Introduction

The cervical spine represents a mobile segment between the head and trunk. Its sagittal alignment involves cognitive, neurosensory and motor control to maintain horizontal gaze during standing, walking and daily activities. In clinical practice, radiographic parameters are used when analyzing cervical alignment, which represents a single momentum in static position. Radiographic alignment is considered when planning multiple-level subaxial or occipitocervical fusion [1, 2]. Cervical alignment is further considered in adolescent idiopathic scoliosis, where a relationship between thoracic hypokyphosis and decreased cervical lordosis exists [3,4,5,6]. In adult spinal deformity, cervical deformity might occur if the ability to maintain compensatory lordosis decreases. This results in anterior C2-plumbline migration and deterioration of health-related quality of life [7]. Although cervical and thoracolumbar relationships were described in deformed spine, there is a lack of normative values that could be used when treating spinal deformities.

Normal cervical and thoracic alignment is linked: Cervical lordosis correlates with T1-slope and thoracic kyphosis in children and adults [8,9,10,11]. The amount of cervical lordosis increases with age [12, 13]. In the thoracolumbar segment, an age-related thoracic kyphosis increase, lumbar lordosis decrease and spinopelvic adaptive phenomena were described [14, 15]. The physiologic amount of lumbar lordosis and thoracic kyphosis and the shape of thoracolumbar curvatures are related to the spinopelvic configuration [16].

It remains unclear how cervical alignment varies according to gender, age and spinopelvic alignment types. The amount of pelvic incidence influences thoracolumbar lordosis and kyphosis distribution, which might influence cervical alignment. We hypothesize that normal variations exist in individuals without spinal deformity.

The purpose of this cross-sectional study was to describe cervical parameters across age groups from the growth period to the elderly and to analyze differences by gender and pelvic incidence.

Materials and methods

Institutional review board approval (FC/2019-91) was obtained. Sagittal full spine radiographs (EOS Imaging, Paris, France) performed in 16 pediatric and adult spine centers were prospectively collected in a national registry from September 2019 to March 2020 and analyzed retrospectively. Clinical charts were screened, and radiographs of patients with limb length discrepancy less than 2 cm, spondylolysis, common low back or radicular leg pain were used. If several radiographs were available, the most recent image was selected. One radiograph was analyzed per patient. Exclusion criteria were: spinal deformity such as scoliosis, Scheuermann’s kyphosis or spondylolisthesis more than grade 1, spinal or pelvic fractures, tumors, infection and neuromuscular disorders, previous spine surgery other than microdiscectomy and severe degenerative changes or osteoporosis leading to thoracolumbar deformity. Intervertebral disc degeneration, facet joint osteoarthritis and degenerative spondylolisthesis without thoracolumbar malalignment did not represent exclusion criteria since these radiographic changes commonly occur during aging.

In each center, a trained operator reconstructed radiographic landmarks between the cranial auditory meatus and femoral heads using KEOPS software (SMAIO, Lyon, France) (Fig. 1). Five senior spine surgeons then checked each reconstruction to minimize inter-rater errors. This method is reliable and superior to manual measurements [17]. Spinopelvic parameters were assessed: pelvic incidence (PI), pelvic tilt (PT) and sacral slope (SS), thoracic kyphosis (TK) between cranial T1 and caudal T12 endplates and lumbar lordosis (LL) between cranial L1 and S1 endplates. TK and LL were expressed as positive values. Among global alignment parameters, sagittal vertical axis (SVA) of C7 and C2 measured the distances between vertebral plumblines and the posterior corner of S1 [18]. The odontoid-hip axis (OD-HA) determined the angle between the vertical axis originating from the midpoint between femoral heads and the axis joining the odontoid tip [19].

Fig. 1
figure 1

KEOPS software reconstruction of each vertebra from C1 to the sacrum, cranial auditory meatus, McGregor line and femoral heads on lateral full spine EOS radiograph

Full size image

Cervical parameters were divided into cranial (C0–C2) and caudal (C3–C7) measurements. Among cranial parameters, McGregor slope represented the angle between the horizontal axis and McGregor line: the tangent to the caudal occiput joining the hard palate. Positive values reflected head flexion and negative values extension. The occipito-C2 angle was measured between McGregor line and the posterior odontoid cortical margin. McGregor-C2 lordosis was determined between McGregor line and C2 endplate. C1–C2 lordosis was measured between the anterior–posterior atlas axis and C2 endplate (Fig. 2). Among caudal parameters, C2–C7 lordosis was measured between caudal C2 and C7 endplates. Cervical lordosis was then divided into a cranial arch between C2 and cervical apex (inflexion point in sigmoid alignment types) and into a caudal arch between cervical apex and cervicothoracic inflexion point. Segmental cervical lordosis was expressed as positive value and kyphosis as negative value. C7-slope and T1-slope were determined as angles between cranial endplates and the horizontal axis (Fig. 3). The cervical morphology was analyzed by counting the number of vertebrae included from C1 to cervicothoracic inflexion point. Cervical alignment patterns were described: global lordosis, global kyphosis, sigmoid shape with proximal kyphosis and distal lordosis, sigmoid shape with proximal lordosis and distal kyphosis (Fig. 4).

