Unbiased Recursive Partitioning to Stratify Patients with Acute Traumatic Spinal Cord Injuries: External Validity in an Observational Cohort Study

Prognostic value of tissue bridges in cervical spinal cord injury: a longitudinal, multicentre, retrospective cohort study

BACKGROUND

The accuracy of prognostication in patients with cervical spinal cord injury (SCI) needs to be improved. We aimed to explore the prognostic value of preserved spinal tissue bridges-injury-spared neural tissue adjacent to the lesion-for prediction of sensorimotor recovery in a large, multicentre cohort of people with SCI.

METHODS

For this longitudinal study, we included patients with acute cervical SCI (vertebrae C1-C7) admitted to one of three trauma or rehabilitation centres: Murnau, Germany (March 18, 2010-March 1, 2021); Zurich, Switzerland (May 12, 2002-March 2, 2019); and Denver, CO, USA (Jan 12, 2010-Feb 16, 2017). Patients were clinically assessed at admission (baseline), at discharge (3 months), and at 12 months post SCI. Midsagittal tissue bridges were quantified from T2-weighted images assessed at 3-4 weeks post SCI. Fractional regression and unbiased recursive partitioning models, adjusted for age, sex, centre, and neurological level of injury, were used to assess associations between tissue bridge width and baseline-adjusted total motor score, pinprick score, and light touch scores at 3 months and 12 months. Patients were stratified into subgroups according to whether they showed better or worse predicted recovery.

FINDINGS

The cohort included 227 patients: 93 patients from Murnau (22 [24%] female); 43 patients from Zurich (four [9%] female); and 91 patients from Denver (14 [15%] female). 136 of these participants (from Murnau and Zurich) were followed up for up to 12 months. At 3 months, per preserved 1 mm of tissue bridge at baseline, patients recovered a mean of 9·3% (SD 0·9) of maximal total motor score (95% CI 7·5-11.2), 8·6% (0·8) of maximal pinprick score (7·0-10·1), and 10·9% (0·8) of maximal light touch score (9·4-12·5). At 12 months post SCI, per preserved 1 mm of tissue bridge at baseline, patients recovered a mean of 10·9% (1·3) of maximal total motor score (8·4-13·4), 5·7% (1·3) of maximal pinprick score (3·3-8·2), and 6·9% (1·4) of maximal light touch score (4·1-9·7). Partitioning models identified a tissue bridge cutoff width of 2·0 mm to be indicative of higher or lower 3-month total motor, pinprick, and light touch scores, and a cutoff of 4·0 mm to be indicative of higher and lower 12-month scores. Compared with models that contained clinical predictors only, models additionally including tissue bridges had significantly improved prediction accuracy across all three centres.

INTERPRETATION

Tissue bridges, measured in the first few weeks after SCI, are associated with short-term and long-term clinical improvement. Thus, tissue bridges could potentially be used to guide rehabilitation decision making and to stratify patients into more homogeneous subgroups of recovery in regenerative and neuroprotective clinical trials.

FUNDING

Wings for Life, International Foundation for Research in Paraplegia, EU project Horizon 2020 (NISCI grant), and ERA-NET NEURON.

Conditional inference tree models to perceive depth of invasion in T1 colorectal cancer

BACKGROUND AND AIM

Accurate diagnosis of invasion depth for T1 colorectal cancer is of critical importance as it decides optimal resection technique. Few reports have previously covered the effects of endoscopic morphology on depth assessment. We developed and validated a novel diagnostic algorithm that accurately predicts the depth of early colorectal cancer.

METHODS

We examined large pathological and endoscopic databases compiled between Jan 2015 and Dec 2018. Training and validation data cohorts were derived and real-world diagnostic performance of two conditional interference tree algorithms (Models 1 and 2) was evaluated against that of the Japan NBI-Expert Team (JNET) classification used by both expert and non-expert endoscopists.

RESULTS

Model 1 had higher sensitivity in deep submucosal invasion than that of JNET alone in both training (45.1% vs. 28.6%, p < 0.01) and validation sets (52.3% vs. 40.0%, p < 0.01). Model 2 demonstrated higher sensitivity than Model 1 (66.2% vs. 52.3%, p < 0.01) in excluding deeper invasion of suspected Tis/T1a lesions.

