Large Multi-institutional Cohort Research Articles

A multi-institutional retrospective cohort of adult-onset medulloblastoma in the modern era

Abstract Purpose Adult onset medulloblastoma (aMB) is a rare tumor with limited available evidence. We present a large multi-institutional retrospective cohort of aMB patients treated in the modern era, with an emphasis on understanding the role of chemotherapy at initial diagnosis. Methods We included 267 consecutive patients with aMB treated at 7 different institutions from 2000-present, controlling for chemotherapy regimen and cycles received. Results Treatment factors were highly intercorrelated with one another and with treating institution. Concurrent chemotherapy was not associated with overall survival (OS). Adjuvant chemotherapy was associated with OS on univariable analyses (HR=0.55, p=0.029) and on multivariable analysis when adjusting for risk status (HR 0.55, p=0.026) but not when also adjusting for treating institution. Proton craniospinal irradiation was associated with improved survival on univariable (HR=0.50, p=0.019) and multivariable analysis adjusting for risk status (HR=.51, p=.024) but not when treating institution was also considered. On subgroup analysis, adjuvant chemotherapy was associated with improved survival in M0 (HR=.55, p=.043) but not M1 disease, in patients with subtotal resection (HR=0.43, p=0.048) but not those with GTR. Similarly, progression-free survival was improved with chemotherapy in patients with M0 (HR=0.57, p=0.032) but not M1 disease, and in patients with subtotal (HR=0.50, p=.054) but not gross total resection. Conclusions There was no benefit of concurrent chemotherapy. Adjuvant chemotherapy was associated with improved overall survival and this effect was driven by select subgroups, specifically those with M0 disease and those with residual tumor. We could not confirm that these associations are independent of treating institution.

Multi-institutional Knowledge-Based (KB) plan prediction benchmark models for whole breast irradiation.

To train and validate KB prediction models by merging a large multi-institutional cohort of whole breast irradiation (WBI) plans using tangential fields. Ten institutions (INST1-INST10, 1481 patients) developed their KB-institutional models for left/right WBI (ten models for right and eight models for left). The transferability of models among centers was assessed based on the overlap of the geometric Principal Component (PC1) of each model when applied to other institutions and/or on the presence of significantly different optimization policies. Centers corresponding to transferable models were asked to join the building of two KB-benchmark models for right/left breast. Dose-volume histogram (DVH) prediction bands (lung/heart) were compared against those of the KB-institutional models. All models were transferable except INST6 (right breast) and INST1 (left breast). Planning data from 6 institutions for right breast and 5 institutions for left breast (out of 9 and 7 institutions with transferable models, respectively) were combined, totaling data from 850 patients. Prediction bands on the test cohorts (n=30/25 right/left) showed a large overlap with bands of each institution model: for the right-breast, the KB-benchmark model predicts slightly lower lung Dmean when compared to KB-institution models, except for INST7. Regarding the left-breast, even greater similarity between KB-benchmark and KB-institution model predictions was found. Multi-institutional KB-benchmark models for WBI were successfully generated. They may be employed by other users, representing the performances reached in a multi-institutional context of experienced centers. KB-benchmark models can also have significant applications for large-scale automatic plan optimization, QA/audit and tutoring/education purposes.

Quadrant, necrosis, and infection criteria for the risk stratification of walled-off necrosis: external validation using multi-institutional data.

Endoscopic ultrasound-guided peripancreatic fluid drainage (EUS-PFD) with on-demand endoscopic necrosectomy, increasingly utilized to manage walled-off necrosis (WON), is associated with substantial morbidity and mortality. This multicenter study aimed to externally validate recently developed quadrant (an abdominal quadrant distribution), necrosis, and infection (QNI) criteria for risk stratification in this setting. Of 423 patients with pancreatic fluid collections treated in a large multi-institutional cohort between 2010 and 2020, 212 with available preprocedural computed tomography images were included. Clinical outcomes between high- and low-risk groups defined according to QNI criteria were compared. The clinical success rate did not differ significantly between the QNI-high and QNI-low groups (74.4% vs. 83.5%, respectively; P = 0.12). Time to clinical success was longer in the QNI-high group compared with the QNI-low group (median, 68 vs. 46 days, respectively; P = 0.0016). Compared with the QNI-low group, the QNI-high group was associated with higher rates of percutaneous interventions, direct endoscopic necrosectomy, mortality, and adverse events. The QNI-high group was independently associated with time to clinical success (hazard ratio 0.64, 95% confidence interval 0.46-0.87; P =0.005) in multivariable Cox proportional hazards regression analysis, and with the risk of ≥grade III adverse events (odds ratio 2.93, 95% confidence interval 1.04-8.20; P = 0.04) in multivariable logistic regression analysis. The QNI criteria effectively stratified time to clinical success and the risk of adverse outcomes for patients receiving EUS-PFD of WON. Further investigations could elucidate appropriate management strategies based on QNI-based risk stratification.

