Get acquainted with three of the seven abstract submissions for EUROSPINE 2019 in Helsinki, which have been elected as Best of Show and will be presented in the Best of Show and Award papers session
Spino-pelvic alignment after short segment transforaminal lumbar interbody fusion (TLIF) – Is correction possible and worthwhile?
Presentation by Markus Loibl, Felix Massen, Tamás Fekete, Daniel Haschtmann, Frank Kleinstück, Dezsö Jeszenszky, Francois Porchet, Anne Mannion
Loss of lumbar lordosis is common in the aging spine, secondary to disc degeneration and disc space collapse, and can be associated with progressive sagittal malalignment. Sagittal alignment is governed by radiological parameters such as pelvic incidence (PI), pelvic tilt (PT), sacral slope (SS) and lumbar lordosis (LL). The restoration of lumbar lordosis may decrease the rate of adjacent segment degeneration, and thereby reduce the risk of undergoing revision surgery. Accordingly, several scores have been proposed to assess the ideal amount of lumbar lordosis. Matching PI and LL, and a low Global Alignment and Proportion (GAP) score are associated with better clinical outcome and lower risk of revision in long fusions. The influence of short transforaminal lumbar interbody fusion (TLIF) on the sagittal profile is controversial. This presentation highlights the clinical and radiological results from 196 patients with no coronal deformity >20° and no previous spine surgery who had undergone TLIF (1-3 segments) for degenerative spinal disorders in our spine centre in 2012.
The first objective of the study was to measure the magnitude of the change in segmental and regional lordosis after short segment TLIF (1-3 segments). Segmental lordosis was increased by 1.3±4.5° per fused segment (p<.01), whereas LL showed no significant difference from pre- to postoperatively. We found a low but significant correlation between the increase in the fused segments lordosis and the decrease in the remaining unfused segments lordosis (R=-0.285, p<.01) (Figure 1). These findings suggest that short segment TLIF has some segmental correction potential, but there is a corresponding reduction of the compensatory hyperlordosis of the remaining segments, such that the overall lumbar lordosis remains largely unchanged.
The second objective was to evaluate the effect of short segment TLIF on spino-pelvic alignment. We formed categories according to LL-PI (balanced, unbalanced, or uncompensated) and the lumbar Global Alignment and Proportion (L-GAP) score (proportioned, moderately disproportioned or severely disproportioned). The proportion of patients in the PI-LL and L-GAP-score categories showed no significant differences pre- to postoperatively. Our results therefore suggest that an individual’s spino-pelvic alignment category is not modified by short segment TLIF.
The final objective was to investigate the effect of spino-pelvic alignment category and restoration of segmental lordosis after short segment TLIF on clinical outcome. The patient-rated outcome, prospectively evaluated using the Core Outcome Measures Index (COMI), improved significantly from baseline to two- and five-year follow-up (Figure 2), but independent of the spino-pelvic alignment and degree of increase in segmental lumbar lordosis after short segment TLIF. Nonetheless, patients that showed an increase in lordosis of the fused segments of at least 3° were more likely to have achieved the minimal clinically important change (MCIC) in COMI score two and five years postoperatively than were those with an increase of less than 3° (Figure 3).
In summary, short segment TLIF is associated with an increase in lordosis within the fused segments, and a subsequent decrease in the lordosis of the unfused segments that otherwise served to maintain a given degree of lordosis for the whole lumbar spine. A good clinical outcome is achieved for the majority of patients at five-years’ follow-up, independent of spino-pelvic alignment, but an increase of the fused segments’ lordosis of at least 3° appears to be associated with better clinical outcome.
Terminal complement complex (TCC): a possible target for intervertebral disc degeneration therapeutics
Presentation by Graciosa Q. Teixeira, Zhiyao Yong, Raquel M. Goncalves, Amelie Kuhn, Michael Ruf, Andreas Nehrlich, Thomas Barth, Uwe M. Mauer, Anita Ignatius, Rolf Brenner, Cornelia Neidlinger-Wilke, Institute of Orthopaedic Research and Biomechanics, Ulm University, Ulm, Germany
Inflammation is known to contribute to disc degeneration (DD). However, there is limited knowledge regarding a possible involvement of the innate immune system, namely of the complement system. The terminal complement complex (TCC), an activation product of the complement system, was previously identified in human pathologic intervertebral discs (Gronblad et al., Spine, 2003), there is still limited knowledge regarding the interplay between disc inflammatory/degenerative environment and TCC formation. Therefore, the present work aims to understand if TCC plays a specific role in the development and progression of DD.
