Recursive Partitioning Research Articles

The role of health system penetration rate in estimating the prevalence of type 1 diabetes in children and adolescents using electronic health records.

Having sufficient population coverage from the electronic health records (EHRs)-connected health system is essential for building a comprehensive EHR-based diabetes surveillance system. This study aimed to establish an EHR-based type 1 diabetes (T1D) surveillance system for children and adolescents across racial and ethnic groups by identifying the minimum population coverage from EHR-connected health systems to accurately estimate T1D prevalence. We conducted a retrospective, cross-sectional analysis involving children and adolescents <20years old identified from the OneFlorida+ Clinical Research Network (2018-2020). T1D cases were identified using a previously validated computable phenotyping algorithm. The T1D prevalence for each ZIP Code Tabulation Area (ZCTA, 5 digits), defined as the number of T1D cases divided by the total number of residents in the corresponding ZCTA, was calculated. Population coverage for each ZCTA was measured using observed health system penetration rates (HSPR), which was calculated as the ratio of residents in the corresponding ZTCA and captured by OneFlorida+ to the overall population in the same ZCTA reported by the Census. We used a recursive partitioning algorithm to identify the minimum required observed HSPR to estimate T1D prevalence and compare our estimate with the reported T1D prevalence from the SEARCH study. Observed HSPRs of 55%, 55%, and 60% were identified as the minimum thresholds for the non-Hispanic White, non-Hispanic Black, and Hispanic populations. The estimated T1D prevalence for non-Hispanic White and non-Hispanic Black were 2.87 and 2.29 per 1000 youth, which are comparable to the reference study's estimation. The estimated prevalence of T1D for Hispanics (2.76 per 1000 youth) was higher than the reference study's estimation (1.48-1.64 per 1000 youth). The standardized T1D prevalence in the overall Florida population was 2.81 per 1000 youth in 2019. Our study provides a method to estimate T1D prevalence in children and adolescents using EHRs and reports the estimated HSPRs and prevalence of T1D for different race and ethnicity groups to facilitate EHR-based diabetes surveillance.

Efficient in-situ image and video compression through probabilistic image representation

Fast and effective image compression for multi-dimensional images has become increasingly important for efficient storage and transfer of massive amounts of high-resolution images and videos. In this paper, we present an efficient in-situ method for multi-dimensional image and video compression called Compression via Adaptive Recursive Partitioning (CARP). CARP uses an optimal permutation of the image pixels inferred from a Bayesian probabilistic model on recursive partitions of the image to reduce its effective dimensionality, achieving a parsimonious representation that preserves information. Furthermore, it adopts a multi-layer Bayesian hierarchical model to achieve in-situ compression along with self-tuning and regularization, with just one single parameter to be specified by the user to achieve the desired compression rate. The properties of our proposed method include high reconstruction quality at a wide range of compression rates while preserving key local details, applicability to a variety of different image/video types and of different dimensions, computational scalability, progressive transmission and ease of tuning. Extensive numerical experiments using a variety of datasets including 2D still images, real-life YouTube videos, and surveillance videos show that CARP compares favorably to—and often uniformly outperforms—a wide range of popular image/video compression approaches, including JPEG, JPEG2000, AVI, BPG, MPEG4, HEVC, AV1, and three neural network-based methods.

Clinical-genomic determinants of immune checkpoint blockade response in head and neck squamous cell carcinoma.

BACKGROUNDRecurrent and/or metastatic (R/M) head and neck squamous cell carcinoma (HNSCC) is generally an incurable disease, with patients experiencing median survival of under 10 months and significant morbidity. While immune checkpoint blockade (ICB) drugs are effective in approximately 20% of patients, the remaining experience limited clinical benefit and are exposed to potential adverse effects and financial costs. Clinically approved biomarkers, such as tumor mutational burden (TMB), have a modest predictive value in HNSCC.METHODSWe analyzed clinical and genomic features, generated using whole-exome sequencing, in 133 ICB-treated patients with R/M HNSCC, of whom 69 had virus-associated and 64 had non-virus-associated tumors.RESULTSHierarchical clustering of genomic data revealed 6 molecular subtypes characterized by a wide range of objective response rates and survival after ICB therapy. The prognostic importance of these 6 subtypes was validated in an external cohort. A random forest-based predictive model, using several clinical and genomic features, predicted progression-free survival (PFS), overall survival (OS), and response with greater accuracy than did a model based on TMB alone. Recursive partitioning analysis identified 3 features (systemic inflammatory response index, TMB, and smoking signature) that classified patients into risk groups with accurate discrimination of PFS and OS.CONCLUSIONThese findings shed light on the immunogenomic characteristics of HNSCC tumors that drive differential responses to ICB and identify a clinical-genomic classifier that outperformed the current clinically approved biomarker of TMB. This validated predictive tool may help with clinical risk stratification in patients with R/M HNSCC for whom ICB is being considered.FUNDINGFundación Alfonso Martín Escudero, NIH R01 DE027738, US Department of Defense CA210784, The Geoffrey Beene Cancer Research Center, The MSKCC Population Science Research Program, the Jayme Flowers Fund, the Sebastian Nativo Fund, and the NIH/NCI Cancer Center Support Grant P30 CA008748.