NIMG-79. A RADIOMIC-RISK SCORE FOR SURVIVAL PREDICTION IN GLIOMA: A LARGE MULTI-INSTITUTIONAL RETROSPECTIVE VALIDATION STUDY

Abstract Gliomas have dismal survival. Recently, radiomics has demonstrated success in developing non-invasive image-based biomarkers for survival prediction in gliomas. However, clinical applicability of radiomics-based biomarkers will ultimately require rigorous validation on large, multi-institutional data. This work attempts to evaluate a radiomic-based survival risk-assessment prognostic score on a multi-institutional cohort of Glioma patients. Our rationale is that optimizing and validating radiomic features that capture intensity-based textural variations and morphology-based changes (e.g., local curvature and global contour) on large multi-institutional cohorts can allow for robust risk assessment in Glioma. n=668 pre-operative T1w MRI scans for Glioma patients from 4 cohorts: University of California San Francisco “UCSF” (n=495), LUMIERE (n=86)), xCures (n=13), and Cleveland Clinic (CCF) (n=74) were analyzed. Subjects from UCSF, CCF, and xCures were used for training, whereas the LUMIERE cohort was used for hold-out testing. Following pre-processing, expert-vetted tumor segmentation into Edema (ED) and Enhancing tumor (ET) regions was performed. A total of n=1558 radiomic features were extracted from the tumor regions (779 per region), namely, 31 global contour, 20 curvature-based, 624 gradient co-occurrence statistics, and 104 Collage features. Cox regression model was employed to create a radiomic risk score (RRS) for every subject, which included radiomics features as well as on clinical and demographic variables (IDH, MGMT, sex, age), for survival analysis. RRS incorporated a total of 43 radiomic features, and the associated clinical and demographic variables (CDS), yielding significant differences between high and low risk groups in the training (RRS: p=0.043, HR = 2, CSD: p=0.0031, HR = 2.8) and testing sets (RRS: p=0.0096, HR = 1.8, CSD: p=0.0071, HR = 1.8). Combining RRS and CDS improved the OS prediction at group level (p=3.6e-6, 0.0011, HR=5.4, 2, for training and testing, respectively). Radiomics-based risk-stratification may be promising for reliable risk-stratification in glioma.

