T1 slope minus cervical lordosis in multilevel anterior cervical discectomy and fusion with or without plate: a mean 6-year follow up

Purpose Cervical alignment between self-locked cage (SLC) and anterior cage-with-plate (ACP) system was debatable, especially in multilevel anterior cervical discectomy and fusion (ACDF). T1 slope minus C2-C7 cervical lordosis (T1S-CL) was critical for cervical-balance while without identied threshold. Thus the study was to compare sagittal alignment between SLC and ACP on 3-level ACDF and to identify the threshold of T1S-CL in 3-level ACDF. of T1S-CL with 20.7° in multilevel ACDF.


Introduction
Anterior cervical discectomy and fusion (ACDF) is a world-wide and well-recontrusted strategy for cervical spondylotic myelopathy (CSM) [1]. The mordern anterior cage-with-plate system (ACP) consisting of polyetheretherketone cage and titanium plates supports adequate stability of the index segments, while it has been reported, albeit transient, potential disadvantages for its pro le and soft tissue dissection [2][3].
The stand-alone self-locked cage (SLC), as the integrated zero-pro le cage-plate devices, have been applied for CSM over a decade [4]. But SLC is not always awless and debates on comlications and e cacy between SLC and ACP have never be slient. Multilevel ACDF, through simple approach and direct decompression, makes up for the weakness of posterior surgery such as laminoplasty (LP) [5]. However, contrasted with single-or double-level, there were much fewer publications on multilevel ACDF with longterm follow-up.
Cervical sagittal alignment, as the partial of spino-pelvic parameters, has been gradually paid attention to as the non-neglectful factors for surgery-programming. The alignment parameters, especially C2-C7 sagittal vertical axis (C2-C7 SVA) and T1 slope (T1S), have been reported in close correlation with quality of life (QOL) while it remains controversial [6][7]. Recently, a concept of T1S minus C2-C7 cervical lordosis (T1S-CL), was proposed, which analogied to the recognized measurement of pelvic incidence (PI) minus lumbar lordosis (LL) in lumbar-pelvic alignment with the threshold of 10° [8]. Although having been de ned resonable range of T1S-CL ≤ 20° in some cases, it is obstructing to reach consensus since the balance point of T1S-CL varies in multifarious approaches. Fortunately, it has been proved, with little dispute, loss of CL was initiating agent for alignment disorders, just like LL in lumbar spine and T1S-CL was the key point for evaluating cervical balance [9].
On the one hand, the requirements for proper reconstruction for alignment was perhaps discrepant between LP and ACDF based on the different surgery-indications [5]. On the other hand, the alignment of 3-level ACDF was malleable because of wide surgical exposure, long-segment fusion and sometimes anterior-plating, which has not been quantized with QOL yet. There, the study performed a long-term follow, aiming at (1) comparing cervical alignment and clinical e cacy between SLC and ACP on 3-level ACDF; (2) identifying the relationship between T1S-CL and QOL.
1 Materials And Methods

Participants
The single-center retrospective comparative study was performed from January 2011 to December 2014. An overall 110 patients with CSM having undergone ACDF were enrolled, who were divided into SLC group with the implants of SLC and ACP group with the system of ACP. The study acquired approval from Ethics Committee of our institution and all participants have signed informed consents. The patients were included required three-level ACDF with at-least 6-week uncontrolling conservation treatment and the operated segments were C3-C6 or C4-C7 by pure either SLC or ACP. Then, the radiographs were intact and clear to be measured. The exclusion criteria were (1) patients with a history cervical spine surgery; (2) cases combined with other disease such as radiculopathy of CS, cervical disc hernia and motor neuron disease; (3) hybird surgery by inserting SLC with ACP or intervertebral cages with arti cial prosthesis; (4) operations for cervical spine tumor, fracture or infection; (5) too fuzzy of radiological parameters to be measured and (6) loss of follow-up or unwlling for consecutive observation.
Eventually, a total of 73 CSM patients with a mean of 6.1-year follow-up (follow-up rate of 66.4%) were included, with 45 patients in SLC group and 28 cases in ACP group. The two cohorts were well-matched in gender, age and body mass indes (BMI) with comparable follow-up period (P 0.05). The C4-C7 was major operated segments (74.5% vs 77.8%) of both groups (P=0.782). Operation time was shorter in SLC group (P <0.05) while hemorrhage was comparable (P 0.05) ( Table 1).

