3]. However, males (n = 109) outnumbered females (n = 16) in this study population leading to imbalanced data. Statistical analysis revealed that the ideal sample size for each cohort should be 63, to prevent type II statistical error. To address this, statistical technique of SMOTE was adopted to maintain balance between the male and female categories. By using this technique, both oversampling and undersampling was done to get 169 males and 145 females using ROSE package in statistical software “R.” The resultant data showed a significant age difference between males (median: 44 years) and females (median: 50 years). Hence, class intervals of age were formulated using the statistical method of visual binning process with one standard deviation [Table 4]. Analysis of these age-matched cohorts revealed statistically significant difference between males and females in all the radiological parameters in at least one age group [Table 5].
Of 109 males, 74 were less than 50 years of age. Comparing all parameters between males less than 50 years of age with those more than or equal to 50 years, statistically significant difference was noted only in LL (lower in the elderly population) and in T1SPI [Table 6]. However, as mentioned earlier, this analysis was underpowered as there were less than 63 subjects in the elderly cohort. Of 16 females, six were less than 50 years of age. Comparing all parameters between females less than 50 years of age with those more than or equal to 50 years, statistically significant difference was noted only in T9SPI, cSVA, and CTPA [Table 7]. However, as mentioned earlier, this analysis was underpowered as there were less than 63 subjects in each of the cohorts.
| Discussion|| |
In an epidemiological study in the United States, Zygourakis et al. noted that the number of adult spinal deformity surgeries increased from 4.16 per 100,000 adults in the year 2001 to 13.9 per 100,000 adults in 2013. Since goals for adult spinal deformity surgeries are based on normative data for the population, it is essential to have a database for the same.
Roussoully et al. identified four different variations of the normal sagittal alignment based on differences in SS. This study demonstrated variations in these parameters in addition to the normal variations expected with age and gender. Their study on European Caucasian population demonstrated that the most common variant in asymptomatic population was type III, which was reflected in our study as well which had 36% (45 of 125 subjects). However, though the least common variant in their study was type II, we noted type IV as the least one with 5.6% incidence (7 of 125 subjects).
PI of the cohort in the present study (47.23 ± 8.04°) closely matches the findings of two Indian studies, that by Singh et al. (48.52 ± 8.99°) and Borkar et al. (49.29 ±5.95°) as well as Chinese Han (44.6 ± 11.2°) and Korean population (47.8 ± 9.3°).,,, However, PI of the Indian cohort studied by Sudhir et al. seems to be more akin to the Caucasian study., This phenomenon is likely related to the fact that the cohort studied by Sudhir et al. might have had a greater proportion of subjects with type IV Roussoully variants, whereas the present study, as mentioned earlier, had the least proportion of this variant. The other likely explanation is that even within India, there exists a lot of ethnic diversity. The combined sample size of the present study along with the other Indian studies by Singh et al. and Borkar et al. for asymptomatic individuals is greater (n = 197) when compared with the sample size (n = 101) of the study by Sudhir et al. Hence, the findings of the present study likely represent the larger section of the ethnically diverse Indian population.
The importance of pelvis as the foundation for spinal balance has been well enunciated in the literature. Pelvic parameters influence lumbar spinal curvature with its consequent effect on thoracic spinal curvature. This consequential relationship is highlighted by the correlations in the present study between LL with PI and SS, indicating that the latter two parameters influence how curved the lumbar spine is above the pelvis in order to maintain sagittal balance in a non-pathological situation. Consequential to this, TK too is correlated to LL as enunciated in the present study. As a continuation of this harmonious relationship, TK was correlated to TS in the present study. Going further cranially, TS was found to correlate with CL which in turn had a significant correlation with cSVA in the present study. Similar correlation has also been echoed by Goldschmidt et al., who have even worked out a trigonometric relationship among the three parameters of TS, CL, and cSVA. Inter-relation of TS with other cervical parameters was also echoed in the study by Zhu et al. In an analysis of evolution of sagittal radiological parameters over age, Attiah et al. noted that as age advances, TK increases with increase in TS and increase in CL. This natural history can be well supported by the correlations between these parameters enunciated in the present study.
Protopsaltis et al. from the International Spine Study Group proposed TPA to account for global sagittal deformity. They noted a close association of this angle with SVA and PT. The advantage with this measurement is that being angular, it can be carried out in non-calibrated radiographs as well. However, the most notable advantage of this measurement is the fact that it is independent of knee flexion, use of assistive devices for standing, and pelvic retroversion while at the same time maintaining good correlation with health-related quality of life outcomes in patients. The present study population had TPA of 5.52±6.82° and showed a significant correlation with PT, PI, PI-LL as well as SVA.