Disc tissues were collected post-mortem from healthy donors, considering two different age groups: i) Young (5F/6M, age 7±7 years old); ii) Elder (4F/4M, age 67±14 years old); as well as iii) patients diagnosed with adolescent idiopathic scoliosis (AIS, 8F/3M, age 15±4) displaying no signs of degeneration; and iv) patients with DD (23F/16M, age 64±12, Pfirrmann grade 3-5), with ethical approval and patients’ informed consent. TCC deposition was investigated in nucleus pulposus (NP), annulus fibrosus (AF) and endplate (EP). Randomly selected Sc and DD patients’ AF, NP and EP expanded cells (passage 2-5) were analyzed for gene expression of TCC-inhibitors CD46, CD55 and CD59. Surface expression of TCC, CD46, CD55 and CD59 was analysed by FACS in fresh and expanded cells. In vitro, cellular TCC deposition was stimulated by 5 per cent human serum medium supplementation (containing components C5 to C9, necessary for TCC formation) and analysed by ELISA. Serum-free medium was used as control. TCC’s lytic activity was measured in the supernatants by erythrocytes lysis test. Statistical analysis was performed with Kruskal-Wallis test.
A significantly higher frequency of TCC+ cells was detected in the NP of DD compared to Elder and Sc groups (p<0.05), and in the EP of both Young (p<0.001) and DD (p<0.05) compared to Elder (Figure 1). Moreover, Young donors presented a significantly higher frequency of TCC+ cells in the EP versus NP (p<0.05). No correlations with age or degeneration degree in DD were observed. Overall, the frequency of TCC+, CD46+, CD55+ and CD59+ AF, NP and EP cells significantly increased with time in culture, becoming similar for Sc and DD cells in passages 2-5. CD46, CD55 and CD59 expression was also similar between Sc and DD cells. Moreover, in presence of human serum, no significant differences were observed between DD and Sc groups for AF, NP or EP expanded cells regarding TCC deposition and cell lysis.
These data suggest that TCC is formed in NP cells of strongly degenerated samples, whereas it is detected in the EP of both Young and DD groups, which might correlate with vascularisation. Moreover, although TCC deposition can be induced in vitro, AF, NP and EP cells isolated from tissues derived from patients with different pathologies seem to lose their native phenotype with time in culture. Further studies are ongoing to understand which microenvironmental factors can activate TCC deposition and if there is a possible functional relevance of the complement system in DD, being a target for new therapeutic approaches.
Artificial intelligence-based adult spinal deformity risk-benefit classification: Hierarchical clustering of 1245 patients and surgeries with machine-based learning and simplified decision trees
Presentation by Ferran Pellisé, Christopher P. Ames, Justin S. Smith, Michael Kelly, Ahmet Alanay, Emre Acaroglu, Francisco Javier Sánchez Pérez-Grueso, Frank Kleinstück, Ibrahim Obeid, Alba Vila-Casademunt, Douglas Burton, Virginie Lafage, Frank Schwab, Christopher I. Shaffrey, Shay Bess, Miquel Serra-Burriel, International Spine Study Group, European Spine Study Group
Introduction: The Schwab-SRS ASD classification is based on disability scores and the sagittal plane and is limited by the lack of preoperative information on associated risk or outcome. The combination of artificial intelligence (AI) based unsupervised learning and expert cluster interpretation will yield a risk-benefit ASD classification, which does not require computer access.
AI-based unsupervised multivariate analyses, such as hierarchical clustering, have been extensively applied across multiple basic science and health fields. Given its marked heterogeneity, application of AI-based hierarchical clustering to ASD could provide data-driven, patient-focused insight. The objective of the present study was to apply AI-based hierarchical clustering analysis of patient types and surgical intervention categories to a large prospectively-collected, multicenter series of surgically treated ASD patients as a step toward defining a classification scheme that optimises overall quality, value, and safety.