PATH-47. AGE-, SEX-, AND GRADE-ASSOCIATED MOLECULAR DIFFERENCES IN A MULTI-INSTITUTIONAL COHORT OF CHOROID PLEXUS TUMORS

Abstract BACKGROUND Choroid plexus tumors (CPT) are central nervous system tumors where prognosis has been correlated to histologic grade, with aggressive clinical behavior and reduced survival rate in higher grade cases. Here, we present a genomic survey of a large multi-institutional cohort of CPTs across all grades. MATERIAL AND METHODS Fifty-eight CPT samples were analyzed by comprehensive genomic profiling (CGP) by a hybrid capture-based DNA sequencing platform (FoundationOne®CDx). Pathology reports, histopathology reviews, and patient clinical data were assessed. RESULTS Our cohort included 11 choroid plexus papillomas (CPP; grade 1), 16 atypical CPP (grade 2), and 31 choroid plexus carcinomas (CPC; grade 3). Males were slightly more affected than females (31 vs. 27 cases). Median patient age at tissue sampling was 11 years (range: 1–72 years). Median age of patients with CPC was younger at 3 years compared CPP at 34 years (p < 0.01). Genomic analyses identified three frequent, mutually exclusive mutations in TP53, TERTp, and DAXX. TP53 mutations (n = 15, 26%) were significantly more prevalent in CPC (n=14 CPC, p=0.0003), and were identified predominantly in children (n=12 CPC at <20 years). By contrast, TERT promoter mutations (TERTp; n = 5, 8.6%) were identified exclusively in CPP (p=0.019), predominantly grade 2 (4 of 5 cases), and exclusively in adult patients, with a median age of 48 years. DAXX mutations (n=4, 6.9%) were more common in CPC (n=3). Other alterations included PTEN (n=4, all adults) and MYCN amplification (n=3, all children <6 years), both exclusively in CPC. In 21 cases, including 9 out of the 11 Grade 1 CPP, no known cancer mutation was detected. CONCLUSION This study provides comprehensive genomic profiling of CPTs, highlighting distinct genetic alterations associated with different histological grades and patient demographics. These findings may inform future research and therapeutic strategies targeting specific molecular pathways in CPTs.

Beyond seizure freedom: Reduction in anti-seizure medication after epilepsy surgery

IntroductionPatients considering therapeutic epilepsy surgery often seek to decrease the number of anti-seizure medications (ASMs) they need. Predicting such reductions remains challenging. Although predictors of seizure freedom after epilepsy surgery are well-established, long-term outcomes remain modest and factors associated with ASM reduction, even in the absence of seizure freedom, may improve surgical planning to align with patient goals. MethodsWe studied a large multi-institutional cohort of patients who underwent epilepsy surgery between 2001 and 2022, with a minimum of two years follow-up. Preoperative features, including duration of epilepsy, epilepsy etiology, non-invasive investigation data, and total number of ASMs prescribed immediately prior to surgery were extracted for each patient. Primary endpoints included likelihood of ASM reduction and ASM freedom at multiple post-operative time points up to 15 years and stratified by seizure control. ResultsA total of 250 patients were followed for a median of 6.0 (range 2, 22) years after intracranial EEG (iEEG) surgery. Significant ASM reduction was only observed in those who underwent subsequent resection, whereas those undergoing neuromodulation saw their ASM usage maintained. Engel I outcomes were the strongest driver of ASM reduction. In patients with persistent seizures, patients with lateralized seizure onset zones (SOZs) also achieved sustained ASM reduction over time. Similarly, an increased number of preoperative ASMs also corresponded to a higher likelihood of ASM reduction across all follow-up periods. Other preoperative factors, including seizure etiology, did not independently influence ASM reduction. ConclusionsEven patients with persistent seizures after epilepsy surgery can observe meaningful ASM reduction during long-term follow-up. ASM reduction may be a relevant secondary outcome measure for epilepsy surgery.

S2113 Gender Disparities in the Impact of Bariatric Surgery in Metabolic Associated Steatohepatitis Disease (MASLD) Patients: Propensity Matched Analysis of a Large Multi-Institutional Cohort

S2113 Gender Disparities in the Impact of Bariatric Surgery in Metabolic Associated Steatohepatitis Disease (MASLD) Patients: Propensity Matched Analysis of a Large Multi-Institutional Cohort

Consensus molecular subtyping of metastatic colorectal cancer expands biomarker-directed therapeutic benefit for patients with CMS1 and CMS2 tumors.

We developed a whole transcriptome sequencing (WTS)-based Consensus Molecular Subtypes (CMS) classifier using FFPE tissue and investigated its prognostic and predictive utility in a large clinico-genomic database of CRC patients (n = 24,939). The classifier was trained against the original CMS datasets using an SVM model and validated in an independent blinded TCGA dataset (88.0% accuracy). Kaplan-Meier estimates of overall survival (OS) and time-on-treatment (TOT) were calculated for each CMS (p < 0.05 considered significant). CMS2 tumors were enriched on left-side of colon and conferred the longest median OS. In RAS-wildtype mCRC, left-sided tumors and CMS2 classification were associated with longer TOT with anti-EGFR antibodies (cetuximab and panitumumab). When restricting to only CMS2, there was no significant difference in TOT between right- versus left-sided tumors. CMS1 tumors were associated with a longer median TOT with pembrolizumab relative to other CMS groups, even when analyzing only microsatellite stable (MSS) tumors. A WTS-based CMS classifier allowed investigation of a large multi-institutional clinico-genomic mCRC cohort, suggesting anti-EGFR therapy benefit for right-sided RAS-WT CMS2 tumors and immune checkpoint inhibitor benefit for MSS CMS1. Routine CMS classification of CRC provides important treatment associations that should be further investigated.

