Optimizing the Definition of Proximal Junctional Kyphosis: A Sensitivity Analysis

STUDY DESIGN

Diagnostic binary threshold analysis.

OBJECTIVE

(1) Perform a sensitivity analysis demonstrating the test performance metrics for any combination of proximal junctional angle (PJA) magnitude and change; (2) Propose a new proximal junctional kyphosis (PJK) criteria.

SUMMARY OF BACKGROUND DATA

Previous definitions of PJK have been arbitrarily selected and then tested through retrospective case series, often showing little correlation with clinical outcomes.

MATERIALS AND METHODS

Surgically treated adult spinal deformity patients (≥4 levels fused) enrolled into a prospective, multicenter database were evaluated at a minimum 2-year follow-up for proximal junctional failure (PJF). Using PJF as the outcome of interest, test performance metrics including sensitivity, positive predictive value, and F1 metrics (harmonic mean of precision and recall) were calculated for all combinations of PJA magnitude and change using different combinations of perijunctional vertebrae. The combination with the highest F1 score was selected as the new PJK criteria. Performance metrics of previous PJK definitions and the new PJK definition were compared.

RESULTS

Of the total, 669 patients were reviewed. PJF rate was 10%. Overall, the highest F1 scores were achieved when the upper instrumented vertebrae -1 (UIV-1)/UIV+2 angle was measured. For lower thoracic cases, out of all the PJA and magnitude/change combinations tested, a UIV-1/UIV+2 magnitude of -28° and a change of -20° was associated with the highest F1 score. For upper thoracic cases, a UIV-1/UIV+2 magnitude of -30° and a change of -24° were associated with the highest F1 score. Using PJF as the outcome, patients meeting this new criterion (11.5%) at 6 weeks had the lowest survival rate (74.7%) at 2 years postoperative, compared with Glattes (84.4%) and Bridwell (77.4%).

CONCLUSIONS

Out of all possible PJA magnitude and change combinations, without stratifying by upper thoracic versus lower thoracic fusions, a magnitude of ≤-28° and a change of ≤-22° provide the best test performance metrics for predicting PJF.

Uncertainty-driven dynamics for active learning of interatomic potentials

Machine learning (ML) models, if trained to data sets of high-fidelity quantum simulations, produce accurate and efficient interatomic potentials. Active learning (AL) is a powerful tool to iteratively generate diverse data sets. In this approach, the ML model provides an uncertainty estimate along with its prediction for each new atomic configuration. If the uncertainty estimate passes a certain threshold, then the configuration is included in the data set. Here we develop a strategy to more rapidly discover configurations that meaningfully augment the training data set. The approach, uncertainty-driven dynamics for active learning (UDD-AL), modifies the potential energy surface used in molecular dynamics simulations to favor regions of configuration space for which there is large model uncertainty. The performance of UDD-AL is demonstrated for two AL tasks: sampling the conformational space of glycine and sampling the promotion of proton transfer in acetylacetone. The method is shown to efficiently explore the chemically relevant configuration space, which may be inaccessible using regular dynamical sampling at target temperature conditions.

Diagnosis-Related Group-Based Payments for Adult Spine Deformity Surgery Significantly Vary across Centers: Results from a Multicenter Prospective Cohort Study

BACKGROUND

To investigate the variation in total episode-of-care (EOC) payment and quality-adjusted life-year (QALY) gain for complex adult spine deformity surgeries in the United States, adjusting for case type and surgeon preferences.

METHODS

Patients aged >18 years with adult spine deformity with Medicare Severity-Diagnosis-Related Groups (DRGs) 453-460 and a minimum of 2 years of follow-up from index surgery were included. Index and total payments were calculated using Medicare's Inpatient Prospective Payment System. All costs were adjusted for inflation to 2020 U.S. dollar values. QALYs gained were calculated using baseline, 1-year, and 2-year Short-Form 6D scores. Mixed-effect models were used to estimate the proportion of variation in total EOC payment and QALY gain.

RESULTS

A total of 330/543 patients from 6 sites were included. Mean age was 62.4 ± 11.9 years, 79% were women, and 92% were white. The mean index and total EOC payment were $77,302 and $93,182, respectively. Patients gained on average 0.15 QALY (P < 0.0001) 2 years after surgery. In unadjusted analysis, 39% of the variation in total EOC payment across the 6 centers was attributable to relative weight of DRG and base rate. Adjusting for patient and procedural factors increased the proportion of variation in total EOC payments across the centers to 56%. Less than 2% of the variation in QALY gain was observed across the 6 centers.

CONCLUSIONS

Medicare-based payments for complex spine deformity fusions are primarily driven by relative weight of the DRG and the hospital's base rate. Patient and procedural factors are unaccounted for in the DRG-based payments made to the providers.

Proximal junctional failure in primary thoracolumbar fusion/fixation to the sacrum/pelvis for adult symptomatic lumbar scoliosis: long-term follow-up of a prospective multicenter cohort of 160 patients

OBJECTIVE

Proximal junctional failure (PJF) is a severe form of proximal junctional kyphosis. Previous reports on PJF have been limited by heterogeneous cohorts and relatively short follow-ups. The authors' objectives herein were to identify risk factors for PJF and to assess its long-term incidence and revision rates in a homogeneous cohort.

METHODS

The authors reviewed data from the Adult Symptomatic Lumbar Scoliosis 1 trial (ASLS-1), a National Institutes of Health-sponsored prospective multicenter study. Inclusion criteria were an age ≥ 40 years, ASLS (Cobb angle ≥ 30° and Oswestry Disability Index [ODI] ≥ 20 or Scoliosis Research Society revised 22-item questionnaire [SRS-22r] score ≤ 4.0 in pain, function, or self-image domains), and primary thoracolumbar fusion/fixation to the sacrum/pelvis of ≥ 7 levels. PJF was defined as a postoperative proximal junctional angle (PJA) change > 20°, fracture of the uppermost instrumented vertebra (UIV) or UIV+1 with > 20% vertebral height loss, spondylolisthesis of UIV/UIV+1 > 3 mm, or UIV screw dislodgment.