Akin to the relevance of PI-LL in lumbar spine surgery, TS-CL too has been noted to be an important parameter predictive of successful outcome after cervical spine surgery., In fact, the two parameters showed a statistically significant correlation with each other in the present asymptomatic group of patients. cSVA, another parameter linked to quality of life, also had a significant correlation with TS-CL in the present study.
CS has been suggested as a singular marker for cervical deformity, correlating with patient-reported outcomes. The present study demonstrated significant correlation between CS with cSVA and with TS-CL, the latter two being independently related to patient-reported outcomes. In 2017, the International Spine Study Group published one more parameter, CTPA, to define severity of cervical deformity and its correlation with patient-reported health outcomes. The present study demonstrated correlation of CTPA with CL, cSVA, TS-CL as well as TK. Similar correlations were also echoed by literature published by the International Spine Study Group., A logical extrapolation of these correlations indicates that only one of these parameters may be adopted in a study evaluating cervical spine surgery.
Numerous studies in the past have tried to identify gender differences in sagittal radiological parameters in the asymptomatic population.,, Various studies have noted higher LL in females when compared with males.,,, Other parameters demonstrating gender differences noted by these workers included PI and PT., In another Indian study, Singh et al. noted higher SS in females, which was replicated in the present study as well. In a large study involving 626 volunteers, Yukawa et al. studied gender- and age-related changes in asymptomatic subjects. They noted differences between males and females in at least one age group in CL, TK, LL, PI, and PT, though there was no definite age-related trend noted in these differences. The present study too showed significant difference between males and females in all the parameters in at least one age group, though there was no discernible age-related trend. The close inter-dependence between various radiological parameters probably explains the consequential difference noted between all the parameters between males and females identified in the present study.
The present study is the largest study of its kind in the Indian population with a detailed analysis of multiple parameters which were hitherto not studied in this population. All radiographs were taken as per a strict protocol with fists on opposite clavicles in a single institution supervised by a single person. In addition to routine radiological parameters, a lot of angular parameters were studied that obviated errors due to calibration as well as minor postural variations.,,, However, the present study is limited by the fact that India is a large country with ethnically diverse population. Hence, the findings of the present study may need validation with further studies. Also, differences noted between males and females were derived from SMOTE data, which may need validation with actual data in further studies. Similar differences in parameters based on age noted in the present study may need further validation with larger studies.
To conclude, the present study provides a guide to base surgical decisions in an era where the role of sagittal radiological parameters is ever increasing. Considering the variations inherent to ethnicity, these data can be fruitfully utilized by clinicians attending to Indian patients. Significant differences were noted in numerous parameters between males and females. In addition, the numerous correlations between various radiological parameters elucidated in the present study can allow the clinician to choose just a few of these parameters to guide their surgical aims. As mentioned earlier, a greater leaning toward angular measurements can obviate errors due to posture and calibration.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
Yogesh Kishorkant Pithwa—Conception and design, analysis of data, drafting of manuscript, critical revision, and supervision of data acquisition.
Sanjeev Sankar Chandran—Conception and design, data acquisition, and drafting of manuscript.
Vishnu Vardhan Rudravaram—Authorship role: Analysis of data, drafting of manuscript, and critical revision.
| References|| |
Roussouly P, Gollogly S, Berthonnaud E, Dimnet J. Classification of the normal variation in the sagittal alignment of the human lumbar spine and pelvis in the standing position. Spine (Phila Pa 1976)2005;30:346-53.
Lee CS, Chung SS, Kang KC, Park SJ, Shin SK . Normal patterns of sagittal alignment of the spine in young adults radiological analysis in a Korean population. Spine (Phila Pa 1976) 2011;36:E1648-54.
Mac-Thiong JM, Roussouly P, Berthonnaud E, Guigui P. Age- and sex-related variations in sagittal sacropelvic morphology and balance in asymptomatic adults. Eur Spine J 2011;20(Suppl 5): 572-7.
Singh R, Yadav SK, Sood S, Yadav RK, Rohilla R. Spino-pelvic radiological parameters in normal Indian population. Sicot J 2018;4:14.
Sudhir G, Acharya S, Kalra KL, Chahal R. Radiographic analysis of the sacropelvic parameters of the spine and their correlation in normal asymptomatic subjects. Global Spine J 2016;6:169-75.