Genetic variants in canonical Wnt signaling pathway associated with pediatric immune thrombocytopenia

AbstractThrough the use of genetic sequencing, molecular variants driving autoimmunity are increasingly identified in patients with chronic and refractory immune cytopenias. With the goal of discovering genetic variants that predispose to pediatric immune thrombocytopenia (ITP) or increase risk for chronic disease, we conducted a genome-wide association study in a large multi-institutional cohort of pediatric patients with ITP. A total of 591 patients were genotyped using an Illumina Global Screening Array BeadChip. Six variants met genome-wide significance in comparison between children with ITP and a cohort of healthy children. One variant in NAV2 was inversely associated with ITP (adjusted odds ratio [aOR], 0.52; P = 3.2 × 10−11). Two other variants in close proximity to NKD1 were also inversely associated with ITP (aOR, 0.43; P = 8.86 × 10−15; aOR, 0.48; P = 1.84 × 10−16). These genes have been linked to the canonical Wnt signaling pathway. No variants met genome-wide significance in comparison of those with ITP that self-resolved in <1 year versus those who developed chronic ITP. This study identifies genetic variants that may contribute to ITP risk and raises a novel pathway with a potential role in ITP pathogenesis.

(012) USE OF REAR TIP EXTENDERS IS ASSOCIATED WITH INCREASED RISK OF POST-OPERATIVE COMPLICATIONS AND PENILE PROSTHESIS REVISION IN PATIENTS UNDERGOING PRIMARY INFLATABLE PENILE PROSTHESIS PLACEMENT: RESULTS FROM A LARGE MULTI-INSTITUTIONAL COLLABORATIVE

Abstract Introduction The utilization of rear tip extenders (RTE) is often imperative during Inflatable Penile Prosthesis (IPP) placement. The impact of RTE on perioperative morbidity of IPP surgery is underexplored. Objective In this study, we assess intraoperative and postoperative outcomes among patients undergoing primary IPP implantation with or without RTE. Methods We performed a multicenter, retrospective analysis of men undergoing primary IPP placement between July 2016 and July 2021. A total of 4308 patients were identified, of which 2963 had no RTE and 1345 had RTE. After 1:1 propensity score matching for all confounding variables, 840 patients without RTE were compared to patients with RTE for intraoperative and postoperative outcomes. Chi-square and Mann–Whitney tests were used for statistical analysis for categorial and continuous variables respectively. For multivariable analysis, we used a forward stepwise model and included significant variables found on preoperative characteristics of patients. Kaplan–Meier time to event analysis were compared with Mantel-Cox log-rank test. Results Patients with RTE use had more postoperative complications (9.6% vs 5.2%, p &amp;lt; 0.001). More specifically, RTE patients had more episodes of postoperative hematoma and bleeding (2.0% vs 0.7%, p = 0.033), device malfunction (2.7% vs 1.8%, p = 0.006), and glans-related complications (0.8% vs 0, p = 0.015). On multivariable analysis, use of RTE (OR 1.58, 95% CI 1.07-2.34, p = 0.021), AMS 700 series device (OR 1.93, 95% CI 1.27-2.94, p = 0.002), penoscrotal approach (OR 2.60, 95% CI 1.61-4.18, p &amp;lt; 0.001), and older age (OR 1.02, 95% CI 1.00-1.04, p = 0.035) were associated with increased risk of postoperative complications. Patients with RTE use had increased rate of postoperative IPP revision (5.0% vs 2.1%, p = 0.002). The use of RTE (OR 1.71, 95% CI 1.00-2.90, p = 0.046), AMS 700 series device (OR 1.73, 95% CI 1.01-2.97, p = 0.048), penoscrotal approach (OR 3.16, 95% CI 1.63-6.17, p &amp;lt; 0.001), and history of Peyronie’s disease (OR 2.24, 95% CI 1.31-3.82, p = 0.003) were associated with increased risk of postoperative IPP revision or explantation. On Kaplan–Meier analysis, use of RTE was associated with increased risk of postoperative complications and IPP revision or explantation within 48 months (log rank p = 0.001 and p = 0.002 respectively). Conclusions In this large multi-institutional cohort, patients with RTE experienced more hematoma, device malfunctioning, and glans related post operative complications. The use of RTE is an independent predictor of post-operative complications and IPP revision or explanation. Disclosure No.