Multi-Institutional Validation of the Recursive Partitioning Analysis for Overall Survival in Patients Undergoing Spine Radiosurgery for Spine Metastasis

The recently published spine radiosurgery (sSRS) recursive partitioning analysis (RPA) for overall survival (OS) separated patients into 3 distinct prognostic groups. We sought to externally validate this RPA using 3 separate multi-institutional datasets. A total of 444 patients were utilized to develop the recently published sSRS RPA predictive of OS in patients with spine metastases. The RPA identified three distinct prognostic classes. RPA Class 1 was defined as KPS >70 and controlled systemic disease (n = 142); RPA Class 2 was defined as KPS>70 with uncontrolled systemic disease or KPS ≤70, age ≥54 and absence of visceral metastases (n = 207); RPA Class 3 was defined as KPS ≤70 and age <54 years or KPS≤70, age ≥54 years and presence of visceral metastases (n = 95). We utilized data from 3 large tertiary care centers to independently validate this RPA. Data from each institution was utilized independently to validate the RPA to minimize confounding based on institutional differences in patient selection. A total of 1,184 patients (221 patients from institution A, 749 institution B, and 214 from institution C) were in the validation cohort and were divided based on their RPA Class. Kaplan-Meier method was used to estimate OS and log-rank test was used to compare OS between RPA classes. In each of the validation cohorts, the median OS was 19.9 months (institution A), 11.0 months (institution B), and 24.4 months (institution C). The patient distribution into RPA classification based on Institution A/B/C was, Class 1 (19.4%, 15.1%, 50.5%), Class 2 (74.7%, 57.7%%, 37.9%), and Class 3 (5.9%, 27.2%%, 11.2%), respectively. The median OS for patients in the validation cohort at Institution A/B/C based on RPA class was Class 1 (54 months, 27.1 months, 50.0 months), Class 2 (15.9 months, 13.0 months, 15.1 months) and Class 3 (6.9 months, 3.5 months, 6.1 months), respectively. Patients in RPA Class 1 had a significantly better OS compared to those in Class 2 of the each of the three external institution validation cohorts (p<0.01). Similarly, patients in RPA Class 2 had a significantly better OS compared to those in Class 3 (p<0.01). The external datasets from three large institutions independently validated the spine SRS RPA successfully for OS in patients undergoing sSRS for spinal metastases. This is the first RPA for OS to have been externally validated using multiple large datasets. Based on this validation, upfront spine SRS is strongly supported for patients in RPA Class 1 and Class 2 and is also cost effective with median OS >11 months for these patients. Patients in RPA Class 3 would benefit most from upfront conventional radiotherapy given their poor expected survival. Given successful external validation, this RPA helps guide physicians to identify those patients with spinal metastases who most benefit from sSRS.

Development of an RPA for Prediction of Radiation Necrosis Following Single Fraction Gamma Knife Radiosurgery for Brain Metastases

Radiation necrosis (RN) is a potential complication following treatment of brain metastases with stereotactic radiosurgery (SRS). Several risk factors for RN have been reported, but to our knowledge there are no recursive partitioning analysis (RPA) models to identify patients at highest risk for RN. We therefore sought to develop a predictive tool to identify patients at highest risk for the development of RN following single fraction SRS. Patients who underwent single fraction SRS for brain metastases from 2017-2021 were identified from a single institutional IRB-approved database. Patients with concern for RN were discussed in a multi-disciplinary setting and a diagnosis of RN was made based on pathologic or radiographic findings. Cox proportional hazards regression was done to identify factors associated with RN. RPA was performed to categorize patients into distinct risk groups for RN. Variables with p<0.1 on univariate analysis from the Cox regression analysis were included in the RPA. Patients with staged SRS, incomplete treatment records, or < 3 months radiographic follow-up were excluded from the analysis. The study population comprised 1,011 lesions from 283 patients with a median follow-up of 9.7 months. The majority of lesions had non-small cell lung cancer (NSCLC) (49%) as the primary site followed by breast (12%) and melanoma (11%). The median prescription dose was 24 Gy (range: 12-24 Gy). RN was diagnosed in 12.2% of lesions, of which 28% (35/123) were symptomatic RN. The median time to RN was 4.9 months. Variables identified for inclusion in the RPA included primary tumor site, use of targeted therapy, tumor location, pre-SRS hemorrhage, post-SRS hemorrhage, prior SRS to other lesions, number of SRS targets, maximum dose, maximum lesion diameter, 70% isodose line, heterogeneity index, conformality index, and gradient index. RPA identified four distinct groups. Group 1 was defined as maximum lesion diameter (MLD) <0.8 cm with primary tumor site other than breast, colorectal (CRC) or NSCLC (n = 174); group 2 was MLD <0.8 cm with breast, CRC, or NSCLC (n = 372). Group 3 was defined as MLD ≥ 0.8 cm without post-SRS hemorrhage (n = 336) and group 4 was MLD ≥0.8 cm with post-SRS hemorrhage (n = 129). Two-year RN free survival for all lesions was 82%, 100% for group 1, 89% for group 2, 76% for group 3, and 58% for group 4. We created the first RPA predictive model for RN following single fraction SRS and identified a subgroup of patients at highest risk. This RPA can help guide clinicians when educating patients on RN risk for brain metastases.