RESULTS

One hundred sixty patients (141 women) were included in this analysis and had a median age of 62 years and a mean follow-up of 4.3 years (range 0.1-6.1 years). Forty-six patients (28.8%) had PJF at a median of 0.92 years (IQR 0.14, 1.23 years) following surgery. Based on Kaplan-Meier analyses, PJF rates at 1, 2, 3, and 4 years were 14.4%, 21.9%, 25.9%, and 27.4%, respectively. On univariate analysis, PJF was associated with greater age (p = 0.0316), greater body mass index (BMI; p = 0.0319), worse baseline patient-reported outcome measures (PROMs; ODI, SRS-22r, and SF-12 Physical Component Summary [PCS]; all p < 0.04), the use of posterior column osteotomies (PCOs; p = 0.0039), and greater postoperative thoracic kyphosis (TK; p = 0.0031) and PJA (p < 0.001). The use of UIV hooks was protective against PJF (p = 0.0340). On regression analysis (without postoperative measures), PJF was associated with greater BMI (HR 1.077, 95% CI 1.007-1.153, p = 0.0317), lower preoperative PJA (HR 0.607, 95% CI 0.407-0.906, p = 0.0146), and greater preoperative TK (HR 1.362, 95% CI 1.082-1.715, p = 0.0085). Patients with PJF had worse PROMs at the last follow-up (ODI, SRS-22r subscore and self-image, and SF-12 PCS; p < 0.04). Sixteen PJF patients (34.8%) underwent revision, and PJF recurred in 3 (18.8%).

CONCLUSIONS

Among 160 primary ASLS patients with a median age of 62 years and predominant coronal deformity, the PJF rate was 28.8% at a mean 4.3-year follow-up, with a revision rate of 34.8%. On univariate analysis, PJF was associated with a greater age and BMI, worse baseline PROMs, the use of PCOs, and greater postoperative TK and PJA. The use of UIV hooks was protective against PJF. On multivariate analysis (without postoperative measures), a higher risk of PJF was associated with greater BMI and preoperative TK and lower preoperative PJA.

Accuracy of Rod Contouring to Desired Angles With and Without a Template: Implications for Achieving Desired Spinal Alignment and Outcomes

STUDY DESIGN

Biomechanical Study.

OBJECTIVE

The search for optimal spinal alignment has led to the development of sophisticated formulas and software for preoperative planning. However, preoperative plans are not always appropriately executed since rod contouring during surgery is often subjective and estimated by the surgeon. We aimed to assess whether rods contoured to specific angles with a French rod bender using a template guide will be more accurate than rods contoured without a template.

METHODS

Ten experienced spine surgeons were requested to contour two 125 × 5.5 mm Ti64 rods to 40°, 60° and 80° without templates and then 2 more rods using 2D metallic templates with the same angles. Rod angles were then measured for accuracy and compared.

RESULTS

Average angles for rods bent without a template to 40°, 60° and 80° were 60.2°, 78.9° and 97.5°, respectively. Without a template, rods were overbent by a mean of 18.9°. When using templates of 40°, 60° and 80°, mean bend angles were 41.5°, 59.1° and 78.7°, respectively, with an average underbend of 0.2°. Differences between the template and non-template groups for each target angle were all significant (p < 0.001).

CONCLUSIONS

Without the template, surgeons tend to overbend rods compared to the desired angle, while surgeons improved markedly with a template guide. This tendency to overbend could have significant impact on patient outcomes and risk of proximal junctional failure and warrants further research to better enable surgeons to more accurately execute preoperative alignment plans.

How Good Are Surgeons at Achieving Their Preoperative Goal Sagittal Alignment Following Adult Deformity Surgery?

STUDY DESIGN

Multicenter, prospective cohort.

OBJECTIVES

Malalignment following adult spine deformity (ASD) surgery can impact outcomes and increase mechanical complications. We assess whether preoperative goals for sagittal alignment following ASD surgery are achieved.

METHODS

ASD patients were prospectively enrolled based on 3 criteria: deformity severity (PI-LL ≥25°, TPA ≥30°, SVA ≥15 cm, TCobb≥70° or TLCobb≥50°), procedure complexity (≥12 levels fused, 3-CO or ACR) and/or age (>65 and ≥7 levels fused). The surgeon documented sagittal alignment goals prior to surgery. Goals were compared with achieved alignment on first follow-up standing radiographs.

RESULTS

The 266 enrolled patients had a mean age of 61.0 years (SD = 14.6) and 68% were women. Mean instrumented levels was 13.6 (SD = 3.8), and 23.2% had a 3-CO. Mean (SD) offsets (achieved-goal) were: SVA = -8.5 mm (45.6 mm), PI-LL = -4.6° (14.6°), TK = 7.2° (14.7°), reflecting tendencies to undercorrect SVA and PI-LL and increase TK. Goals were achieved for SVA, PI-LL, and TK in 74.4%, 71.4%, and 68.8% of patients, respectively, and was achieved for all 3 parameters in 37.2% of patients. Three factors were independently associated with achievement of all 3 alignment goals: use of PACs/equivalent for surgical planning (P < .001), lower baseline GCA (P = .009), and surgery not including a 3-CO (P = .037).

CONCLUSIONS

Surgeons failed to achieve goal alignment of each sagittal parameter in ∼25-30% of ASD patients. Goal alignment for all 3 parameters was only achieved in 37.2% of patients. Those at greatest risk were patients with more severe deformity. Advancements are needed to enable more consistent translation of preoperative alignment goals to the operating room.

Can unsupervised cluster analysis identify patterns of complex adult spinal deformity with distinct perioperative outcomes?

OBJECTIVE

The objective of this study was to use an unsupervised cluster approach to identify patterns of operative adult spinal deformity (ASD) and compare the perioperative outcomes of these groups.

METHODS

A multicenter data set included patients with complex surgical ASD, including those with severe deformities, significant surgical complexity, or advanced age who underwent a multilevel fusion. An unsupervised cluster analysis allowing for 10% outliers was used to identify different deformity patterns. The perioperative outcomes of these clusters were then compared using ANOVA, Kruskal-Wallis, and chi-square tests, with p values < 0.05 considered significant.