Creation of a predictive calculator to determine adequacy of occlusion of the woven endobridge (WEB) device in intracranial aneurysms-A retrospective analysis of the WorldWide WEB Consortium database.

Endovascular treatment with the woven endobridge (WEB) device has been widely utilized for managing intracranial aneurysms. However, predicting the probability of achieving adequate occlusion (Raymond-Roy classification 1 or 2) remains challenging. Our study sought to develop and validate a predictive calculator for adequate occlusion using the WEB device via data from a large multi-institutional retrospective cohort. We used data from the WorldWide WEB Consortium, encompassing 356 patients from 30 centers across North America, South America, and Europe. Bivariate and multivariate regression analyses were performed on a variety of demographic and clinical factors, from which predictive factors were selected. Calibration and validation were conducted, with variance inflation factor (VIF) parameters checked for collinearity. A total of 356 patients were included: 124 (34.8%) were male, 108 (30.3%) were elderly (≥65 years), and 118 (33.1%) were current smokers. Mean maximum aneurysm diameter was 7.09 mm (SD 2.71), with 112 (31.5%) having a daughter sac. In the multivariate regression, increasing aneurysm neck size (OR 0.706 [95% CI: 0.535-0.929], p = 0.13) and partial aneurysm thrombosis (OR 0.135 [95% CI: 0.024-0.681], p = 0.016) were found to be the only statistically significant variables associated with poorer likelihood of achieving occlusion. The predictive calculator shows a c-statistic of 0.744. Hosmer-Lemeshow goodness-of-fit test indicated a satisfactory model fit with a p-value of 0.431. The calculator is available at: https://neurodx.shinyapps.io/WEBDEVICE/. The predictive calculator offers a substantial contribution to the clinical toolkit for estimating the likelihood of adequate intracranial aneurysm occlusion by WEB device embolization.

Can Pelvic Lymph Node Dissection in Prostate Cancer Patients with a 5% Briganti Nomogram Cut-off Value Provide an Oncological Benefit? A Large Multi-Institutional Cohort Study in Japan.

The Briganti nomogram (cut-off value 5%) is commonly used to determine the indications for pelvic lymph node dissection (PLND) in patients with prostate cancer. We retrospectively analyzed the potential oncological benefit of PLND based on the 5% cut-off value on the Briganti nomogram. We obtained the data from the Medical Investigation Cancer Network (MICAN) Study, which included 3,463 patients who underwent a radical prostatectomy (RP) at nine institutions in Japan between 2010 and 2020. We included patients with Briganti scores ≥ 5% and a follow-up period ≥6 months and excluded patients categorized in the very high-risk group (based on NCCN categories); a final total of the cases of 1,068 patients were analyzed. The biochemical recurrence (BCR)-free survival was significantly worse in the patients who underwent PLND compared to those who did not (p=0.019). A multivariate analysis showed that high prostate-specific antigen (PSA) levels (p<0.001) and an advanced T-stage (p=0.018) were significant prognostic factors for BCR, whereas PLND had no effect on BCR (p=0.059). Thus, PLND in patients with prostate cancer whose Briganti score was 5% did not provide any oncological benefit. Further research is necessary to determine the indication criteria for conducting PLND.