Prognostic Value of Lymph Node-to-Primary Tumor Ratio of PET Standardized Uptake Value for Nasopharyngeal Carcinoma: A Recursive Partitioning Risk Stratification Analysis

To evaluate the prognostic value of the lymph node-to-primary tumor ratio (NTR) of positron emission tomography (PET) standardized uptake value (SUV) for nasopharyngeal carcinoma (NPC) patients treated with induction chemotherapy (IC). Four hundred and sixty-seven locoregionally advanced NPC patients with pretreatment 18F-fluorodeoxyglucose (18F-FDG) PET/computed tomography (CT) scans between September 2017 and November 2020 were retrospectively reviewed. All patients underwent IC plus intensity-modulated radiotherapy (IMRT). The receiver operating characteristic (ROC) analysis was used to determine the optimal cut-off value of SUV NTR. Kaplan-Meier method was used to evaluate survival rates. The recursive partitioning analysis (RPA) was performed to construct a risk stratification model. The optimal cut-off value of SUV NTR was 0.74. Multivariate analyses showed that SUV NTR and overall stage were independent predictors for distant metastasis-free survival (DMFS) and regional recurrent-free survival (RRFS). Therefore, an RPA model based on the endpoint of DMFS was generated and categorized the patients into three distinct risk groups: RPA I (low-risk: SUV NTR<7.4 and stage III), RPA II (medium-risk: SUV NTR<7.4 and stage IVa, or SUV NTR≥7.4 and stage III), and RPA III (high-risk: SUV NTR≥7.4 and stage IVa), with a 3-year DMFS of 98.9%, 93.4%, and 84.2%, respectively. ROC analysis showed that the RPA model had superior predictive efficacy than the SUV NTR or overall stage alone. SUV NTR was an independent prognosticator for distant metastasis and regional recurrence in locoregionally advanced NPC. The RPA risk stratification model base on SUV NTR provides improved DMFS and RRFS prediction over the 8th edition of the TNM staging system.

Defining the Oligometastatic State and Predicting Overall Survival in Uterine Cancers using a Recursive Partitioning Analysis: A National Cancer Database Analysis

Oligometastatic disease is an intermediate state between locoregional disease and widely metastatic disease. Studies have shown an overall survival advantage in oligometastatic patients treated aggressively in multiple cancers. However, few studies have explored this relationship in oligometastatic gynecological cancers. In this study, we sought to identify if an oligometastatic state with improved overall survival exists in uterine cancer and whether treatment patterns, or patient factors play a role in survival. The National Cancer Database was retrospectively analyzed in patients diagnosed with uterine cancer between 2004-2019. We excluded patients with non-metastatic disease at diagnosis, lack of metastatic sites listed, multiple primaries and missing survival data. Survival analyses were performed using Cox proportional hazards analyses and the Kaplan-Meier method and differences between groups were evaluated with log-rank tests. Recursive partitioning analysis (RPA) was performed using the rpart package in R. Amongst 641,276 women with uterine cancer, 17,343 were ultimately in our study population after exclusion who had a median age of 64 (range 18-90), median overall survival of 11.1 months, and a total of 24,337 metastases that represented the following metastatic sites: lung (35%), other (24%) (sites "other" than bone, brain liver or lung), liver (15%), bone (12%), distant lymph nodes (10%), brain (3%) and carcinomatosis (1%). Multivariate cox proportional hazards analysis identified the following variables as significantly associated with survival, which were then used for RPA: one metastatic site, treatment, grade, age, race, Charlson-Deyo score. The data was split 70/30% for training and validation, respectively. A survival model was built using RPA on the training data with 10-fold cross-validation and had accuracy of 76.25% (p = 3.393e-0.5). Survival prediction was performed on the internal validation data set with 76.69% accuracy (p = 0.004). RPA stratified 4 groups with the following median survivals: 5.9, 8.6, 29.5, and 43.8 months (log-rank test p<0.0001). The group with the best overall survival was defined by 1 metastatic site (carcinomatosis, distant lymph nodes and "other"), combination treatment with systemic therapy and radiation or surgery and low-grade histology. An oligometastatic state with involvement of one metastatic site does exist with improved survival in uterine cancer. The RPA defined groups may aid in identifying patients that are best suited for aggressive definitive combination therapy versus those that would be better suited for comfort care. Analysis is ongoing and will further delineate the relationship between specific treatment modalities and overall survival in patients with one metastatic site. This study may set the foundation for a prospective clinical trial.

Dynamic Contrast Enhanced MRI as a Biomarker of Tumor Response and Oncologic Outcomes in Head and Neck Cancer: Results of a Single Institution Prospective Imaging Study