CONCLUSION

We discovered that machine-learning classifiers, including JNET and macroscopic features, provide the best non-invasive screen to exclude deeper invasion for suspected Tis/T1 lesions. Adding this algorithm improves depth diagnosis of T1 colorectal lesions for both expert and non-expert endoscopists.

Lower motoneuron dysfunction impacts spontaneous motor recovery in acute cervical spinal cord injury

Paresis after spinal cord injury is caused by damage to upper and lower motoneurons and may differentially impact neurological recovery. This prospective monocentric longitudinal observational study investigated the extent and severity of lower motoneuron dysfunction and its impact on upper extremity motor recovery after acute cervical spinal cord injury. Pathological spontaneous activity at rest and/or increased discharge rates of motor unit action potentials recorded by needle electromyography (EMG) were taken as parameters for lower motoneuron dysfunction and its relation to the extent of myelopathy in the first available spine MRI was determined. Motor recovery was assessed by standardized neurological examination within the first 4 weeks (acute stage) and up to 1 year (chronic stage) after injury. Eighty-five muscles of 17 individuals with cervical spinal cord injury (neurological level of injury from C1 to C7) and a median age of 54 (28-59) were examined. The results showed that muscles with signs of lower motoneuron dysfunction peaked at the lesion center (Χ²[2,n=85]=6.6, p=0.04) and that the severity of lower motoneuron dysfunction correlated with T2-weighted hyperintense MRI signal changes in routine spine MRI at the lesion site (spearman ρ=0.31, p=0.01). Muscles exhibiting signs of lower motoneuron dysfunction, as indicated by pathological spontaneous activity at rest and/or increased discharge rates of motor unit action potentials, were associated with more severe paresis in both the acute and chronic stages after spinal cord injury (spearman ρ acute=-0.22, p=0.04 and chronic=-0.31, p=0.004). Moreover, the severity of lower motoneuron dysfunction in the acute stage was also associated with a greater degree of paresis (spearman ρ acute=-0.24, p=0.03 and chronic=-0.35, p=0.001). While both muscles with and without signs of lower motoneuron dysfunction were capable of regaining strength over time, those without lower motoneuron dysfunctions had a higher potential to reach full strength. Muscles with signs of lower motoneuron dysfunction in the acute stage displayed increased amplitudes of motor unit action potentials with chronic-stage needle EMG, indicating reinnervation through peripheral collateral sprouting as compensatory mechanism (Χ²[1,n=72]=4.3, p=0.04). Thus, lower motoneuron dysfunction represents a relevant factor contributing to motor impairment and recovery in acute cervical spinal cord injury. Defined recovery mechanisms (peripheral reinnervation) may at least partially underlie spontaneous recovery in respective muscles. Therefore, assessment of lower motoneuron dysfunction could help refine prediction of motor recovery following spinal cord injury.

Development of a machine learning algorithm for predicting in-hospital and 1-year mortality after traumatic spinal cord injury

BACKGROUND CONTEXT

Current prognostic tools such as the Injury Severity Score (ISS) that predict mortality following trauma do not adequately consider the unique characteristics of traumatic spinal cord injury (tSCI).

PURPOSE

Our aim was to develop and validate a prognostic tool that can predict mortality following tSCI.

STUDY DESIGN

Retrospective review of a prospective cohort study.

PATIENT SAMPLE

Data was collected from 1245 persons with acute tSCI who were enrolled in the Rick Hansen Spinal Cord Injury Registry between 2004 and 2016.

OUTCOME MEASURES

In-hospital and 1-year mortality following tSCI.

METHODS

Machine learning techniques were used on patient-level data (n=849) to develop the Spinal Cord Injury Risk Score (SCIRS) that can predict mortality based on age, neurological level and completeness of injury, AOSpine classification of spinal column injury morphology, and Abbreviated Injury Scale scores. Validation of the SCIRS was performed by testing its accuracy in an independent validation cohort (n=396) and comparing its performance to the ISS, a measure which is used to predict mortality following general trauma.

RESULTS

For 1-year mortality prediction, the values for the Area Under the Receiver Operating Characteristic Curve (AUC) for the development cohort were 0.84 (standard deviation=0.029) for the SCIRS and 0.55 (0.041) for the ISS. For the validation cohort, AUC values were 0.86 (0.051) for the SCIRS and 0.71 (0.074) for the ISS. For in-hospital mortality, AUC values for the development cohort were 0.87 (0.028) and 0.60 (0.050) for the SCIRS and ISS, respectively. For the validation cohort, AUC values were 0.85 (0.054) for the SCIRS and 0.70 (0.079) for the ISS.