LGG-60. DISTRIBUTION AND CLINICAL SIGNIFICANCE OF FGFR ALTERATIONS IN PEDIATRIC AND AYA GLIOMAS

Abstract Fibroblast growth factor receptor (FGFR) alterations are prevalent in human cancers. In a large Canadian population-based cohort, encompassing over 1,300 pediatric and Adolescent and young adult (AYA) gliomas, FGFR mutations occurred in 10% of pediatric low-grade gliomas (LGG) and 20% of IDH-WT AYA LGGs. Here we assembled a large multi-institutional cohort including more than 300 mutated gliomas, aged 6 months to 87 years, 53% of which were LGG. High-grade gliomas were more common with increasing age; over 90% of pediatric cases were LGGs, compared to 30% in the AYA group. Fusions, predominantly seen in the pediatric group (70% of LGGs), contrast with FGFR3-TACC3 fusions frequent found in adult high-grade gliomas. Interestingly, FGFR mutations sometimes existed as sole drivers or co-occurred with other alterations (~30%), it is unclear how this alters tumor behavior, although gain of additional alterations could be implied in malignant transformation and poor prognosis. Despite higher incidence of high-grade gliomas in AYAs, their progression-free survival was significantly better than that of pediatric LGGs (p=0.007), largely due to the prevalence of hemispheric tumors amenable to gross total resection. Approximately 30% of our LGG cohort underwent adjuvant therapy; with available data for 28 patients, objective responses were noted in 2/12 to chemotherapy, 2/9 to MEK inhibitors, and 3/7 to FGFR inhibitors, respectively. With the increasing promise of precision targeted therapy, molecular profiling and assessment for FGFR alterations is essential. Although encouraging, further studies are needed to assess the benefit of FGFR inhibition in these patients.

Prognosis and tumor-infiltrating lymphocytes (TILs) in estrogen receptor (ER)-low metastatic breast cancer (MBC): Analysis from a large multi-institutional cohort and the TONIC phase II trial with nivolumab.

1100 Background: Although 1% as ER cutoff to discriminate ER+/HER2- from triple-negative (TN) BC serves the purpose of maximizing access to endocrine therapy, it may be suboptimally informative in terms of prognosis and immune checkpoint inhibitor (ICI) benefit. In the early setting, we reported that ER-low BC (ER 1-9%/HER2-) is immunologically and prognostically similar to TN (ER 0%) [Massa, 2023]. Here we focus on MBC and aim to assess prognosis of ER-low BC, including the conversion from ER+(≥10%)/HER2- to ER-low at relapse. We also compared ER-low and TN in terms of outcome after ICI, and TILs. Methods: We included two cohorts of MBC patients (pts): a multicentric retrospective cohort and pts with ER&lt;10%/HER2- MBC enrolled in the TONIC trial testing nivolumab +/- induction therapy. All pts had ER and HER2 status assessed on metastatic tumor lesion; in the retrospective cohort ER and HER2 status on primary tumor was also available. Survival endpoints in the retrospective cohort: overall survival (OS, from BC diagnosis), post-relapse survival (PRS, from BC relapse). Exploratory endpoints from the TONIC trial: clinical benefit rate (CBR), OS and PFS (both from randomization). TILs were assessed on MBC samples (both cohorts) and on matched primary tumor (retrospective cohort). Results: We included 1112 pts in the retrospective cohort: the prevalence of ER-low BC was 3.6% in primary tumor and 5.9% in MBC lesions. In the TONIC trial, 15/110 pts had ER-low MBC. In the retrospective cohort, ER-low and TN MBC showed similar outcome, and significantly unfavorable vs ER+/HER2- MBC (Table). Among pts with ER+/HER2- primary BC, 6.7% converted to ER-low and 14.7% to TN at relapse. ER+/HER2- BC pts converting to ER-low had similar outcome than those converted to TN, and significantly poorer than pts with stable ER+/HER2- (Table). In the TONIC trial, ER-low MBC pts, compared to TN, showed similar response to ICI (CBR: 20.0% vs 22.1%, p=1), similar PFS and numerically worse OS (Table). TIL levels on MBC samples were similar between ER-low and TN both in the retrospective (median: 6% vs 5%, p=0.83) and the TONIC cohorts (median: 5.5% vs 5.0%, p=0.56) and were higher than ER+/HER2- (median: 3%). Conclusions: We confirmed that ER-low and TN MBC show similar poor outcome and the shift from ER+/HER2- to ER-low has a comparable unfavorable prognostic impact compared to conversion to TN. Our results support the hypothesis that ER-low and TN are immunologically akin and similarly prone to respond to ICI. [Table: see text]

Overall survival prediction and risk stratification in patients with neuroblastoma through machine learning in the large multi-institutional PRIMAGE cohort.