We aim to determine the correlation between vascular parameters of Dynamic contrast enhanced (DCE) MRIs and tumor response and outcomes in head and neck (HNC) patients treated with definitive radiation therapy (RT). Eighty-two HNC patients are included in this prospective study in one institute. All patients had malignant head and neck neoplasm indicative of curative- intent treatment. Patients were imaged using MRIs pre-, mid-, and post-RT completion at 8-12 weeks. T2-weighted sequences were used for tumor contouring then it was co-registered to respective DCE images. The response to treatment was checked at mid-radiotherapy (mid-RT) and at the end of RT. Mid-RT MRI was co-registered to baseline images and the manually segmented baseline primary tumor regions of interest were propagated to mid-RT images. Quantitative maps (Ktrans, Kep, Ve and Vp) were generated with the extended Tofts pharmacokinetic models and were used for analysis. These vascular parameters were presented as a mean value and percentile using histogram analysis and the following parameters were extracted using an in-house programming environment script: mean, 5th, 10th, 20th, 30th, 40th, 50th (i.e., median), 60th, 70th, 80th, 90th, 95th percentile. The non-parametric Wilcoxon signed-rank test was used to assess the changes of mid-RT DCE parameters compared to baseline. Recursive partitioning analysis (RPA) was used to identify the delta DCE threshold associated with relapse. We assessed the identified thresholds' correlation with oncological and survival endpoints using Cox regression with and without standard clinical variables. The median age for patients is 61 years old (33-78 range). Never smokers are 39 (47%), 35 (43%) are former smoker and 8 (10%) are current smoker with a mean value of 14 pack per year and 26 standard deviations. Using AJCC 8th edition, 39 (47%) are stage I and 19 (23%) are stage II and stage III and IV are 15 (18%) and 9 (10%) respectively. HPV positive are 72 (88%). For patients with GTV-P at baseline (n = 60), 11 (18%) had mid-RT CR at the primary site which increased to 50 (83%) post-RT. The LC and RFS for the entire cohort were 91.4%, and 79.2% respectively. In GTV-P, none of the pre-radiotherapy DCE parameters were correlated with LC or RFS. Wilcoxon signed rank test was statistically significant in 80, 90 and 95 percentiles with (p<0.05). RPA analysis identified different thresholds for each DCE parameter, and its inclusions to the multivariate model improved its performance. In GTV-P, RPA analysis identified ΔKtrans 40 percentiles >15.6% at mid-RT as the most significant point. When this value of ΔKtrans added to the multivariate analysis it was associated with a significantly better model performance in RFS (p = 0.00001). DCE parameters are a very promising tool to correlate with response and outcomes in H&N cancer patients. Future work is warranted for external validation of our findings.

Dawn of Staging for Head and Neck Soft Tissue Sarcoma: Validation of the Novel 8th Edition AJCC T Classification and Proposed Stage Groupings

After decades of stagnation, the 8th edition TNM (TNM8) introduced a new T classification for head and neck (HN) soft tissue sarcomas (STS). New size cutoffs of 2 and 4 cm define T1-3, and a novel T4 category is defined by local invasion of adjoining structures. These size cutoffs had been chosen arbitrarily to advance data collection in this unique disease site since literature showed approximately 70% of HN STS did not reach the previous size threshold (5 cm) for the existing T1 category. The definition of the TNM8 T categories also align with mucosal HN cancers. No stage grouping for HN STS was defined since this new classification required more data collection to derive stage groups. This study aims to validate the TNM8 T classification and to propose stage groupings. Clinical data of all adult (>16 years) HN STS patients treated from 1988 - 2019 with curative intent in our tertiary cancer center were retrieved from a prospective database, and supplemented with chart review. As per TNM8, cutaneous angiosarcoma, embryonal and alveolar rhabdomyosarcoma, Kaposi sarcoma, and dermatofibrosarcoma protuberans were excluded due to their different behavior. Multivariate analysis (MVA) identified prognostic factors for overall survival (OS). Adjusted hazard ratios (AHR) and recursive partitioning analysis (RPA) were used to derive stage groupings. Stage grouping performance for OS was assessed and also compared against the existing TNM8 groups for non-HN STS. A total of 221 patients (N1: 2; M1: 2) were included. Of the 219 M0 patients, 63% were males; median tumor size was 3.0 cm (range: 0.3-14.0); the proportion of TNM8 T1-T4 were 35%, 34%, 26%, and 5%, respectively. Median follow up was 5.9 years. Five-year OS was 79%. MVA confirmed the prognostic value of T category (T4 HR 7.73, 95% CI 3.62-16.5) and grade (G2/3 vs G1 HR 3.7, 95% CI 1.82-7.53), in addition to age (HR 1.03, 95% CI 1.01-1.04) (all p<0.001) for OS. AHR model derived T1-3_Grade 1 as stage 1; T1-3_Grade 2/3 as stage II; and T4_any Grade or any T_N1 as stage III (Table 1); the corresponding 5-year OS was 93%, 73%, and 38%, respectively. Both patients with M1 died within 1.5 years after diagnosis and M1 disease was designated stage IV. The AHR-grouping outperformed the RPA and non-HN TNM8 stage grouping for hazard consistency, hazard discrimination, percent variance explained, hazard difference, and sample size balance. The novel T4 category introduced in TNM8 is associated with a >7 fold increased risk of death. Grade continues to be a critical prognostic factor in HN STS. The TNM8 HN STS T classifications have been validated, and the proposed new stage groupings with TNM8 incorporating grade have excellent performance for OS.