RESULTS

Two hundred eighty-six patients were classified into four clusters of deformity patterns: hyper-thoracic kyphosis (hyper-TK), severe coronal, severe sagittal, and moderate sagittal. Hyper-TK patients had the lowest disability (mean Oswestry Disability Index [ODI] 32.9 ± 17.1) and pain scores (median numeric rating scale [NRS] back score 6, leg score 1). The severe coronal cluster had moderate functional impairment (mean physical component score 34.4 ± 12.3) and pain (median NRS back score 7, leg score 4) scores. The severe sagittal cluster had the highest levels of disability (mean ODI 49.3 ± 15.6) and low appearance scores (mean 2.3 ± 0.7). The moderate cluster (mean 68.8 ± 7.8 years) had the highest pain interference subscores on the Patient-Reported Outcomes Measurement Information System (mean 65.2 ± 5.8). Overall 30-day adverse events were equivalent among the four groups. Fusion to the pelvis was most common in the moderate sagittal (89.4%) and severe sagittal (97.5%) clusters. The severe coronal cluster had more osteotomies per case (median 11, IQR 6.5-14) and a higher rate of 30-day implant-related complications (5.5%). The severe sagittal and hyper-TK clusters had more three-column osteotomies (43% and 32.3%, respectively). Hyper-TK patients had shorter hospital stays.

CONCLUSIONS

This cohort of patients with complex ASD surgeries contained four natural clusters of deformity, each with distinct perioperative outcomes.

Rod fractures in thoracolumbar fusions to the sacrum/pelvis for adult symptomatic lumbar scoliosis: long-term follow-up of a prospective, multicenter cohort of 160 patients

OBJECTIVE

Previous reports of rod fracture (RF) in adult spinal deformity are limited by heterogeneous cohorts, low follow-up rates, and relatively short follow-up durations. Since the majority of RFs present > 2 years after surgery, true occurrence and revision rates remain unclear. The objectives of this study were to better understand the risk factors for RF and assess its occurrence and revision rates following primary thoracolumbar fusions to the sacrum/pelvis for adult symptomatic lumbar scoliosis (ASLS) in a prospective series with long-term follow-up.

METHODS

Patient records were obtained from the Adult Symptomatic Lumbar Scoliosis-1 (ASLS-1) database, an NIH-sponsored multicenter, prospective study. Inclusion criteria were as follows: patients aged 40-80 years undergoing primary surgeries for ASLS (Cobb angle ≥ 30° and Oswestry Disability Index ≥ 20 or Scoliosis Research Society-22r ≤ 4.0 in pain, function, and/or self-image) with instrumented fusion of ≥ 7 levels that included the sacrum/pelvis. Patients with and without RF were compared to assess risk factors for RF and revision surgery.

RESULTS

Inclusion criteria were met by 160 patients (median age 62 years, IQR 55.7-67.9 years). At a median follow-up of 5.1 years (IQR 3.8-6.6 years), there were 92 RFs in 62 patients (38.8%). The median time to RF was 3.0 years (IQR 1.9-4.54 years), and 73% occurred > 2 years following surgery. Based on Kaplan-Meier analyses, estimated RF rates at 2, 4, 5, and 8 years after surgery were 11%, 24%, 35%, and 49%, respectively. Baseline radiographic, clinical, and demographic characteristics were similar between patients with and without RF. In Cox regression models, greater postoperative pelvic tilt (HR 1.895, 95% CI 1.196-3.002, p = 0.0065) and greater estimated blood loss (HR 1.02, 95% CI 1.005-1.036, p = 0.0088) were associated with increased risk of RF. Thirty-eight patients (61% of all RFs) underwent revision surgery. Bilateral RF was predictive of revision surgery (HR 3.52, 95% CI 1.8-6.9, p = 0.0002), while patients with unilateral nondisplaced RFs were less likely to require revision (HR 0.39, 95% CI 0.18-0.84, p = 0.016).

CONCLUSIONS

This study provides what is to the authors' knowledge the highest-quality data to date on RF rates following ASLS surgery. At a median follow-up of 5.1 years, 38.8% of patients had at least one RF. Estimated RF rates at 2, 4, 5, and 8 years after surgery were 11%, 24%, 35%, and 49%, respectively. Greater estimated blood loss and postoperative pelvic tilt were significant risk factors for RF. These findings emphasize the importance of long-term follow-up to realize the true prevalence and cumulative incidence of RF.

Validation of Adult Spinal Deformity Surgical Outcome Prediction Tools in Adult Symptomatic Lumbar Scoliosis

STUDY DESIGN

A post hoc analysis.

OBJECTIVE

Advances in machine learning (ML) have led to tools offering individualized outcome predictions for adult spinal deformity (ASD). Our objective is to examine the properties of these ASD models in a cohort of adult symptomatic lumbar scoliosis (ASLS) patients.

SUMMARY OF BACKGROUND DATA

ML algorithms produce patient-specific probabilities of outcomes, including major complication (MC), reoperation (RO), and readmission (RA) in ASD. External validation of these models is needed.

METHODS

Thirty-nine predictive factors (12 demographic, 9 radiographic, 4 health-related quality of life, 14 surgical) were retrieved and entered into web-based prediction models for MC, unplanned RO, and hospital RA. Calculated probabilities were compared with actual event rates. Discrimination and calibration were analyzed using receiver operative characteristic area under the curve (where 0.5=chance, 1=perfect) and calibration curves (Brier scores, where 0.25=chance, 0=perfect). Ninety-five percent confidence intervals are reported.

RESULTS

A total of 169 of 187 (90%) surgical patients completed 2-year follow up. The observed rate of MCs was 41.4% with model predictions ranging from 13% to 68% (mean: 38.7%). RO was 20.7% with model predictions ranging from 9% to 54% (mean: 30.1%). Hospital RA was 17.2% with model predictions ranging from 13% to 50% (mean: 28.5%). Model classification for all three outcome measures was better than chance for all [area under the curve=MC 0.6 (0.5-0.7), RA 0.6 (0.5-0.7), RO 0.6 (0.5-0.7)]. Calibration was better than chance for all, though best for RA and RO (Brier Score=MC 0.22, RA 0.16, RO 0.17).

CONCLUSIONS

ASD prediction models for MC, RA, and RO performed better than chance in a cohort of adult lumbar scoliosis patients, though the homogeneity of ASLS affected calibration and accuracy. Optimization of models require samples with the breadth of outcomes (0%-100%), supporting the need for continued data collection as personalized prediction models may improve decision-making for the patient and surgeon alike.

Patient satisfaction after multiple revision surgeries for adult spinal deformity

OBJECTIVE

Revision surgery is often necessary for adult spinal deformity (ASD) patients. Satisfaction with management is an important component of health-related quality of life. The authors hypothesized that patients who underwent multiple revision surgeries following ASD correction would exhibit lower self-reported satisfaction scores.