CONCLUSIONS

The SCIRS can predict in-hospital and 1-year mortality following tSCI more accurately than the ISS. The SCIRS can be used in research to reduce bias in estimating parameters and can help adjust for coefficients during model development. Further validation using larger sample sizes and independent datasets is needed to assess its reliability and to evaluate using it as an assessment tool to guide clinical decision-making and discussions with patients and families.

Characterizing Natural Recovery of People With Initial Motor Complete Tetraplegia

OBJECTIVE

To determine the differences in neurologic recovery in persons with initial cervical American Spinal Cord Injury Association Impairment Scale (AIS) grades A and B over time.

DESIGN

Retrospective analysis of data from people with traumatic cervical spinal cord injury (SCI) enrolled in the National Spinal Cord Injury Model Systems (SCIMS) database from 2011-2019.

SETTING

SCIMS centers.

PARTICIPANTS

Individuals (N=187) with traumatic cervical (C1-C7 motor level) SCI admitted with initial AIS grade A and B injuries within 30 days of injury, age 16 years or older, upper extremity motor score (UEMS) ≤20 on both sides, and complete neurologic data at admission and follow-up between 6 months and 2 years.

INTERVENTIONS

Not applicable.

MAIN OUTCOME MEASURES

Conversion in AIS grades, UEMS and lower extremity motor scores (LEMS), and sensory scores.

RESULTS

Mean time to initial and follow-up examinations were 16.1±7.3 days and 377.5±93.4 days, respectively. Conversion from an initial cervical AIS grades A and B to motor incomplete status was 13.4% and 50.0%, respectively. The mean UEMS change for people with initial AIS grades A and B did not differ (7.8±6.5 and 8.8±6.1; P=.307), but people with AIS grade B experienced significantly higher means of LEMS change (2.3±7.4 and 8.8±13.9 (P≤.001). The increased rate of conversion to motor incomplete status from initial AIS grade B appears to be the primary driving factor of increased overall motor recovery. Individuals with initial AIS grade B had greater improvement in sensory scores.

CONCLUSIONS

While UEMS recovery is similar in persons with initial AIS grades A and B, the rate of conversion to motor incomplete status, LEMS, and sensory recovery are significantly different. This information is important for clinical as well as research considerations.

Trajectory-Based Classification of Recovery in Sensorimotor Complete Traumatic Cervical Spinal Cord Injury

OBJECTIVE

To test the hypothesis that sensorimotor complete traumatic cervical spinal cord injury (SCI) is a heterogenous clinical entity comprising several subpopulations that follow fundamentally different trajectories of neurologic recovery.

METHODS

We analyzed demographic and injury data from 655 patients who were pooled from 4 prospective longitudinal multicenter studies. Group-based trajectory modeling was applied to model neurologic recovery trajectories over the initial 12 months postinjury and to identify predictors of recovery trajectories. Neurologic outcomes included upper extremity motor score, total motor scores, and American Spinal Injury Association Impairment Scale (AIS) grade improvement.

RESULTS

The analysis identified 3 distinct trajectories of neurologic recovery. These clinical courses included (1) marginal recovery trajectory, characterized by minimal or no improvement in motor strength or change in AIS grade status (remained grade A); (2) moderate recovery trajectory, characterized by low baseline motor scores that improved approximately 13 points or AIS conversion of 1 grade point; (3) good recovery trajectory, characterized by baseline motor scores in the upper quartile that improved to near maximum values within 3 months of injury. Patients following the moderate or good recovery trajectories were younger, had more caudally located injuries, had a higher degree of preserved motor and sensory function at baseline examination, and exhibited a greater extent of motor and sensory function in the zone of partial preservation.

CONCLUSION

Cervical complete SCI can be classified into one of 3 distinct subpopulations with fundamentally different trajectories of neurologic recovery. This study defines unique clinical phenotypes based on potential for recovery, rather than baseline severity of injury alone. This approach may prove beneficial in clinical prognostication and in the design and interpretation of clinical trials in SCI.