Longitudinal Monitoring of Quantitative Imaging Kinetics of Primary Tumor and Nodal Volumes Using the MR-Linac Device in Head and Neck Cancer Patients

We aim to characterize the serial quantitative apparent diffusion coefficient (ADC) changes of the target disease volume using diffusion-weighted imaging (DWI) acquired weekly during radiation therapy (RT) on a 1.5T MR-Linac and correlate these changes with tumor response and oncologic outcomes for head and neck squamous cell carcinoma (HNSCC) patients. Thirty patients with pathologically confirmed HNSCC and received curative-intent RT at the University of Texas MD Anderson Cancer Center, were included in this prospective study. Baseline and weekly MRIs (weeks 1-6) were obtained, and various ADC parameters (mean, 5th, 10th, 20th, 30th, 40th, 50th, 60th, 70th, 80th, 90th and 95th percentile) were extracted from the target regions of interest (ROIs). Pre-RT and weekly ADC parameters were correlated with response during RT, loco-regional control, and the development of relapse using the Mann-Whitney U test. The Wilcoxon signed-rank test was used to compare the weekly ADC versus baseline values. Weekly volumetric changes (Δvolume) for each ROI were correlated with ΔADC using Spearman's Rho test. Recursive partitioning analysis (RPA) was performed to identify the optimal ΔADC threshold associated with different oncologic outcomes. There was an overall significant rise in all ADC parameters during different time points of RT compared to baseline values for both GTV-P & GTV-N. The increased ADC values for GTV-P were statistically significant only for primary tumors achieving CR during RT. RPA identified GTV-P ΔADC 5th percentile >13% at the 3rd week of RT as the most significant parameter associated with CR for GTV-P during RT (p <0.001). Baseline ADC parameters didn't significantly correlate with response to RT or other oncologic outcomes. There was a significant decrease in residual volume of both GTV-P & GTV-N throughout the course of RT. Additionally, a significant negative correlation between mean ΔADC and Δvolume for GTV-P at the 3rd and 4th week of RT was detected (r = -0.39, p = 0.044 & r = -0.45, p = 0.019, respectively). Assessment of ADC kinetics at regular intervals throughout RT is potentially able to predict the response to RT and oncologic outcome. Further studies with larger cohorts and multi-institutional data are needed for validation of our results.

Variation in global network properties across risk factors for adolescent internalizing symptoms: evidence of cumulative effects on structure and connectivity.

Identifying adolescents at risk of internalizing problems is a key priority. However, studies have tended to consider such problems in simple ways using diagnoses, or item summaries. Network theory and methods instead allow for more complex interaction between symptoms. Two key hypotheses predict differences in global network properties for those at risk: altered structure and increased connectivity. The current study evaluated these hypotheses for nine risk factors (e.g. income deprivation and low parent/carer support) individually and cumulatively in a large sample of 12-15 year-olds (N = 34 564). Recursive partitioning and bootstrapped networks were used to evaluate structural and connectivity differences. The pattern of network interactions was shown to be significantly different via recursive partitioning for all comparisons across risk-present/absent groups and levels of cumulative risk, except for income deprivation. However, the magnitude of differences appeared small. Most individual risk factors also showed relatively small effects for connectivity. Exceptions were noted for gender and sexual minority risk groups, as well as low parent/carer support, where larger effects were evident. A strong linear trend was observed between increasing cumulative risk exposure and connectivity. A robust approach to considering the effect of risk exposure on global network properties was demonstrated. Results are consistent with the ideas that pathological states are associated with higher connectivity, and that the number of risks, regardless of their nature, is important. Gender/sexual minority status and low parent/carer support had the biggest individual impacts on connectivity, suggesting these are particularly important for identification and prevention.

Phase II Randomized Trial of Single- vs. Two-Fraction Spine Stereotactic Radiosurgery for the Treatment of Vertebral Metastases

As systemic therapies improve significantly, more patients with limited metastatic disease are undergoing spine stereotactic radiosurgery (sSRS). High dose (≥24 Gy) single fraction sSRS has been associated with a vertebral compression fracture (VCF) risk of up to 40%. Comparatively, lower dose (16-18 Gy) single fraction sSRS is associated with a fracture risk of 15-20%, with the risk increasing as the dose increases. To mitigate the risk of VCF, while optimizing higher dose delivery, some have advocated utilizing two-fraction sSRS regimen. Therefore, we designed a phase II randomized trial in which we hypothesized that single fraction sSRS is non-inferior to two-fraction sSRS with respect to VCF but offers patients greater convenience. Inclusion criteria include age ≥18, Karnofsky performance score ≥70, vertebral metastasis from C3 to L5, maximum of three separate sites of metastases, limited paraspinal extension (<5 cm), and no rapid neurological decline. Patients must also be either Recursive Partitioning Analysis Class 1 (KPS >70 AND controlled systemic disease) or Class 2 (KPS >70, uncontrolled systemic disease OR KPS ≤70, age ≥54, no visceral metastases). Exclusion criteria include multiple primary cancers, primary neoplasm of the spine, prior surgery at the site of sSRS, spinal cord compression, bony retropulsion resulting in neurologic deficit, inability to undergo/contraindication to MRI, or diffuse multi-level metastatic spine disease. Our primary hypothesis is that single fraction sSRS (experimental arm - 16-18 Gy) is non-inferior to two fraction sSRS (standard arm - 24 Gy). Furthermore, we hypothesize that both treatment arms will have similar local control, pain control, quality of life and toxicity profiles. The primary endpoint of this trial is the development or progression of VCF at 6 months. Secondary endpoints include local control, pain control, quality of life and toxicity all of which will be assessed at 12 months. For the sample size calculation, we assumed a VCF risk of 17% in the experimental arm and a 7% risk in the standard arm. Based on these calculations, we aim to enroll 130 patients, 65 in each arm. This trial is currently enrolling patients actively, and approximately 30% of expected enrollment has been completed to date. This trial is registered on clinicaltrials.gov: NCT04218617. Pending full accrual. Pending full accrual.