METHODS

This was a retrospective cohort study of 668 patients who underwent ASD surgery and were eligible for a minimum 2-year follow-up. Visits were stratified by occurrence prior to the index surgery (period 0), after the index surgery only (period 1), after the first revision only (period 2), and after the second revision only (period 3). Patients were further stratified by prior spine surgery before their index surgery. Scoliosis Research Society-22 (SRS-22r) health-related quality-of-life satisfaction subscore and total satisfaction scores were evaluated at all periods using multiple linear regression and adjustment for age, sex, and Charlson Comorbidity Index.

RESULTS

In total, 46.6% of the study patients had undergone prior spine surgery before their index surgery. The overall revision rate was 21.3%. Among patients with no spine surgery prior to the index surgery, SRS-22r satisfaction scores increased from period 0 to 1 (from 2.8 to 4.3, p < 0.0001), decreased after one revision from period 1 to 2 (4.3 to 3.9, p = 0.0004), and decreased further after a second revision from period 2 to 3 (3.9 to 3.3, p = 0.0437). Among patients with spine surgery prior to the index procedure, SRS-22r satisfaction increased from period 0 to 1 (2.8 to 4.2, p < 0.0001) and decreased from period 1 to 2 (4.2 to 3.8, p = 0.0011). No differences in follow-up time from last surgery were observed (all p > 0.3). Among patients with multiple revisions, 40% experienced rod fracture, 40% proximal junctional kyphosis, and 33% pseudarthrosis.

CONCLUSIONS

Among patients undergoing ASD surgery, revision surgery is associated with decreased satisfaction, and multiple revisions are associated with additive detriment to satisfaction among patients initially undergoing primary surgery. These findings have direct implications for preoperative patient counseling and establishment of postoperative expectations.

Comparable satisfaction and clinical outcomes after surgery for adolescent idiopathic scoliosis in the adult (AISA) between the US and Japan

BACKGROUND

The impact of ethnicity on the surgery outcomes of adolescent idiopathic scoliosis in the adult (AISA) is poorly understood. This study aimed to compare the surgery outcomes for AISA between the United States (US) and Japan (JP).

METHODS

171 surgically treated AISA (20-40y) were consecutively collected from 2 separate multicenter databases. Patients were propensity-score matched for age, gender, curve type, levels fused, and 2y postop spinal alignment. Demographic and radiographic parameters were compared between the US and JP at baseline and 2y post-op.

RESULTS

A total of 108 patients were matched by propensity score (age; US vs. JP: 29 ± 6 vs. 29 ± 7y, females: 76 vs. 76%, curve type [Schwab-SRS TypeT; TypeD; TypeL; TypeN]: 35; 35; 30; 0 vs. 37; 33; 30; 0%)] levels fused: 10 ± 4 vs. 10 ± 4, 2y thoracic curve:17 ± 13 vs. 17 ± 12°, 2y CSVL: 10 ± 8 vs. 11 ± 9 mm). Similar clinical improvement was achieved between US and JP (function; 4.2 ± 0.9 vs 4.3 ± 0.6, p = 0.60, pain; 3.8 ± 0.9 vs 4.1 ± 0.8, p = 0.13, satisfaction; 4.3 ± 0.9 vs 4.2 ± 0.7, p = 0.61, total; 4.0 ± 0.8 vs 4.1 ± 0.5, p = 0.60). The correlation analyzes indicated that postoperative SRS-22 subdomains correlated differently with satisfaction (all subdomains moderately correlated with satisfaction in the US while only pain and mental health correlated moderately with satisfaction in JP ([function: r = 0.61 vs 0.29, pain: r = . 72 vs 0.54, self-image: r = 0.72 vs 0.37, mental health: r = 0.64 vs 0.55]).

CONCLUSIONS

Surgery for AISA was similarly effective in the US and JP. Satisfaction for spinal surgery among patients in different countries may not be different unless the procedure limits an individual's unique lifestyle that the patient expected to resume.

Changes in health-related quality of life measures associated with degree of proximal junctional kyphosis

PURPOSE

To explore the changes in health-related quality of life parameters observed in patients experiencing varying degrees of proximal junctional kyphosis following corrective adult spinal deformity fusions.

METHODS

Inclusion: adult spinal deformity patients > 18 y/o, undergoing spinal fusion. PJK: ≥ 10° measure of the sagittal Cobb angle between the inferior endplate of the UIV and the superior endplate of the UIV + 2. Severe PJK: > 28° PJK. Mild PJK: ≥ 10^oand ≤ 28°. ANOVA, followed by ANCOVA, compared the change in HRQoLs between time points (BL, 1Y, 2Y) among PJK groups. Correlation-related change in PJK and change in HRQoL for mild and severe groups.

RESULTS

969 patients (age: 64.5 y/o,75% F, posterior levels fused:12.3) were studied. 59% no PJK, 32% mild PJK, 9% severe PJK. No differences in HRQoLs were seen between no PJK and PJK groups at baseline, one year, and 2 years. Adjusted analysis revealed Severe PJK patients improved less in SRS-22 Satisfaction (NoPJK: 1.6, MildPJK: 1.6, SeverePJK: 1.0; p = 0.022) scores at 2 years. Linear regression analysis only found clinical improvement in SRS-22 Satisfaction to correlate with the change of the PJK angle by 2 years (R = 0.176, P = 0.008). No other HRQoL metric correlated with either the incidence of PJK or the change in the PJK angle by one or 2 years.

CONCLUSIONS

These results maintain that patients presenting with and without proximal junctional kyphosis report similar health-related qualities of life following corrective adult spinal deformity surgery, and SRS-22 Satisfaction may be a clinical correlate to the degree of PJK. Rather than proving proximal junctional kyphosis to have a minimal clinical impact overall on HRQoL metrics, these data suggest that future analysis of this phenomenon requires different assessments.

LEVEL OF EVIDENCE

Level of evidence: III.

Development of Risk Stratification Predictive Models for Cervical Deformity Surgery

BACKGROUND

As corrective surgery for cervical deformity (CD) increases, so does the rate of complications and reoperations. To minimize suboptimal postoperative outcomes, it is important to develop a tool that allows for proper preoperative risk stratification.

OBJECTIVE

To develop a prognostic utility for identification of risk factors that lead to the development of major complications and unplanned reoperations.

METHODS

CD patients age 18 years or older were stratified into 2 groups based on the postoperative occurrence of a revision and/or major complication. Multivariable logistic regressions identified characteristics that were associated with revision or major complication. Decision tree analysis established cutoffs for predictive variables. Models predicting both outcomes were quantified using area under the curve (AUC) and receiver operating curve characteristics.