Fig. 2
figure 2

Cranial cervical parameters: McGregor slope (A), occipito-C2 angle (B), McGregor-C2 lordosis (C) and C1–C2 lordosis (D)

Full size image
Fig. 3
figure 3

Caudal cervical parameters: C2–C7 lordosis (A), cranial arch lordosis (B), caudal arch lordosis (C), C7- and T1-slope (D)

Full size image
Fig. 4
figure 4

Morphologic variations of cervical alignment patterns: global lordosis (A), global kyphosis (B), sigmoid shape with proximal kyphosis and distal lordosis (C), sigmoid shape with proximal lordosis and distal kyphosis (D)

Full size image

Statistical evaluation was performed with R software, version 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria). A Bayesian inference using Markov chain Monte Carlo techniques with low informative priors was used. To infer the mean of indices and correlation coefficients between two indices, point estimates corresponded to the median of posterior distributions and credibility intervals to their 2.5 and 97.5 percentiles. Cervical alignment parameters were compared by gender, age groups and PI. Ages were grouped by 5-year spans in children and adolescents, as growth-related changes might occur, and by 15-year spans in adults. The thresholds for PI were determined according to first and third quartiles: low PI < 45°, medium PI 45°–60° and high PI > 60°. Significance tests were based on the probability of superiority for the difference between 2 means and for correlation coefficients (strong if ρ < − 0.5 or ρ > 0.5). For PI the medium group and for ages the group 20–34 years were considered as references. The significance level was set at a 0.95 probability of superiority.

Results

A total of 2599 individuals, 1488 females and 1111 males, were included. The ages ranged from 5 to 93 years. The distribution by age was: 5–9 years n = 94, 10–14 years n = 516, 15–19 years n = 448, 20–34 years n = 224, 35–49 years n = 424, 50–64 years n = 449, 65–79 years n = 340 and > 80 years n = 104. The distribution by PI was: low PI n = 972, medium PI n = 1123 and high PI n = 504. Table 1 gives an overview of SVA-C7, SVA-C2, OD-HA, TK, LL, PI, PT and SS per gender, PI and age groups.

Table 1 Overview of distribution of global alignment and spinopelvic parameters per gender pelvic incidence and age groups: mean ± SD
Full size table

Cranial cervical parameters

Table 2 demonstrates the distribution of cranial cervical parameters by gender, PI and age. Variations of McGregor slope were small without significant differences between gender and age groups. Individuals with low PI had larger values (Pr = 0.9606). The occipito-C2 angle was similar in males and females. This parameter was significantly higher in low PI (Pr > 0.9999) and in adolescents from 10 to 19 years compared to young adults aged 20–34 years (Pr > 0.95). A progressive decrease was observed with age. McGregor-C2 lordosis was significantly smaller in males (Pr > 0.9999) and in low PI (Pr = 0.9732). A progressive increase with age was observed. Values were significantly smaller in individuals younger than 15 years (Pr > 0.95). C1–C2 lordosis was significantly smaller in males (Pr > 0.9999). This parameter was similar in PI groups. C1–C2 lordosis increased during growth with significantly smaller values in children younger than 10 years (Pr > 0.9999), and it then kept constant in adults (Fig. 5).

Table 2 Distribution of cranial cervical parameters (mean ± SD) and prediction of significant difference (Pr > 0.95) between median of posterior distributions and credibility intervals (95% CI)
Full size table
Fig. 5
figure 5

Distribution of cranial cervical parameters per age group. Red squares indicate median values

Full size image

Caudal cervical parameters

Table 3 displays caudal cervical parameters by gender, PI and age. Differences of C2–C7 lordosis were non-significant between males and females. Individuals with low PI had significantly smaller C2–C7 lordosis (Pr > 0.9999). A progressive increase was observed after 50 years. Cranial arch lordosis was significantly smaller in males (Pr = 0.9901). Differences between PI groups were non-significant. This parameter kept constant during adulthood. Caudal arch lordosis was significantly larger in males (Pr > 0.9999). Differences between PI groups were non-significant. This parameter increased progressively during adulthood. When compared to young adults, there was a significant difference in the age group between 35 and 49 years (Pr = 0.9693) and after 50 years (Pr > 0.9999). Similar observations were made for C7- and T1-slope. Both parameters were significantly larger in males (Pr > 0.9999). Differences between PI groups were non-significant. A significant increase with aging was observed for both parameters in the group 35–49 years (Pr > 0.95) and after 50 years (Pr > 0.9999) (Fig. 6).