Conditional Survival Estimates for Malignant Glioma Patients by RPA Class: Secondary Analysis of RTOG 9006

The RTOG recursive partitioning analysis (RPA) for malignant glioma patients is a key prognostic factor correlated with survival for patients with a malignant glioma, which was established using data from RTOG 9006. While such data is useful when counseling patients on survival outcomes at the time of diagnosis, the utility of such data diminishes over time. Conditional survival (CS) is a way to provide 'real-time' estimates of how survival changes over time. The primary objective of this secondary analysis is to report conditional survival estimates for patients with malignant gliomas enrolled on RTOG 9006 stratified by RPA class. RTOG 9006 was a phase III trial of patients diagnosed with malignant gliomas randomized to BCNU and hyper-fractionated radiation therapy (RT) vs. BCNU and conventionally fractionated RT that showed no significant differences in toxicities or oncologic outcomes. We performed a post-hoc analysis of RTOG 9006 using data obtained from the NCI NCTN Data Archive to evaluate conditional survival estimates for different RPA subgroups. Eligible patients included those with a known RPA class who were enrolled on RTOG 9006 (n = 632). Conditional survival estimates from the time of diagnosis and following 1-, 3-, and 5-y of survival were calculated using the Kaplan-Meier method. Cox proportional hazards modeling was performed to evaluate the prognostic significance of treatment arm, histology, extent of surgical resection, and RPA class following 1-, 3-, and 5-y of survival. For RPA Group I (n = 68), OS at 1-y from diagnosis was 95.58%. At 1- (n = 65), 3- (49) and 5-y (42) of survivorship, the chances of surviving an additional 1-y were 84.62, 91.84, 98.62% respectively (See Table 1). RPA class continued to be significantly correlated with survival following 1- and 3-y of survival irrespective of treatment arm. Treatment arm was not significantly correlated with conditional survival at any time point following treatment completion, and there was no interaction between treatment arm and RPA subgroups. Histology was significant at all time points. Conditional risk of death for patients treated for malignant gliomas in RPA Group I remains low at levels similar to those at initial diagnosis. Conditional risk continuously improves for RPA Groups IV-V as patients live longer. In most groups, the lowest survival odds were found in patients who have already survived 1 year.

Predicting Survival for Veno-Arterial ECMO Using Conditional Inference Trees-A Multicenter Study.

Despite increasing use and understanding of the process, veno-arterial extracorporeal membrane oxygenation (VA-ECMO) therapy is still associated with considerable mortality. Personalized and quick survival predictions using machine learning methods can assist in clinical decision making before ECMO insertion. This is a multicenter study to develop and validate an easy-to-use prognostic model to predict in-hospital mortality of VA-ECMO therapy, using unbiased recursive partitioning with conditional inference trees. We compared two sets with different numbers of variables (small and comprehensive), all of which were available just before ECMO initiation. The area under the curve (AUC), the cross-validated Brier score, and the error rate were applied to assess model performance. Data were collected retrospectively between 2007 and 2019. 837 patients were eligible for this study; 679 patients in the derivation cohort (median (IQR) age 60 (49 to 69) years; 187 (28%) female patients) and a total of 158 patients in two external validation cohorts (median (IQR) age 57 (49 to 65) and 70 (63 to 76) years). For the small data set, the model showed a cross-validated error rate of 35.79% and an AUC of 0.70 (95% confidence interval from 0.66 to 0.74). In the comprehensive data set, the error rate was the same with a value of 35.35%, with an AUC of 0.71 (95% confidence interval from 0.67 to 0.75). The mean Brier scores of the two models were 0.210 (small data set) and 0.211 (comprehensive data set). External validation showed an error rate of 43% and AUC of 0.60 (95% confidence interval from 0.52 to 0.69) using the small tree and an error rate of 35% with an AUC of 0.63 (95% confidence interval from 0.54 to 0.72) using the comprehensive tree. There were large differences between the two validation sets. Conditional inference trees are able to augment prognostic clinical decision making for patients undergoing ECMO treatment. They may provide a degree of accuracy in mortality prediction and prognostic stratification using readily available variables.

Development and validation of a novel TNM staging N-classification of oral cavity squamous cell carcinoma.