RESULTS

A total of 109 patients with CD were included in this study. By 1 year postoperatively, 26 patients experienced a major complication and 17 patients underwent a revision. Predictive modeling incorporating preoperative and surgical factors identified development of a revision to include upper instrumented vertebrae > C5, lowermost instrumented vertebrae > T7, number of unfused lordotic cervical vertebrae > 1, baseline T1 slope > 25.3°, and number of vertebral levels in maximal kyphosis > 12 (AUC: 0.82). For developing a major complication, a model included a current smoking history, osteoporosis, upper instrumented vertebrae inclination angle < 0° or > 40°, anterior diskectomies > 3, and a posterior Smith Peterson osteotomy (AUC: 0.81).

CONCLUSION

Revisions were predicted using a predominance of radiographic parameters while the occurrence of major complications relied on baseline bone health, radiographic, and surgical characteristics.

Operative treatment outcomes for adult cervical deformity: a prospective multicenter assessment with mean 3-year follow-up

OBJECTIVE

Adult cervical deformity (ACD) has high complication rates due to surgical complexity and patient frailty. Very few studies have focused on longer-term outcomes of operative ACD treatment. The objective of this study was to assess minimum 2-year outcomes and complications of ACD surgery.

METHODS

A multicenter, prospective observational study was performed at 13 centers across the United States to evaluate surgical outcomes for ACD. Demographics, complications, radiographic parameters, and patient-reported outcome measures (PROMs; Neck Disability Index, modified Japanese Orthopaedic Association, EuroQol-5D [EQ-5D], and numeric rating scale [NRS] for neck and back pain) were evaluated, and analyses focused on patients with ≥ 2-year follow-up.

RESULTS

Of 169 patients with ACD who were eligible for the study, 102 (60.4%) had a minimum 2-year follow-up (mean 3.4 years, range 2-8.1 years). The mean age at surgery was 62 years (SD 11 years). Surgical approaches included anterior-only (22.8%), posterior-only (39.6%), and combined (37.6%). PROMs significantly improved from baseline to last follow-up, including Neck Disability Index (from 47.3 to 33.0) and modified Japanese Orthopaedic Association score (from 12.0 to 12.8; for patients with baseline score ≤ 14), neck pain NRS (from 6.8 to 3.8), back pain NRS (from 5.5 to 4.8), EQ-5D score (from 0.74 to 0.78), and EQ-5D visual analog scale score (from 59.5 to 66.6) (all p ≤ 0.04). More than half of the patients (n = 58, 56.9%) had at least one complication, with the most common complications including dysphagia, distal junctional kyphosis, instrumentation failure, and cardiopulmonary events. The patients who did not achieve 2-year follow-up (n = 67) were similar to study patients based on baseline demographics, comorbidities, and PROMs. Over the course of follow-up, 23 of the total 169 enrolled patients were reported to have died. Notably, these represent all-cause mortalities during the course of follow-up.

CONCLUSIONS

This multicenter, prospective analysis demonstrates that operative treatment for ACD provides significant improvement of health-related quality of life at a mean 3.4-year follow-up, despite high complication rates and a high rate of all-cause mortality that is reflective of the overall frailty of this patient population. To the authors' knowledge, this study represents the largest and most comprehensive prospective effort to date designed to assess the intermediate-term outcomes and complications of operative treatment for ACD.

Evolution of Adult Cervical Deformity Surgery Clinical and Radiographic Outcomes Based on a Multicenter Prospective Study: Are Behaviors and Outcomes Changing With Experience?

STUDY DESIGN

Retrospective cohort study.

OBJECTIVE

Assess changes in outcomes and surgical approaches for adult cervical deformity surgery over time.

SUMMARY OF BACKGROUND DATA

As the population ages and the prevalence of cervical deformity increases, corrective surgery has been increasingly seen as a viable treatment. Dramatic surgical advancements and expansion of knowledge on this procedure have transpired over the years, but the impact on cervical deformity surgery is unknown.

MATERIALS AND METHODS

Adult cervical deformity patients (18 yrs and above) with complete baseline and up to the two-year health-related quality of life and radiographic data were included. Descriptive analysis included demographics, radiographic, and surgical details. Patients were grouped into early (2013-2014) and late (2015-2017) by date of surgery. Univariate and multivariable regression analyses were used to assess differences in surgical, radiographic, and clinical outcomes over time.

RESULTS

A total of 119 cervical deformity patients met the inclusion criteria. Early group consisted of 72 patients, and late group consisted of 47. The late group had a higher Charlson Comorbidity Index (1.3 vs. 0.72), more cerebrovascular disease (6% vs. 0%, both P <0.05), and no difference in age, frailty, deformity, or cervical rigidity. Controlling for baseline deformity and age, late group underwent fewer three-column osteotomies [odds ratio (OR)=0.18, 95% confidence interval (CI): 0.06-0.76, P =0.014]. At the last follow-up, late group had less patients with: a moderate/high Ames horizontal modifier (71.7% vs. 88.2%), and overcorrection in pelvic tilt (4.3% vs. 18.1%, both P <0.05). Controlling for baseline deformity, age, levels fused, and three-column osteotomies, late group experienced fewer adverse events (OR=0.15, 95% CI: 0.28-0.8, P =0.03), and neurological complications (OR=0.1, 95% CI: 0.012-0.87, P =0.03).

CONCLUSION

Despite a population with greater comorbidity and associated risk, outcomes remained consistent between early and later time periods, indicating general improvements in care. The later cohort demonstrated fewer three-column osteotomies, less suboptimal realignments, and concomitant reductions in adverse events and neurological complications. This may suggest a greater facility with less invasive techniques.

Three-Column Osteotomy in Adult Spinal Deformity: An Analysis of Temporal Trends in Usage and Outcomes

BACKGROUND

Three-column osteotomies (3COs), usually in the form of pedicle subtraction or vertebral column resection, have become common in adult spinal deformity surgery. Although a powerful tool for deformity correction, 3COs can increase the risks of perioperative morbidity.