Characterizing Natural Recovery after Traumatic Spinal Cord Injury

The predominant tool used to predict outcomes after traumatic spinal cord injury (SCI) is the International Standards for Neurological Classification of Spinal Cord Injury (ISNCSCI), in association with the American Spinal Injury Association (ASIA) Impairment Scale (AIS). These measures have evolved based on analyses of large amounts of longitudinal neurological recovery data published in numerous separate studies. This article reviews and synthesizes published data on neurological recovery from multiple sources, only utilizing data in which the sacral sparing definition was applied for determination of completeness. Conversion from a complete to incomplete injury is more common in tetraplegia than paraplegia. The majority of AIS conversion and motor recovery occurs within the first 6-9 months, with the most rapid rate of motor recovery occurring in the first three months after injury. Motor score changes, as well as recovery of motor levels, are described with the initial strength of muscles as well as the levels of the motor zone of partial preservation influencing the prognosis. Total motor recovery is greater for patients with initial AIS B than AIS A, and greater after initial AIS C than with motor complete injuries. Older age has a negative impact on neurological and functional recovery after SCI; however, the specific age (whether >50 or >65 years) and underlying reasons for this impact are unclear. Penetrating injury is more likely to lead to a classification of a neurological complete injury compared with blunt trauma and reduces the likelihood of AIS conversion at one year. There are insufficient data to support gender having a major effect on neurological recovery after SCI.

Unbiased Recursive Partitioning Enables Robust and Reliable Outcome Prediction in Acute Spinal Cord Injury

Neurological disorders usually present very heterogeneous recovery patterns. Nonetheless, accurate prediction of future clinical end-points and robust definition of homogeneous cohorts are necessary for scientific investigation and targeted care. For this, unbiased recursive partitioning with conditional inference trees (URP-CTREE) have received increasing attention in medical research, especially, but not limited to traumatic spinal cord injuries (SCIs). URP-CTREE was introduced to SCI as a clinical guidance tool to explore and define homogeneous outcome groups by clinical means, while providing high accuracy in predicting future clinical outcomes. The validity and predictive value of URP-CTREE to provide improvements compared with other more common approaches applied by clinicians has recently come under critical scrutiny. Therefore, a comprehensive simulation study based on traumatic, cervical complete spinal cord injuries provides a framework to investigate and quantify the issues raised. First, we assessed the replicability and robustness of URP-CTREE to identify homogeneous subgroups. Second, we implemented a prediction performance comparison of URP-CTREE with traditional statistical techniques, such as linear or logistic regression, and a novel machine learning method. URP-CTREE's ability to identify homogeneous subgroups proved to be replicable and robust. In terms of prediction, URP-CTREE yielded a high prognostic performance comparable to a machine learning algorithm. The simulation study provides strong evidence for the robustness of URP-CTREE, which is achieved without compromising prediction accuracy. The slightly lower prediction performance is offset by URP-CTREE's straightforward interpretation and application in clinical settings based on simple, data-driven decision rules.

Neurophysiological Changes in the First Year After Cell Transplantation in Sub-acute Complete Paraplegia

Neurophysiological testing can provide quantitative information about motor, sensory, and autonomic system connectivity following spinal cord injury (SCI). The clinical examination may be insufficiently sensitive and specific to reveal evolving changes in neural circuits after severe injury. Neurophysiologic data may provide otherwise imperceptible circuit information that has rarely been acquired in biologics clinical trials in SCI. We reported a Phase 1 study of autologous purified Schwann cell suspension transplantation into the injury epicenter of participants with complete subacute thoracic SCI, observing no clinical improvements. Here, we report longitudinal electrophysiological assessments conducted during the trial. Six participants underwent neurophysiology screening pre-transplantation with three post-transplantation neurophysiological assessments, focused on the thoracoabdominal region and lower limbs, including MEPs, SSEPs, voluntarily triggered EMG, and changes in GSR. We found several notable signals not detectable by clinical exam. In all six participants, thoracoabdominal motor connectivity was detected below the clinically assigned neurological level defined by sensory preservation. Additionally, small voluntary activations of leg and foot muscles or positive lower extremity MEPs were detected in all participants. Voluntary EMG was most sensitive to detect leg motor function. The recorded MEP amplitudes and latencies indicated a more caudal thoracic level above which amplitude recovery over time was observed. In contrast, further below, amplitudes showed less improvement, and latencies were increased. Intercostal spasms observed with EMG may also indicate this thoracic “motor level.” Galvanic skin testing revealed autonomic dysfunction in the hands above the injury levels. As an open-label study, we can establish no clear link between these observations and cell transplantation. This neurophysiological characterization may be of value to detect therapeutic effects in future controlled studies.