For oral cavity squamous cell carcinoma (OSCC), extent of extranodal extension (ENE) (minor, ≤2mm; major, >2mm) is differentially prognostic, whereas limitations exist with the 8th edition of American Joint Committee on Cancer/International Union Against Cancer TNM N-classification (TNM-8-N). Resected OSCC patients at four centers were included and extent of ENE was recorded. Thresholds for optimal overall survival (OS) discrimination of lymph node (LN) features were established. After dividing into training and validation sets, two new N-classifications were created using 1) recursive partitioning analysis (RPA), and 2) adjusted hazard ratios (aHRs) and were ranked against TNM-8-N and two published proposals. A total of 1460 patients were included (pN0: 696; pN+: 764). Of the pN+cases, 135 (18%) had bilateral/contralateral LNs; 126 (17%) and 244 (32%) had minor and major ENE, and two (0.3%) had LN(s)>6cm without ENE (N3a). LN number (1 and>1 vs. 0: aHRs, 1.92 [95% confidence interval (CI), 1.44-2.55] and 3.21 [95% CI, 2.44-4.22]), size (>3 vs. ≤3cm: aHR, 1.88 [95% CI, 1.44-2.45]), and ENE extent (major vs. minor: aHR, 1.40 [95% CI, 1.05-1.87]) were associated with OS, whereas presence of contralateral LNs was not (aHR, 1.05 [95% CI, 0.81-1.36]). The aHR proposal provided optimal performance with these changes to TNM-8-N: 1) stratification of ENE extent, 2) elimination of N2c and 6-cm threshold, and 3) stratification of N2b by 3cm threshold. A new N-classification improved staging performance compared to TNM-8-N, by stratifying by ENE extent, eliminating the old N2c category and the 6 cm threshold, and by stratifying multiple nodes by size.

Glasgow prognostic score for assessing the efficacy of whole-brain radiation therapy in cases of recursive partitioning analysis class 2 and class 3 multiple brain metastases: a retrospective study

PurposeWhole-brain radiotherapy (WBRT) may not be beneficial for patients with brain metastases (BMs). The Glasgow Prognostic Score (GPS) is a suggested prognostic factor for malignancies. However, GPS has never been assessed in patients with BMs who have undergone WBRT. The purpose of this study was to determine whether GPS can be used to identify subgroups of patients with BMs who have a poor prognosis, such as recursive partitioning analysis (RPA) Class 2 and Class 3, and who will not receive clinical prognostic benefits from WBRT.Materials and methodsA total of 180 Japanese patients with BMs were treated with WBRT between May 2008 and October 2015. We examined GPS, age, Karnofsky Performance Status (KPS), RPA, graded prognostic assessment (GPA), number of lesions, tumor size, history of brain surgery, presence of clinical symptoms, and radiation doses.ResultsThe overall median survival time (MST) was 6.1 months. seventeen patients (9.4%) were alive more than 2 years after WBRT. In univariate analysis, KPS ≤ 70 (p = 0.0066), GPA class 0–2 (p = 0.0008), > 3 BMs (p = 0.012), > 4 BMs (p = 0.02), patients who received ≥ 3 Gy per fraction (p = 0.0068), GPS ≥ 1 (p = 0.0003), and GPS ≥ 2 (p = 0.0009) were found to significantly decrease the MST. Patients who had brain surgery before WBRT (p = 0.036) had a longer survival. On multivariate analysis, GPS ≥ 1 (p = 0.008) was found to significantly decrease MST.ConclusionOur results suggest that GPS ≥ 1 indicates a poor prognosis in patients undergoing WBRT for intermediate and poor prognosis BMs.

OS13.6.A SURVIVAL IS ASSOCIATED WITH CIRCULATING B-CELLS AND AGE AMONG OLIGODENDROGLIOMA PATIENTS

Abstract BACKGROUND Though patients with oligodendroglioma have improved outcomes versus astrocytomas, there remains marked variability in their survival experience that is unexplained by known prognostic factors. Immunomethylomics uses DNA methylation as a substitute for flow cytometry to characterize immune cells in whole blood. We applied immunomethylomics to archived samples from patients with oligodendroglioma enrolled in the UCSF Adult Glioma Study (AGS) to assess the impact of immune-related characteristics on prognosis. MATERIAL AND METHODS We used immunomethylomics to evaluate immune cell fractions in archived whole blood from the AGS. We identified 141 patients with oligodendroglioma (IDH and TERT mutant, and 1p19q co-deleted) and compared them to other adult diffuse glioma subtypes (N = 347) and controls (N = 454). RESULTS The median age at diagnosis of the 141 oligodendrogliomas was 44 years (range 19 - 83), 57% were male, and 75% white. A total of 60% of patients had received chemotherapy, with or without radiation, prior to blood draw and 25% were taking dexamethasone at the time of sample collection. Oligodendroglioma patients had lower proportions of B cell fractions (total, memory, naïve) than people without glioma. Still, there were higher proportions of B cell (total, memory) and lymphocyte fractions (total, CD4-total and memory), basophil, eosinophil, and NK cells in oligodendrogliomas compared to other glioma subtypes. Median overall survival for the oligodendrogliomas was 17.5 years (95% CI 16.9 - NA) with median follow-up time of 12 years (95% CI 11 - 14). Recursive partitioning analysis delineated two statistically significantly different survival groups for oligodendroglioma patients based on an interaction between age at diagnosis and B naïve cell proportions. The group with the best outcome had a median survival of 24.2 years (95% CI 20.4 - NR) comprised of patients &amp;lt;/= 42 years of age with a higher proportion of B naïve cells (&amp;gt;1.91%) (N = 37). The group with the worse outcome had a median survival of 16.9 years (14.8 - NR) and included patients &amp;lt;/= 42 years of age with a lower proportion of B naïve cells (&amp;lt;/= 1.91%) (N = 28) OR patients &amp;gt; 42 years of age (N = 76) (p = 0.00032). Additionally, proportions of lymphocyte fractions (CD4 and CD8 total, memory, and naïve) were significantly lower in the worse outcome group. Grade or prior treatment did not differ between the groups. CONCLUSION We identified an interaction between age and B naïve cell proportions that distinguished survival outcomes for patients with oligodendrogliomas. Immunomethylomics is a powerful tool for assessing immune status and these findings suggest that blood-immune characteristics may play a vital role in the prognosis of patients with gliomas.