METHODS

Operative patients with adult spinal deformity (Cobb angle of >20°, sagittal vertical axis [SVA] of >5 cm, pelvic tilt of >25°, and/or thoracic kyphosis of >60°) with available baseline and 2-year radiographic and health-related quality-of-life (HRQoL) data were included. Patients were stratified into 2 groups by surgical year: Group I (2008 to 2013) and Group II (2014 to 2018). Patients with 3COs were then isolated for outcomes analysis. Severe sagittal deformity was defined by an SVA of >9.5 cm. Best clinical outcome (BCO) was defined as an Oswestry Disability Index (ODI) of <15 and Scoliosis Research Society (SRS)-22 of >4.5. Multivariable regression analyses were used to assess differences in surgical, radiographic, and clinical parameters.

RESULTS

Seven hundred and fifty-two patients with adult spinal deformity met the inclusion criteria, and 138 patients underwent a 3CO. Controlling for baseline SVA, PI-LL (pelvic incidence minus lumbar lordosis), revision status, age, and Charlson Comorbidity Index (CCI), Group II was less likely than Group I to have a 3CO (21% versus 31%; odds ratio [OR] = 0.6; 95% confidence interval [CI] = 0.4 to 0.97) and more likely to have an anterior lumbar interbody fusion (ALIF; OR = 1.6; 95% CI = 1.3 to 2.3) and a lateral lumbar interbody fusion (LLIF; OR = 3.8; 95% CI = 2.3 to 6.2). Adjusted analyses showed that Group II had a higher likelihood of supplemental rod usage (OR = 21.8; 95% CI = 7.8 to 61) and a lower likelihood of proximal junctional failure (PJF; OR = 0.23; 95% CI = 0.07 to 0.76) and overall hardware complications by 2 years (OR = 0.28; 95% CI = 0.1 to 0.8). In an adjusted analysis, Group II had a higher likelihood of titanium rod usage (OR = 2.7; 95% CI = 1.03 to 7.2). Group II had a lower 2-year ODI and higher scores on Short Form (SF)-36 components and SRS-22 total (p < 0.05 for all). Controlling for baseline ODI, Group II was more likely to reach the BCO for the ODI (OR = 2.8; 95% CI = 1.2 to 6.4) and the SRS-22 total score (OR = 4.6; 95% CI = 1.3 to 16).

CONCLUSIONS

Over a 10-year period, the rates of 3CO usage declined, including in cases of severe deformity, with an increase in the usage of PJF prophylaxis. A better understanding of the utility of 3CO, along with a greater implementation of preventive measures, has led to a decrease in complications and PJF and a significant improvement in patient-reported outcome measures.

LEVEL OF EVIDENCE

Therapeutic Level III . See Instructions for Authors for a complete description of levels of evidence.

Proximal and distal reciprocal changes following cervical deformity malalignment correction

OBJECTIVE

Hyperextension of C0-2 is a debilitating compensatory mechanism used to maintain horizontal gaze, analogous to high pelvic tilt in the lumbopelvic complex to maintain an upright posture. This study aims to investigate the impact of cervical deformity (CD) correction on this hyperextension. The authors hypothesize that correction of cervical sagittal malalignment allows for relaxation of C0-2 hyperextension and improved clinical outcomes.

METHODS

A retrospective review was conducted of a multicenter database of patients with CD undergoing spinal realignment and fusion caudal to C2 and cephalad to the pelvis. Range of motion (ROM) and reserve of extension (ROE) were calculated across C2-7 and C0-2. The association between C2-7 correction and change in C0-2 ROE was investigated while controlling for horizontal gaze, followed by stratification into ΔC2-7 percentiles.

RESULTS

Sixty-five patients were included (mean age 61.8 ± 9.6 years, 68% female). At baseline, patients had cervical kyphosis (C2-7, -11.7° ± 18.2°; T1 slope-cervical lordosis mismatch, 38.6° ± 18.6°), negative global alignment (sagittal vertical axis [SVA] -12.8 ± 71.2 mm), and hyperlordosis at C0-2 (mean 33.2° ± 11.8°). The mean ROM was 25.7° ± 17.7° and 21.3° ± 9.9° at C2-7 and C0-2, respectively, with an ROE of approximately 9° for each segment. Limited C0-2 ROM and ROE correlated with the Neck Disability Index (r = -0.371 and -0.394, p < 0.01). The mean number of levels fused was 7.0 ± 3.1 (24.6% anterior, 43.1% posterior), with 87.7% undergoing at least an osteotomy. At 1 year, mean C2-7 increased to 5.5° ± 13.4°, SVA became neutral (11.5 ± 54.8 mm), C0-2 hyperlordosis decreased to 27.8° ± 11.7°, and thoracic kyphosis (TK) increased to -49.4° ± 18.1° (all p < 0.001). Concurrently, mean C0-2 ROM increased to 27.6° ± 8.1° and C2-7 ROM decreased significantly to 9.0° ± 12.3° without a change in ROE. Controlling for horizontal gaze, change in C2-7 lordosis significantly correlated with increased TK (r = -0.617, p < 0.001), decreased C0-2 (r = -0.747, p < 0.001), and increased C0-2 ROE (r = 0.550, p = 0.002).

CONCLUSIONS

CD correction can significantly impact cephalad and caudal compensation in the upper cervical and thoracic spine. Restoration of cervical alignment resulted in increased C0-2 ROE and TK and was also associated with improved clinical outcome.

Outcomes of operative treatment for adult spinal deformity: a prospective multicenter assessment with mean 4-year follow-up

OBJECTIVE

The current literature has primarily focused on the 2-year outcomes of operative adult spinal deformity (ASD) treatment. Longer term durability is important given the invasiveness, complications, and costs of these procedures. The aim of this study was to assess minimum 3-year outcomes and complications of ASD surgery.

METHODS

Operatively treated ASD patients were assessed at baseline, follow-up, and through mailings. Patient-reported outcome measures (PROMs) included scores on the Oswestry Disability Index (ODI), Scoliosis Research Society-22r (SRS-22r) questionnaire, mental component summary (MCS) and physical component summary (PCS) of the SF-36, and numeric rating scale (NRS) for back and leg pain. Complications were classified as perioperative (≤ 90 days), delayed (90 days to 2 years), and long term (≥ 2 years). Analyses focused on patients with minimum 3-year follow-up.

