SPINOPELVIC VARIABILITY ACCORDING TO THE ROUSSOULY CLASSIFICATON CURVE TYPE

ABSTRACT Objective To evaluate the variability of spinopelvic sagittal parameters and the distribution of lordosis in the lumbar spine in a sample of patients. Methods This is a cross-sectional study considering full-spine radiographs of a patient sample. The patients were classified according to the Roussouly classification and both radiographic spinopelvic alignment parameters and the lordosis measurement of each lumbar spinal segment were considered. The radiographic parameters were correlated with the Roussouly classification type. Results Ninety patients were included in the study. There was significant correlation between pelvic incidence (PI) and lumbar lordosis (LL) (R=0.89; p<0.0001). The values of PI were significantly higher in Roussouly types 3 and 4 than in types 1 and 2 (p<0.001), as were the values of LL L1-S1(p<0.001). Considering the total sample, 67% of LL L1-S1 was located between L4-S1, but with variations by the Roussouly classification curve types. Conclusion This study demonstrated a high correlation between the values of PI and LL, as well as the importance of the distal lumbar segment (L4-S1) in the overall value of LL L1-S1, which was even higher in patients with a lower PI value (Roussouly types 1 and 2). Level of evidence II; Retrospective analysis of a prospective database (Cohort); Diagnostic study.


INTRODUCTION
The study of the sagittal balance of the sagittal spine has become popular in recent decades with prolific evidence showing the correlation between the loss of spinopelvic sagittal alignment and the deterioration of function and the quality of life. [1][2][3][4] Thus, numerous radiographic parameters have been described for an understanding of spinopelvic sagittal alignment, as well as for the recognition of loss of alignment. 1,2,5,6 Among these radiographic parameters, pelvic shape and position parameters stand out, 7 as well as lumbar lordosis (LL), measured between L1 and S1, and the relationship between LL and pelvic incidence (PI). 5 However, some authors question the concept of fixed anatomical intervals for different spinal curvatures, including thoracic kyphosis measured between T4 and T12 and lumbar lordosis between T1 and S1. Berthonaud et al., for example, presented the concept of "inflection point" as a functional variable that corresponds to the point of transition between lumbar lordosis and thoracic kyphosis, regardless of the anatomical level where this occurs, and the concept of the variability of the extension of the curvatures of the spine. 8 Roussouly et al. presented a classification system for normal variants of sagittal alignment of the spine, taking the form and inclination of the pelvis and the distribution of lordosis throughout the lumbar segment into account, considering the "inflection point". 9 Four types of curves were described 9 and recently a fifth type was included. 10 The objective of this study was to evaluate the variability of spinopelvic sagittal alignment parameters in a sample of patients, as well as the distribution of lordotic curvature in the lumbar segment of the spine.

METHODS
This is a prospective, cross-sectional analysis of a radiographic database of patients treated at an outpatient spine pathology clinic of a single service. The study was approved by the Institutional Review Board of the service responsible for the study (CAAE: 97266618.0.0000.5463), which waived collection of the signed Informed Consent Form as the study considered only data already established in a radiographic database.
Full spinal radiography examinations (panoramic X-rays of the spine) of individuals older than 18 years of age were included, which allowed the evaluation and measurement of the spinopelvic parameters of interest. Examinations of patients with a previous history of neurological or spinal surgery, neurological or neuromuscular disease, trauma, or neoplastic disease of the spine and complaints of hip, knee, foot, or ankle disability that might alter the position of the joints were excluded.

Data collection
All radiographic examinations were performed at the same service following a standardized technique, with patients standing, comfortable, with the elbow in full flexion and the shoulder in 45° flexion with the hands relaxed and the fingers resting on the clavicle or the malar bone. 11,12 The radiographic parameters of interest were measured and analyzed using Surgimap Spine software (Nemaris Inc. New York, USA), validated for the measurement of radiographic spinal parameters. 13 The lordosis of each segment was measured (L1-L2, L2-L3, L3-L4, L4-L5, and L5-S1), as well as the LL between L1 and S1. (Figure 1) We also considered the parameters PI, pelvic tilt (PT), sacral slope (SS), and the discrepancy between PI and LL (PI-LL). The demographic data and medical records of the patients were also considered. The patients were classified according to the sagittal alignment characteristics using the system proposed by Roussouly et al. 9 Statistical analysis was performed using R software, version 3.4.9 (R Foundation for Statistical Computing, Vienna, Austria). The data consisted of quantitative variables and the normality of distribution was verified by the Shapiro-Wilk test. Radiographic parameters were compared among the different Roussouly classification types using the ANOVA test. The significance level considered was 5%.