Relationship between transcriptional expression of pyruvate dehydrogenase and local control of disease in patients with oral cavity carcinomas

BackgroundThe altered cellular metabolism is one of the hallmarks of the cancer cells, favoring the process of aerobic glycolysis, known as the Warburg effect. The pyruvate dehydrogenase (PDH) complex is one of the elements involved in this metabolic process. The present study aims to evaluate the relationship between the transcriptional expression of PDHB and the risk of local recurrence in patients with oral cavity carcinomas. MethodsWe determined the transcriptional expression of PDHB in biopsies from 41 patients with oral cavity carcinomas treated with surgery. The PDHB expression was categorized according to the local control of the disease with a recursive partitioning analysis. ResultsDuring the follow-up period 13 patients (31.7%) had a local recurrence of the tumor. Considering local disease control as the dependent variable, the recursive partitioning analysis classified the patients in two categories according to high (n=16, 39.0%) or low (n=25, 61.0%) PDHB expression. Five-year local recurrence-free survival for patients with high PDHB expression was 84.8% (95% CI: 65.2-100%), and for patients with low expression it was 54.3% (95% CI: 34.3–74.2 %) (P=0.034). The results of multivariate analysis showed that patients with a low PDHB expression had a 4.90 times higher risk of local recurrence of the tumor (95% CI: 1.02–22.68, P=0.042). ConclusionThere is a relationship between the metabolic characteristics of the tumor and its aggressiveness. According to our results, patients with oral cavity carcinomas with low transcriptional expression levels of PDHB have a significantly higher risk of local tumor recurrence.

Treatment effect of remdesivir on the mortality of hospitalised COVID-19 patients in Switzerland across different patient groups: a tree-based model analysis.

Remdesivir has shown benefits against COVID-19. However, it remains unclear whether, to what extent, and among whom remdesivir can reduce COVID-19-related mortality. We explored whether the treatment response to remdesivir differed by patient characteristics. We analysed data collected from a hospital surveillance study conducted in 21 referral hospitals in Switzerland between 2020 and 2022. We applied model-based recursive partitioning to group patients by the association between treatment levels and mortality. We included either treatment (levels: none, remdesivir within 7 days of symptom onset, remdesivir after 7 days, or another treatment), age and sex, or treatment only as regression variables. Candidate partitioning variables included a range of risk factors and comorbidities (and age and sex unless included in regression). We repeated the analyses using local centring to correct the results for the propensity to receive treatment. Overall (n = 21,790 patients), remdesivir within 7 days was associated with increased mortality (adjusted hazard ratios 1.28-1.54 versus no treatment). The CURB-65 score caused the most instability in the regression parameters of the model. When adjusted for age and sex, patients receiving remdesivir within 7 days of onset had higher mortality than those not treated in all identified eight patient groups. When age and sex were included as partitioning variables instead, the number of groups increased to 19-20; in five to six of those branches, mortality was lower among patients who received early remdesivir. Factors determining the groups where remdesivir was potentially beneficial included the presence of oncological comorbidities, male sex, and high age. Some subgroups of patients, such as individuals with oncological comorbidities or elderly males, may benefit from remdesivir.

Sex-Specific Implant Fixation Can Reduce Revision Rates in Total Hip Arthroplasty: Evidence From the Swiss National Joint Registry

BackgroundPatient’s sex is considered a risk factor for revision following primary total hip arthroplasty (THA), but sex-specific treatment guidelines are lacking. The purpose was to assess sex-specificity of risk factors for periprosthetic femoral fractures (PFFs) and aseptic stem loosening (ASL) in a nationwide register study. MethodsAll uncemented and hybrid THAs for hip osteoarthritis registered in the Swiss National Joint Registry were considered. 86,423 THAs were analyzed. Comparable THA subsets for both sexes were obtained through propensity score matching (1:1). A sex-specific analysis of risk factors for early PFF or ASL was performed using recursive partitioning analyses. ResultsIn women, PFFs were most significantly associated with uncemented THA fixation (P < .0001) and age (P < .01, threshold: 70.5 years). The ASLs were solely associated with patient age of <65 years (P = .023). In men, PFFs were associated exclusively with an American Society of Anesthesiologists (ASA) score >2 (P = .026). The ASLs were not correlated to any of the potential risk factors analyzed. A mathematical simulation indicated that avoiding uncemented THA fixation in women ≥70.5 years of age decreased the number of revisions within the observational period by 21% in this subset and by 4.9% in the entire patient population. ConclusionUncemented THA should be avoided in women >70.5 years due to the increased risk of early PFF, while the mode of stem fixation did not influence revision risk in men. A sex-specific regimen for THA fixation has the potential to markedly reduce early THA revision rates.