Table 3 Distribution of caudal cervical parameters (mean ± SD) and prediction of significant difference (Pr > 0.95) between median of posterior distributions and credibility intervals (95% CI)
Full size table
Fig. 6
figure 6

Distribution of caudal cervical parameters per age group. Red squares indicate median values

Full size image

Cervical morphology

Cervicothoracic inflexion levels were at C6 in 526 (20.2%), C7 in 978 (37.6%), T1 in 768 (29.6%), T2 in 256 (9.9%) and T3 in 71 (2.7%) individuals. Variations in number of vertebrae from C1 to the cervicothoracic inflexion were small (Table 4). The cervical segment included more vertebrae in low PI (Pr > 0.9999) and in the group 10–14 years (Pr = 0.9813). The distribution by cervical alignment patterns (Fig. 4) was: global lordosis n = 1323 (50.9%), global kyphosis n = 34 (1.3%), sigmoid shape with proximal kyphosis and distal lordosis n = 894 (34.4%) and sigmoid shape with proximal lordosis and distal kyphosis n = 348 (13.4%).

Table 4 Distribution of number of cervical vertebrae from C1 to the cervicothoracic inflexion point (mean ± SD) and prediction of significant difference (Pr > 0.95) between median of posterior distributions and credibility intervals (95% CI)
Full size table

Correlation between cervical, global and spinopelvic alignment

Table 5 displays correlation coefficients ρ between cervical and global alignment parameters. SVA-C2 correlated significantly with caudal arch lordosis (ρ = 0.5526; Pr = 0.9997) and non-significantly with T1-slope (ρ = 0.5217; Pr = 0.9337) and C7-slope (ρ = 0.5001; Pr = 0.5026). No correlation was observed for SVA-C7 and OD-HA. Table 6 demonstrates correlation coefficients between cervical and spinopelvic parameters. TK correlated significantly with C2–C7 lordosis (ρ = 0.5359; Pr = 0.9943), caudal arch lordosis (ρ = 0.7267; Pr > 0.9999), C7-slope (ρ = 0.7073; Pr > 0.9999) and T1-slope (ρ = 0.8491; Pr > 0.9999). There was no correlation between cervical parameters and LL, PI, PT or SS.

Table 5 Correlation between cervical and global alignment parameters: correlation coefficients ρ strong if ρ < − 0.5 or ρ > 0.5 and prediction of significance only if Pr > 0.95 indicated below each coefficient
Full size table
Table 6 Correlation between cervical and spinopelvic parameters: correlation coefficients ρ strong if ρ < − 0.5 or ρ > 0.5, and prediction of significance only if Pr > 0.95 indicated below each coefficient
Full size table

Discussion

Analyzing cervical alignment in spinal deformity has gained interest [3,4,5,6,7]. Cervical malalignment represents a consequence of thoracolumbar deformity as interactions with the mobile cervical segment exist [11, 20]. Although age-related changes of cervical parameters have been reported on the one hand [11,12,13] and spinopelvic and global alignment parameters on the other hand [14, 15], there is a lack of knowledge about the interaction of both. PI- and age-related alignment targets have been defined in adult spinal deformity [21], seeking for optimal thoracolumbar alignment, thus limiting the risk of secondary cervical decompensation [7]. Therefore, the present study aimed to provide a comprehensive description of cranial and caudal cervical alignment according to gender, PI and age, to provide a repository of reference parameters when treating spinal deformity.

Cranial parameters are mainly influenced by head position, which is often analyzed by the chin-brow vertical angle. Moses et al. [22] demonstrated a correlation with McGregor slope. Le Huec et al. [9] reported a mean value of 1.59° (range − 18° to 16°) in adults. Abelin-Genevois et al. [8] reported a mean McGregor-C2 lordosis of 15.2° in children and 18.3° in adolescents, with larger values in girls. In asymptomatic adults, mean values range from 12.7° to 22.4° [9, 23,24,25,26]. Our results are in line with these findings. Gender-specific differences exist for McGregor-C2 lordosis, and a progressive increase with age was observed. McGregor slope and occipito-C2 angle decreased with age, describing a progressive head extension. Significant differences existed between pediatric and young adult populations, which might be explained by a lower muscle tone of cervicothoracic erector spinae and trapezius muscles in children and adolescents. The amount of PI influenced the occipitocervical orientation, especially in low PI, which is associated lower TK and LL. Positive McGregor slope values indicated a flexed head position, which resulted in larger occipito-C2 angles and smaller McGregor-C2 lordosis. C1–C2 lordosis accounts for the main part of cervical lordosis. Mean values range from 26.0° to 30.3° in children and adolescents [8] and 20.8°–31.9° in adults [9, 12, 24,25,26,27,28]. Our findings fit within these ranges. C1–C2 lordosis increased during growth, whereas values kept stable during adulthood. Gender-specific differences existed with smaller values in males.