Methods: Two independent and compatible prospective multicenter ASD databases, one from the US and the other from Europe, were queried and merged. Inclusion criteria were: age ≥18 and at least one of the following: scoliosis>20°, SVA>5cm, PT>25°, or thoracic kyphosis>60°.
Patient characteristics at baseline were divided into objective measurements (demographics, radiographic parameters) and PROMs (ODI, SRS-22r, SF-36v2). Surgical characteristics included: prior spine surgeries, approach, number of fused levels, use of pelvic fixation, interbody fusions and osteotomies, operative time, bleeding, and length of hospital stay.
Two dendrograms based on hierarchical clustering, built upon ward distances and optimised with the gap method, were generated by fitting the data separately to the patient parameters and to the surgical parameters.
Results: 1,245 patients were included (mean 55.7 years; 77.6 per cent female) in this analysis. Mean surgical time was 346 minutes, and mean EBL was 1660 cc. On average, patients had 10.7 vertebral levels fused and LOS averaged 8.9 days. Major complications were reported in 36.2 per cent of patients.
Patient-based dendrogram and clusters
A total of three clusters were found to be optimal given the sample size and patient characteristics (gap statistic K=0.67). All variables were statistically significantly different across groups (overall p-value <0.004).
The first cluster included 200 patients and is termed the “Young Coronal” cluster, since these patients were the youngest of three clusters (mean age=47.6 years) and had deformity that was predominantly scoliosis (Major Cobb=50.4°). These patients had relatively normal mean global alignment in both the sagittal (SVA=17.3 mm) and coronal (GCA=27.7 mm) planes, had a low incidence of previous spine surgery (8 per cent), and had the best baseline PROM scores.
“Old Revision” cluster (516 patients) is based on the relatively older mean age (62.3 years) and high incidence of previous surgery (48 per cent). These patients had the highest mean SVA (88.1 mm), intermediate mean GCA=28.1 mm, and the lowest major Cobb angle (32.7°). Patients in the Old Revision cluster had the poorest baseline mean PROM values.
“Old Primary” cluster (527 patients) is based on relatively older mean age (61.0 years) and low incidence of previous spine surgery (7 per cent). These patients had intermediate mean values for global alignment, major Cobb angle (36.6°), and baseline PROM scores.
Surgically-based dendrogram and clusters
A total of four surgical types was found to be optimal given the sample size and patient characteristics (gap statistic K=0.68): 3-column osteotomy (PSO/3CO, n=254), interbody fusion (IBF with [n=296] or without decompression [n=216]), single PCO [n=258], and multiple PCO [n=219]. All variables were statistically significantly different across groups (overall p-value <0.001).
Efficiency Grid
The intersection of patient-based clusters with the surgery-based clusters yielded 12 subgroups over nine outcomes, eight PROMs and the incidence of major complications. This structure allows comparison of the ratio of risk-to-benefit of surgical interventions over homogeneous patients.
Conclusion:
Unsupervised AI-based hierarchical
clustering provides a novel ASD classification methodology which facilitates
inclusion and simultaneous analysis of significantly more overall patient
demographics, frailty factors, radiographic measurements, and functional
status-based data points than existing classification schemes. In addition to
creating a novel AI-based ASD classification which may enhance outcome and
complication prediction, pattern identification may facilitate treatment
optimisation by educating surgeons on which treatment patterns yield optimal
improvement with lowest
risk.
Reference: Ames CP, Smith JS, Pellisé et al. Artificial Intelligence Based Hierarchical Clustering of Patient Types and Intervention Categories in Adult Spinal Deformity Surgery: Towards a New Classification Scheme that Predicts Quality and Value. Spine. 2019 Jul 1;44(13):915-926
The Best of Show session will take place on Friday, 18 October, from 10.30am-12pm. After presentation of all seven abstracts that were selected best in the blinded peer-reviewed selection process beforehand, the audience gets to vote via the conference app and decides who gets to win EUR 2,000 and a certificate.