ACDF procedures
A right-sided incision was utilized for both SLC and ACP groups and standard Smith-Robinson approach was performed. Then, three consecutive cages were orderly inserted after complete decompression. Zeropro le-anchored spacer ROI-C (LDR, Troyes, France) were embeded into in SLC group while MC+ (LDR, Troyes, France) or Solis poly-ether-etherketone cages (Stryker, Michigan, USA) combined with anterior plates (DePuySynthes, New Jersey, USA) were utilized as ACP system. All operations were performed by the same senior surgeon and patients were directed to wear collar for 2 months after ACDF, followed by long-term regulating function training in our institution.

Cervical sagittal alignment
Cervical spine alignment were dipicted by 4 radiological parameters. (1) C2-C7 cervical lordosis (CL): the angle was from lower endplate of C2 to lower endplate of C7; (2) C2-C7 SVA: the vertical was from C2 plumb line to posterior margin of the upper endplate of C7; (3) T1S: the angle was from upper endplate of T1 to horizontal line and (4) T1S-CL: was used for cervical spine balance evaluation (Fig 1). In addition, adjacent segment degeneration (ASD) was also in assessment, which was judged by the presence of disk space narrowing moret han 50%, new or enlarged osteophytes, endplate sclerosis or increased calci cation of the anterior longitudinal ligament [10]. All measurements were obtained from standard lateral X-ray of cervical spine at preoperation (baseline) and last follow-up in December 2019.
The parameters were independently measured by two blind-observers. The reliability and reproducibility of inter-and inntra-observer were quantitized by intraclass correlation coe cient (ICC). ICC ranged from 0 to 1, where ICC <0.40 suggested a poor reliability and ICC 0.75 represented a high reliability. The interobserver ICC showed 0.79 for CL, 0.83 for C2-C7 SVA, 0.75 for T1S and 0.81 for T1S-CL, respectively and intra-observer ICC were 0.85 for CL, 0.87 for C2-C7 SVA, 0.77 for T1S and 0.80 for T1S-CL. All measurements suggested a quali ed reliability.

QOL evaluation
Neck disability index (NDI) and Japanese Orthopedic Association (JOA) score were both recorded at preoperation and nal-visit for evaluating QOL. Recovery rate (RR) of JOA was calculated as: RR (%)= (Post-op JOA-Pre-op JOA)/(17-Pre-op JOA)×100. A higher NDI meant a worse QOL and NDI was class ed into 5 grades. 0-10 and 10-20 points respectively represented an excellent and good function of cervical spine and 40-50 showed an extreme disabled ststus [11]. The NDI of 20 was di ned as the threshold of nureological function and QOL after ACDF at last follow-up.

Statistical analysis
Chi-square analysis were used to compare dichotomous between SLC and ACP group. Independent sample t test and Mann-Whitney U-test were used to for comparisons on continuous variable between groups and paired t test was for variables beween baseline and last follow-up. Pearson correlation analysis was applied between radiological parameters and clinical outcomes. Both linear and logistic regression model were established to determine a possible threshold of sagittal parameters, which was most closely correlated to clinical outcomes. A threshold of NDI of 20 points was determined when linear regression analysis performed and cutoff value of predictors by ROC curve was described following logistic regression. The range of T1S-CL was de ned corresponding to C2-C7 SVA, which was regarded as a threshold before NDI signi cantly affected. SPSS Statistics 22.0 (International Business Machines Corporation, Armonk, NY, USA) was the software for analysis and P <0.05 showed statistically signi cant.