Borkar SA, Sharma R, Mansoori N, Sinha S, Kale SS. Spinopelvic parameters in patients with lumbar degenerative disc disease, spondylolisthesis, and failed back syndrome: Comparison vis-à-vis normal asymptomatic population and treatment implications. J Craniovertebr Junction Spine 2019;10:167-71.
Rajnics P, Templier A, Skalli W, Lavaste F, Illes T. The importance of spinopelvic parameters in patients with lumbar disc lesions. Int Orthop 2002;26:104-8.
Fernández A, García S, Galar M, Prati RC, Krawczyk B, Herrera F. Learning from Imbalanced Data Sets. 1st ed. Cham:Springer Publications; 2018.
Zygourakis CC, Liu CY, Keefe M, Moriates C, Ratliff J, Dudley RA, et al
. Analysis of national rates, cost, and sources of cost variation in adult spinal deformity. Neurosurgery 2018;82:378-87.
Zhu Z, Xu L, Zhu F, Jiang L, Wang Z, Liu Z, et al
. Sagittal alignment of spine and pelvis in asymptomatic adults: Norms in Chinese populations. Spine (Phila Pa 1976) 2014;39:E1-6.
Legaye J, Duval-Beaupère G, Hecquet J, Marty C. Pelvic incidence: A fundamental pelvic parameter for three-dimensional regulation of spinal sagittal curves. Eur Spine J 1998;7:99-103.
Goldschmidt E, Angriman F, Agarwal N, Trevisan M, Zhou J, Chen K, et al
; International Spine Study Group (ISSG). A new piece of the puzzle to understand cervical sagittal alignment: Utilizing a novel angle δ to describe the relationship among T1 vertebral body slope, cervical lordosis, and cervical sagittal alignment. Neurosurgery 2020;86:446-51.
Zhu Y, An Z, Zhang Y, Wei H, Dong L. Predictive formula of cervical lordosis in asymptomatic young population. J Orthop Surg Res 2020;15:2.
Attiah M, Gaonkar B, Alkhalid Y, Villaroman D, Medina R, Ahn C,et al
. Natural history of the aging spine: A cross-sectional analysis of spinopelvic parameters in the asymptomatic population. J Neurosurg Spine2019:1-6.
Protopsaltis T, Schwab F, Bronsard N, Smith JS, Klineberg E, Mundis G, et al
; International Spine Study Group. The T1 pelvic angle, a novel radiographic measure of global sagittal deformity, accounts for both spinal inclination and pelvic tilt and correlates with health-related quality of life. J Bone Joint Surg Am 2014;96:1631-40.
Passias PG, Vasquez-Montes D, Poorman GW, Protopsaltis T, Horn SR, Bortz CA, et al
; ISSG. Predictive model for distal junctional kyphosis after cervical deformity surgery. Spine J 2018;18:2187-94.
Protopsaltis T, Terran J, Soroceanu A, Moses MJ, Bronsard N, Smith J, et al
; International Spine Study Group. T1 slope minus cervical lordosis (TS-CL), the cervical answer to PI-LL, defines cervical sagittal deformity in patients undergoing thoracolumbar osteotomy. Int J Spine Surg 2018;12:362-70.
Ling FP, Chevillotte T, Leglise A, Thompson W, Bouthors C, Le Huec JC. Which parameters are relevant in sagittal balance analysis of the cervical spine? A literature review. Eur Spine J 2018;27:8-15.
Protopsaltis TS, Ramchandran S, Tishelman JC, Smith JS, Neuman BJ, Mundis MM Jr et al
; International Spine Study Group. The importance of C2 slope, a singular marker of cervical deformity, correlates with patient-reported outcomes. Spine (Phila Pa 1976) 2020;45:184-92.
Protopsaltis T, Bronsard N, Soroceanu A, Henry JK, Lafage R, Smith J, et al
; International Spine Study Group. Cervical sagittal deformity develops after PJK in adult thoracolumbar deformity correction: Radiographic analysis utilizing a novel global sagittal angular parameter, the CTPA. Eur Spine J 2017;26:1111-20.
Vialle R, Levassor N, Rillardon L, Templier A, Skalli W, Guigui P. Radiographic analysis of the sagittal alignment and balance of the spine in asymptomatic subjects. J Bone Joint Surg Am 2005;87:260-7.
Yukawa Y, Kato F, Suda K, Yamagata M, Ueta T, Yoshida M. Normative data for parameters of sagittal spinal alignment in healthy subjects: An analysis of gender specific differences and changes with aging in 626 asymptomatic individuals. Eur Spine J 2018;27:426-32.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6], [Table 7]