The literature indicates a large variability of C2–C7 lordosis, due to varying measurement methods, different alignment patterns and age-related changes. In children and adolescents, mean values of 6.5° and 0.7° were reported [8]. In adults, means range from 4.9° to 40.0° [9,10,11,12,13, 20, 22,23,24,25,26,27,28]. In our study, values were higher in the pediatric population, but similar in adults. Scheer et al. [11] described a C2–C7 lordosis increase from 9.4° at 20–34 years to 22.2° over 60 years. Hardacker et al. [12] demonstrated an increase from 37° at 20–30 years to 47° at 61–70 years. Iyer et al. [13] showed an increase from 4.5° at 21–30 years to 15.7° over 71 years. Our results showed that this lordosis increase mainly occurred in the caudal arch, whereas the cranial arch kept stable. Caudal arch lordosis increased with age in parallel to C7-slope, T1-slope and TK. An age-related TK increase was described by Iyer et al. [14] and Hu et al. [15]. This suggests that progressive caudal cervical hyperextension compensating increasing proximal TK to maintain horizontal gaze. This phenomenon was more prominent in males and individuals with high PI. C7-slope and T1-slope represent similar parameters. Mean C7-slope values were 21.3° and 17.4° in children and adolescents [8] and 19.64° in adults [9]. Iyer et al. [13] described a T1-slope increase from 21.7° at 21–30 years to 35.3° over 71 years. Chen et al. [28] demonstrated an increase from 23.0° under 21 years to 28.7° at 41–60 years, like our findings.

Correlation analyses confirmed a strong relationship between C2–C7 lordosis, C7-slope, T1-slope and TK in pediatric and adult populations [8,9,10,11, 13, 20, 24]. Núñez-Pereira et al. [24] found a weak (r = − 0.3) but significant correlation between C7-slope and SS. Lee et al. [10] and Scheer et al. [11] investigated correlation chains between cervical, thoracolumbar and spinopelvic parameters. Correlations exist between cervical and thoracic parameters around the cervicothoracic junction, between TK and LL and between LL and spinopelvic parameters. A direct correlation between cervical parameters, LL and spinopelvic parameters was not evidenced. Among global alignment parameters, Iyer et al. [13] demonstrated weak but significant correlations for SVA-C2–C7 with T1-slope (r = 0.306) and C2–C7 lordosis (r = 0.221). In our study, only SVA-C2 correlated with caudal arch lordosis and T1-slope. Global alignment results showed that SVA-C2 and TK increased with age. This indicates a relationship between progressive anterior C2-plumbline migration, TK changes and compensatory hyperextension of the caudal cervical arch. This phenomenon might influence cervical morphology and the level of cervicothoracic inflexion point which varied from C6 to T3. In our population, 50.9% presented an evenly distributed cervical lordosis. Kyphotic alignment represented only 1.3%. This pattern is more common in patients with scoliosis and thoracic hypokyphosis [3,4,5,6]. Yu et al. [4] and Charles et al. [5] described sigmoid shapes in scoliosis, which represented compensation phenomena of thoracic malalignment. Sigmoid patterns were also found in our asymptomatic population. Proximal kyphosis and distal lordosis represented 34.4%, possibly related to degenerative changes, adaptive C2–plumbline translation and caudal arch increase with age. Proximal lordosis and distal kyphosis represented 13.4%, where TK extended into the caudal cervical spine.

This cross-sectional study has limitations as the evolution of alignment parameters cannot be assessed over a lifetime. Quality-of-life scores and muscle strength were not assessed, although these clinical factors might influence radiographic alignment. Moderate intervertebral disc degeneration and osteoarthritis represent an inevitable age-related phenomenon. Severe changes leading to deformity were excluded. The influence of degenerative changes was not specifically analyzed.

Conclusion

Cervical alignment parameters vary according to age, gender and PI. In the cranial cervical segment, the main changes occur during growth. In the caudal cervical segment, lordosis increases mainly in the lower arch during adulthood, which is strongly related to thoracic kyphosis increase with age. The caudal cervical arch acts as a compensatory segment.