Comparisons between SLC and ACP group
For all cases underwent ACDF, the CL at preoperation and last follow-up were respectively 9.6±12.9° and 14.9±8.6° (P <0.01); the T1S got larger and T1SCL decreased at last compared to baseline (P <0.05), while no statistical change of C2-C7 SVA. At baseline, there were no signi cances on CL, C2-C7 SVA and T1S (P 0.05) between SLC and ACP group while T1S-CL was larger in ACP group than SLC (P=0.046). All parameters at last follow-up were of no differences between groups, as well as their changes (all P 0.05). For intra-group comparisons, CL got improved at last follow-up compared to that of reoperation in both groups (both P=0.003), so were T1S (P=0.015 and P=0.039, respectively). C2-C7 SVA of both group and T1S-CL at SLC group kept stable contrasted to baseline while T1S-CL in ACP group decreased at last (P=0.021). The incidence of upper and lower ASD were respectively 43.8% and 68.5% and there were no signi cances of upper or lower ASD between groups (P 0.05) ( Table 2). The number of CL-increasd from SLC (29 cases) and ACP group (23 cases In total, all cases underwent ACDF from both groups acquired signi cantly inprovement on NDI and JOA compared to baseline (all P <0.001) (Fig 2) . There were negetive correlation between NDI and JOA at baseline (r=-0.256 and P=0.029) and nal visit (r=-0.300, P=0.014).
There was no in uening factors for NDI at baseline through linear regression analysis, so was JOA at baseline and at last follow-up. When multiple linear regression performed on NDI at nal visit, it showed C2-C7 SVA was independent risk factor (P <0.001) and the tting formula was NDI = 2.85 + 0.49 × C2-C7 SVA, where the threshold of C2-C7 SVA was 35 mm once substituting NDI of 20 (Table 5). Then CL, C2-C7 SVA and T1S-CL were included as independent factors for forward stepwise multiple logistic regression on nal NDI, which revealed that C2-C7 SVA was signi cant factor (χ2=34.02, p =0.001) for NDI (odds ratio =1.51). The ROC-AUC was 0.955 with Youden index of 0.847 and the cutoff value of C2-C7 SVA was 23.4 mm. Thus, two regression models predicted a threshold C2-C7 SVA of 29.2 mm (Fig 4 (A-B), Fig 5). 3 Discussion ACDF was achieved in cases with radiographic signs of instability or no motion at the target levels, with or without facet degeneration. To date, there have been few reports on the biomechanical effect of ACDF involving three or more levels and its clinical indications and contraindications are unclear yet. The main indication of multilevel ACDF dued to symptomatic multilevel cervical degenerative disc disease with radiculopathy or myelopathy with strict and accurate re ection of neuro-dysfunction on body-dominating area [3,12]. Whether through ACP or SLC system, 3-level ACDF have been proved reliable and effective on decompression, stability and rebuilding of alighment, especially with the more advanced operating-skill and consequently extended sugical-indications on CSM [13]. The newer launched SLC was approved by some researchers for its zero-pro le design, simple inplanation, probable fewer comlications such as dysphagia contrasted to former-born ACP system. While it holds debatable since the quli ed costeffectiveness brought by ACP keeps irreplaceable [14]. Therefore, a long-term observation on multilevel SLC versus ACP in this study may support strength on the viewpoints.
In this tudy, the sagittal alignment parameters are comparable between groups after surgery and most are of no signi cances except T1SCL. Although speci c approaches are largely determined by surgeon based on nonuniform surgical-indications, there is a agreement that ACP are likely to be used on cases with straighter spine and imbalanced inclination since kyphotic deformities can be corrected by plate streching [15], which is one of the somewhat discrepancies on indications between ACP and SLC. The improvement of CL and T1S at nal with no signi cants suggests a well-reconstruction in both groups. As a retrospective study, the change of alignment is induced to extremely avoid selection and reporting bias. The indifference of variable changes between ACP and SLC further consolidates the equivalent e cacy in alignment reestablishment. A main concerning after ACDF was the potential ASD with Page 7/21 increased rigidity. Studies have shown that anierior plate is more likely to accelerate degenerative changes in adjacent segments [16]. However, a meta-analysis performed by Zhang et al. [17] showed no difference in ASD incidence between the SSC and ACP groups with a multilevel ACDF. Our data showed a long-term incidence of 70% on ASD with no intergroup signi cance, in consistence with Zhang et al. In addition, no case here underwent reoperation for adjacent segment pathology indicated a comparable and limited impact on ASD progression.
The rebuilting of sagittal alignment was emphasized as a main goal in multilevel ACDF and cervical curvation acquires improved in this series after integrating data of the two group with little heterogeneity.
Cervical alignment could be corrected by release of anterior tissue, removal of osteophyte, prepocess of disc and endplate bed, the shape of cages and sustaining of implants [5,18]. In addition, three levels occupies majority of the overall cervical spine and correction on operated segments is more suitable for stress distribution physiologically [19]. Alberto et al.