RESULTS

Of 569 patients, 427 (75%) with minimum 3-year follow-up (mean ± SD [range] 4.1 ± 1.1 [3.0-9.6] years) had a mean age of 60.8 years and 75% were women. Operative treatment included a posterior approach for 426 patients (99%), with a mean ± SD 12 ± 4 fusion levels. Anterior lumbar interbody fusion was performed in 35 (8%) patients, and 89 (21%) underwent 3-column osteotomy. All PROMs improved significantly from baseline to last follow-up, including scores on ODI (45.4 to 30.5), PCS (31.0 to 38.5), MCS (45.3 to 50.6), SRS-22r total (2.7 to 3.6), SRS-22r activity (2.8 to 3.5), SRS-22r pain (2.3 to 3.4), SRS-22r appearance (2.4 to 3.5), SRS-22r mental (3.4 to 3.7), SRS-22r satisfaction (2.7 to 4.1), NRS for back pain (7.1 to 3.8), and NRS for leg pain (4.8 to 3.0) (all p < 0.001). Degradations in some outcome measures were observed between the 2-year and last follow-up evaluations, but the magnitudes of these degradations were modest and arguably not clinically significant. Overall, 277 (65%) patients had at least 1 complication, including 185 (43%) perioperative, 118 (27%) delayed, and 56 (13%) long term. Notably, the 142 patients who did not achieve 3-year follow-up were similar to the study patients in terms of demographic characteristics, deformities, and baseline PROMs and had similar rates and types of complications.

CONCLUSIONS

This prospective multicenter analysis demonstrated that operative ASD treatment provided significant improvement of health-related quality of life at minimum 3-year follow-up (mean 4.1 years), suggesting that the benefits of surgery for ASD remain durable at longer follow-up. These findings should prove useful for counseling, cost-effectiveness assessments, and efforts to improve the safety of care.

Surgical Planning for Adult Spinal Deformity: Anticipated Sagittal Alignment Corrections According to the Surgical Level

STUDY DESIGN

Retrospective cohort study.

OBJECTIVES

Establish simultaneous focal and regional corrective guidelines accounting for reciprocal global and pelvic compensation.

METHODS

433 ASD patients (mean age 62.9 yrs, 81.3% F) who underwent corrective realignment (minimum L1-pelvis) were included. Sagittal parameters, and segmental and regional Cobb angles were assessed pre and post-op. Virtual postoperative alignment was generated by combining post-op alignment of the fused spine with the pre-op alignment on the unfused thoracic kyphosis and the pre-op pelvic retroversion. Regression models were then generated to predict the relative impact of segmental (L4-L5) and regional (L1-L4) corrections on PT, SVA (virtual), and TPA.

RESULTS

Baseline analysis revealed distal (L4-S1) lordosis of 33 ± 15°, flat proximal (L1-L4) lordosis (1.7 ± 17°), and segmental kyphosis from L2-L3 to T10-T11. Post-op, there was no mean change in distal lordosis (L5-S1 decreased by 2°, and L4-L5 increased by 2°), while the more proximal lordosis increased by 18 ± 16°. Regression formulas revealed that Δ10° in distal lordosis resulted in Δ10° in TPA, associated with Δ100 mm in SVA or Δ3° in PT; Δ10° in proximal lordosis yielded Δ5° in TPA associated with Δ50 mm in SVA; and finally Δ10° in thoraco-lumbar junction yielded Δ2.5° in TPA associated with Δ25 mm in SVA and no impact on PT correction.

CONCLUSIONS

Overall impact of lumbar lordosis restoration is critically determined by location of correction. Distal correction leads to a greater impact on global alignment and pelvic retroversion. More specifically, it can be assumed that 1° L4-S1 lordosis correction produces 1° change in TPA / 10 mm change in SVA and 0.5° in PT.

Patient-Reported Outcomes After Complex Adult Spinal Deformity Surgery: 5-Year Results of the Scoli-Risk-1 Study

STUDY DESIGN

Prospective cohort.

OBJECTIVE

To prospectively evaluate PROs up to 5-years after complex ASD surgery.

METHODS

The Scoli-RISK-1 study enrolled 272 ASD patients undergoing surgery from 15 centers. Inclusion criteria was Cobb angle of >80°, corrective osteotomy for congenital or revision deformity, and/or 3-column osteotomy. The following PROs were measured prospectively at intervals up to 5-years postoperative: ODI, SF36-PCS/MCS, SRS-22, NRS back/leg. Among patients with 5-year follow-up, comparisons were made from both baseline and 2-years postoperative to 5-years postoperative. PROs were analyzed using mixed models for repeated measures.

RESULTS

Seventy-seven patients (28.3%) had 5-year follow-up data. Comparing baseline to 5-year data among these 77 patients, significant improvement was seen in all PROs: ODI (45.2 vs. 29.3, P < 0.001), SF36-PCS (31.5 vs. 38.8, P < 0.001), SF36-MCS (44.9 vs. 49.1, P = 0.009), SRS-22-total (2.78 vs. 3.61, P < 0.001), NRS-back pain (5.70 vs. 2.95, P < 0.001) and NRS leg pain (3.64 vs. 2.62, P = 0.017). In the 2 to 5-year follow-up period, no significant changes were seen in any PROs. The percentage of patients achieving MCID from baseline to 5-years were: ODI (62.0%) and the SRS-22r domains of function (70.4%), pain (63.0%), mental health (37.5%), self-image (60.3%), and total (60.3%). Surprisingly, mean values (P > 0.05) and proportion achieving MCID did not differ significantly in patients with major surgery-related complications compared to those without.

CONCLUSIONS

After complex ASD surgery, significant improvement in PROs were seen at 5-years postoperative in ODI, SF36-PCS/MCS, SRS-22r, and NRS-back/leg pain. No significant changes in PROs occurred during the 2 to 5-year postoperative period. Those with major surgery-related complications had similar PROs and proportion of patients achieving MCID as those without these complications.

Pelvic Non-Response Following Treatment of Adult Spinal Deformity: Influence of Realignment Strategies on Occurrence

PURPOSE

Despite adequate correction, the pelvis may fail to readjust, deemed pelvic non-response. To assess alignment outcomes(pelvic non-response[PNR], PJK, postoperative cervical deformity[CD]) following ASD surgery utilizing different realignment strategies.

METHODS

ASD patients with 2-year(2Y) data were included. PNR defined as undercorrected in age-adjusted pelvic tilt(PT) at 6W and maintained at 2Y. Patients classified by alignment utilities: [a] Improvement in SRS-Schwab SVA,[b] Matching in age-adjusted PI-LL,[c] Matching in Roussouly,[d] aligning Global Alignment and Proportionality(GAP) score. Multivariable regression analyses, controlling for age, baseline deformity, and surgical factors, assessed rates of PNR, PJK, and CD development following realignment.