RESULTS
Ninety patients met the inclusion criteria and were considered in the study. (Table 1
Considering the Roussouly type 1 patients, lordosis of the distal segment of the lumbar spine (L4-S1) corresponded to 76.3% of the LL L1-S1 with the inflection point located at L2-L3. In the Roussouly type 2 patients, the lordosis between L4-S1 corresponded to 62% of the LL L1-S1 with the inflection point located at L1-L2. In the Roussouly type 3 patients, the lordosis between L4-S1 corresponded to 70.8% with the inflection point located at T12-L1. In the Roussouly type 4 patients, the lordosis between L4-S1 corresponded to 60.9% with the inflection point located at T12-L1.

DISCUSSION
It is well established that LL is dependent on the PI value and an increase in the PI value is directly proportional to the absolute value of the LL measured between L1-S1. 14 However, in addition to the absolute value, a variation in the PI value also influences the behavior of the lumbar curvature, especially the extent of lordosis distribution. This concept was observed by Roussouly and was the basis for the classification system for the normal variants of sagittal spinal alignment presented. 9 This study evaluated a sample of patients for spinopelvic sagittal alignment and classification according to the system proposed by Roussouly.
Laouissat et al. 10 investigated the accuracy of the Roussouly classification, in a study with 296 individuals. Comparing their results with the results of our study, type 1 was the least frequent in both (12% vs. 15.6%, respectively), while type 3 was the most common in both (46% vs. 42.4%). In the article cited, type 2 accounted for 22% while in our study type 2 accounted for 20%. Type 4 made up 20% of the population in the Laouissat et al. study and 22.2% in our study.
Regarding the relationship between the radiographic parameters and the Roussouly classification, in this study we observed that the PI value was higher for types 3 and 4 than for types 1 and 2, with the    Roussouly type 4 value being higher than that of type 3. This result agrees with what was published in the original article by Roussouly et al., 9 as well as with the Laouissat et al. article. 10 Likewise, the LL L5-S1 value was higher in types 3 and 4, being higher in type 4 than in type 3, as compared to types 1 and 2, just as observed in the cited articles. 9,10 The literature has shown that 2/3 of the total LL L1-S1 is located in the distal segment of the lumbar spine (L4-S1). 9,15 Data from the present study showed that, in the total sample, segment L4-S1 was responsible for 67.2% of the LL L1-S1 value. Considering the different Roussouly classification types, the distribution of lordosis varied by lumbar segment: Type 1, with patients with low PI (mean 40°) and short lordosis, usually comprising three vertebrae, 76% of the LL L1-S1 being located between L4-S1; Type 2, also with low PI (46°) and a less pronounced lumbar curve (LL L1-S1 mean 48°), 62% of the LL L1-S1 located between L4-S1; Type 3, with a mean PI of 54°, with an increase in the number of vertebrae comprising the lordosis and 70% of the LL L1-S1 located between L4-S1; and Type 4, with a higher PI (mean 67°), a more pronounced and extensive LL (mean 70°), with 60% of the LL L1-S1 located between L4-S1.
A recent study evaluated spinopelvic sagittal alignment considering a sample of 268 individuals. 16 It was also observed that 67% of the lumbar lordosis was located between L4 and S1, similar to that observed in our study and the other published articles. Moreover, the correlation between LL and PI was analyzed and an arithmetic expression was obtained from the linear regression model in which LL L1-S1 = 0.54xPI + 27.6 (R = 0.56). Thus, they demonstrated that in patients with lower PI (PI < 50°) the value of LL is expected to be higher than the PI value, while in patients with higher PI it is expected that the LL value will be lower than that of the PI, respecting the concept presented in the article by Schwab et al. that LL = PI ± 9°. 14 In our study, we also observed a strong correlation between PI and LL, and by the linear regression model the value of LL L1-S1 = 0.83xPI + 13, with R=0.89 and R 2 =0.80 (p < 0.001).

CONCLUSIONS
This study presented an analysis of spinopelvic sagittal alignment in a sample of Brazilian patients and calculated the frequency distribution rate according to the Roussouly classification. The high correlation between PI and LL values was confirmed. We observed the important influence of the distal lumbar segment (L4-S1) in the overall values of LL L1-S1, even more significant in patients with lower PI (Roussouly types 1 and 2).