[20] found a straight cervical spine was related to increased SVA and larger T1SCL,suggesting an interact among parameters. The loss of CL forced an increase of T1S-CL tending to imbalance, imitating the larger PI-LL in loss of LL with pelvis retroversion, ensuing the gravity of head shifts forward to keep horizontal sight with excessive C2-C7 SVA. Cervical balance was evaluated by various factors, where cervical imbalance was de ned as T1S > 40° or C2-C7 SVA > 40 mm without consensus [6,21]. Grasso et al. [22] proposed T1S was an important parameter while T1S-CL, like PI-LL, was a better representative than T1S alone, the latter involving two indexes re ecting a further interlocking status. Hyun et al. [9] rstly determined analogous relationship existed in posterior cervical fusion using T1S-CL, where the threshold of C2-C7 SVA was 50 mm corresponded to a T1S-CL value of 26.1°.
There is correlation between NDI and parameters at nal visit but not at preoperation. It is considered that QOL of patients with CSM was affected by many factors before surgery, especially the severity of compression on spinal cord and no single factor can absolutely dominate neuro-function. Then the compression is elimated by the same surgeon, NDI was associated with appropriate sagittal alignment, in this condition, cervical balance status approximates physiological anatomy with most comfortable biomechanics and minimal power consumption, as well as lower tension of paraspinal muscle [23]. The identi ed relationship of NDI instead of JOA dues to JOA was the patient-reported assessment tool to address each of these domains. Unlike JOA, the NDI is a mix of functional-and pain-status inquiries and more ts the signi cance of QOL [24][25].
The reasonable range of C2-C7 SVA of 29.2 mm corresponded to T1S-CL of 20.7° is narrower than posterior cervical fusion reported by Hyun et al., which emphasizes the different threshold in various approaches. Staub et al. [26] implied normative CL can be predicted via T1S-CL= 16.5° ± 2° based on 103 patients with cervical spine deformity. It is considered that there were differences between anterior and posterior approach on surgical-indications, internal-xtions, posterior muscle striping and preservation of ligament complex, where the solid xtion by lateral mass screw and bowstring effect from posterior muscle requires wider range of T1SCL [5,27]. In addition, spinal canal volume changes with extionextension posture while posterior such as LP enable a steady enlargement of spinal canal compared to ACDF [28], allowing larger deviation from neutral alignment without affecting QOL. Other factors such as no debongding on ossi cation of posterior ligament, disc and osteophyte in posterior approach probably tolerates extensive T1S-CL and C2-C7 SVA. Furthermore, the cutoff value of NDI is 25 in publications while most cases (95.9%) are lower than 25 after ACDF, which will bring reporting bias on de niting threshold of T1S-CL and consequently NDI of 20 is adopted.
This study rstly identi es the signi cance of cervical sagittal parameters in three-level ACDF. Then the cutoff value was quanti ed in this population through the relationship with QOL, just like PI-LL for evaluation of lumbar-pelvic matching. When T1S-CL 20.7° after ACDF, the cervical spine will be involved in imbalance with a straight alignment and a predictive unsatisfactory QOL, which supports strategy for speci c program for ACDF [9,29]. The conclusions proposes partial theory for better exploration on the overall sagittal spine-pelvic biomechanics. There are some limitations: Firstly, the sample size of both groups are small and a larger population may support a strength veri cation. Then, the observation focus on regional parameters on cervical spine instead of the whole spine, which probably bring distinctive points through effect of biomechanics chain. The comparisons are mainly performed on alignments, without other radiological parameters such as range of motion and complications,which were mentioned in publications and not replenished in. The threshold of T1S-CL in this study are just suitable for 3-level ACDF on CSM, the rationality for other types of cervical spine disease need further exploration.

Conclusions
Three-level ACDF provides a satisfactory and stable QOL on the treatment of CSM with a mean 6-year follow up. There is comparable capacity on sagittal alignment reconstruction and incidence of ASD between SLC and ACP system. NDI is in close correlation with CL, C2-C7 SVA and T1S-CL after ACDF. In speci c, C2-C7 SVA less than 29.2 mm is corresponded to a matching of T1S-CL less than 20.7°, where cervical spine is in balance after ACDF with a better QOL. This study has obtained ethics approval and consent of the ethics committee in our hospital.

Consent for publication:
Not applicable Availability of data and material: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.      Comparisons on NDI and JOA between SLC and ACP group and between at baseline and at follow-up endpoint. **: P <0.01 between the same variables Diagram on the interaction among alignment parameters. C2-7 SVA negetively correlated to CL while positively to T1S and T1S-CL. The increase of CL corresponded to a higher T1S to coordinate a maintainance of gravity of head and horizontal sight with a less C2-C7 SVA, the cervical spine was in trend of negetive-imblance. Inversely, a straight cervical spine was accompanied with lower T1S and positive cevical-imbalance, where the C2-C7 SVA positively enlarged.