RESULTS

686 patients met inclusion criteria. Rates of postop PJK and CD were not significant in the PNR group(both P>0.15). PNR patients less often met substantial clinical benefit in ODI by 2Y(OR: 0.6,[0.4-0.98]). Patients overcorrected in age-adjusted PI-LL, matching Roussouly, or proportioned in GAP at 6W had lower rates of PNR(all P<0.001). Incremental addition of classifications led to 0% occurrence of PNR, PJK and CD. Stratifying by baseline PT severity, Low and Moderate deformity demonstrated the least incidence of PNR(7.7%) when proportioning in GAP at 6W, while severe PT benefited most from matching in Roussouly(all P<0.05).

CONCLUSIONS

Following ASD corrective surgery, 24.9% of patients showed residual pelvic malalignment. This occurrence was often accompanied by undercorrection of lumbopelvic mismatch and less improvement of pain. However, overcorrection in any strategy incurred higher rates of PJK. We recommend surgeons identify a middle ground using one, or more, of the available classifications to inform correction goals in this regard.

LEVEL OF EVIDENCE

III.

Highest Achievable Outcomes for Patients Undergoing Cervical Deformity Corrective Surgery by Frailty

BACKGROUND

Frailty is influential in determining operative outcomes, including complications, in patients with cervical deformity (CD).

OBJECTIVE

To assess whether frailty status limits the highest achievable outcomes of patients with CD.

METHODS

Adult patients with CD with 2-year (2Y) data included. Frailty stratification: not frail (NF) <0.2, frail (F) 0.2 to 0.4, and severely frail (SF) >0.4. Analysis of covariance established estimated marginal means based on age, invasiveness, and baseline deformity, for improvement, deterioration, or maintenance in Neck Disability Index (NDI), Modified Japanese Orthopaedic Association (mJOA), and Numerical Rating Scale Neck Pain.

RESULTS

One hundred twenty-six patients with CD included 29 NF, 83 F, and 14 SF. The NF group had the highest rates of deterioration and lowest rates of improvement in cervical Sagittal Vertical Axis and horizontal gaze modifiers. Two-year improvements in NDI by frailty: NF: -11.2, F: -16.9, and SF: -14.6 (P = .524). The top quartile of NF patients also had the lowest 1-year (1Y) NDI (7.0) compared with F (11.0) and SF (40.5). Between 1Y and 2Y, 7.9% of patients deteriorated in NDI, 71.1% maintained, and 21.1% improved. Between 1Y and 2Y, SF had the highest rate of improvement (42%), while NF had the highest rate of deterioration (18.5%).

CONCLUSION

Although frail patients improved more often by 1Y, SF patients achieve most of their clinical improvement between 1 and 2Y. Frailty is associated with factors such as osteoporosis, poor alignment, neurological status, sarcopenia, and other medical comorbidities. Similarly, clinical outcomes can be affected by many factors (fusion status, number of pain generators within treated levels, integrity of soft tissues and bone, and deformity correction). Although accounting for such factors will ultimately determine whether frailty alone is an independent risk factor, these preliminary findings may suggest that frailty status affects the clinical outcomes and improvement after CD surgery.

Individual differences in postoperative recovery trajectories for adult symptomatic lumbar scoliosis

OBJECTIVE

The Adult Symptomatic Lumbar Scoliosis-1 (ASLS-1) trial demonstrated the benefit of adult symptomatic lumbar scoliosis (ASLS) surgery. However, the extent to which individuals differ in their postoperative recovery trajectories is unknown. This study's objective was to evaluate variability in and factors moderating recovery trajectories after ASLS surgery.

METHODS

The authors used longitudinal, multilevel models to analyze postoperative recovery trajectories following ASLS surgery. Study outcomes included the Oswestry Disability Index (ODI) score and Scoliosis Research Society-22 (SRS-22) subscore, which were measured every 3 months until 2 years postoperatively. The authors evaluated the influence of preoperative disability level, along with other potential trajectory moderators, including radiographic, comorbidity, pain/function, demographic, and surgical factors. The impact of different parameters was measured using the R2, which represented the amount of variability in ODI/SRS-22 explained by each model. The R2 ranged from 0 (no variability explained) to 1 (100% of variability explained).

RESULTS

Among 178 patients, there was substantial variability in recovery trajectories. Applying the average trajectory to each patient explained only 15% of the variability in ODI and 21% of the variability in SRS-22 subscore. Differences in preoperative disability (ODI/SRS-22) had the strongest influence on recovery trajectories, with patients having moderate disability experiencing the greatest and most rapid improvement after surgery. Reflecting this impact, accounting for the preoperative ODI/SRS-22 level explained an additional 56%-57% of variability in recovery trajectory, while differences in the rate of postoperative change explained another 7%-9%. Among the effect moderators tested, pain/function variables-such as visual analog scale back pain score-had the biggest impact, explaining 21%-25% of variability in trajectories. Radiographic parameters were the least influential, explaining only 3%-6% more variance than models with time alone. The authors identified several significant trajectory moderators in the final model, such as significant adverse events and the number of levels fused.

CONCLUSIONS

ASLS patients have highly variable postoperative recovery trajectories, although most reach steady state at 12 months. Preoperative disability was the most important influence, although other factors, such as number of levels fused, also impacted recovery.

Extending machine learning beyond interatomic potentials for predicting molecular properties

Machine learning (ML) is becoming a method of choice for modelling complex chemical processes and materials. ML provides a surrogate model trained on a reference dataset that can be used to establish a relationship between a molecular structure and its chemical properties. This Review highlights developments in the use of ML to evaluate chemical properties such as partial atomic charges, dipole moments, spin and electron densities, and chemical bonding, as well as to obtain a reduced quantum-mechanical description. We overview several modern neural network architectures, their predictive capabilities, generality and transferability, and illustrate their applicability to various chemical properties. We emphasize that learned molecular representations resemble quantum-mechanical analogues, demonstrating the ability of the models to capture the underlying physics. We also discuss how ML models can describe non-local quantum effects. Finally, we conclude by compiling a list of available ML toolboxes, summarizing the unresolved challenges and presenting an outlook for future development. The observed trends demonstrate that this field is evolving towards physics-based models augmented by ML, which is accompanied by the development of new methods and the rapid growth of user-friendly ML frameworks for chemistry.