10054 Background: Neuroblastoma (NB) is the most prevalent solid cancer in childhood in which imaging plays a pivotal role at every step of the patient's journey. This investigation aimed to develop a machine learning model using clinical, molecular (MycN amplification), and magnetic resonance (MR) radiomics features at diagnosis to predict patient’s overall survival (OS) and improve their risk stratification. Methods: A database comprising clinical, molecular, and International Neuroblastoma Risk Group (INRG) staging on 513 patients with accessible MR imaging (discovery cohort) was employed for model training, validation, and testing. An additional 22 patients from non-discovery cohort hospitals served as an external validation group. Tumor segmentations of NB were manually and semi-automatically performed on corresponding T2-weighted MR images by an experienced radiologist. In total, 107 radiomics features were extracted and harmonized across MR scan manufacturers and magnetic field strengths using the nested ComBat methodology to correct both batch effects. These radiomics features, combined with clinical and molecular data, were utilized as input for the models. A nested cross-validation approach was employed for model development to determine the optimal preprocessing, machine learning algorithm, and model configuration. Results: The discovery cohort yielded a C-index of 0.788 ± 0.049 in the test partitions, with a random survival forest (RSF) exhibiting the best performance. In the validation cohort, a C-index of 0.934 was achieved. Interpretability analysis identified lesion heterogeneity, size, and molecular variables as crucial factors in OS prediction. The model demonstrated superior predictive performance and patient stratification compared to conventional staging systems in both cohorts. Conclusions: The RSF predictive model exhibited high performance, emphasizing the significant contribution of radiomics features and alignment with established clinical and molecular variables. The model stratified NB patients into low-, intermediate-, and high-risk categories and suggests that radiomics features potentially could improve current risk stratification systems. Additional external validation is warranted as this may present new evidence for enhancing patient care and clinical decision-making for NB patients.

Efficacy and survival of nivolumab treatment for recurrent/unresectable esophageal squamous-cell carcinoma: real-world clinical data from a large multi-institutional cohort

BackgroundReal-world clinical outcomes of and prognostic factors for nivolumab treatment for esophageal squamous-cell carcinoma (ESCC) remain unclear. This study aimed to evaluate real-world outcomes of nivolumab monotherapy in association with relevant clinical parameters in recurrent/unresectable advanced ESCC patients.MethodsThis population-based multicenter cohort study included a total of 282 patients from 15 institutions with recurrent/unresectable advanced ESCC who received nivolumab as a second-line or later therapy between 2014 and 2022. Data, including the best overall response, progression-free survival (PFS), and overall survival (OS), were retrospectively collected from these patients.ResultsObjective response and disease control rates were 17.0% and 47.9%, respectively. The clinical response to nivolumab treatment significantly correlated with development of overall immune-related adverse events (P < .0001), including rash (P < .0001), hypothyroidism (P = .03), and interstitial pneumonia (P = .004). Organ-specific best response rates were 20.6% in lymph nodes, 17.4% in lungs, 15.4% in pleural dissemination, and 13.6% in primary lesions. In terms of patient survival, the median OS and PFS was 10.9 and 2.4 months, respectively. Univariate analysis of OS revealed that performance status (PS; P < .0001), number of metastatic organs (P = .019), C-reactive protein-to-albumin ratio (CAR; P < .0001), neutrophil–lymphocyte ratio (P = .001), and PMI (P = .024) were significant. Multivariate analysis further identified CAR [hazard ratio (HR) = 1.61, 95% confidence interval (CI) 1.15–2.25, P = .0053)] in addition to PS (HR = 1.65, 95% CI 1.23–2.22, P = .0008) as independent prognostic parameters.ConclusionsCAR and PS before nivolumab treatment are useful in predicting long-term survival in recurrent/unresectable advanced ESCC patients with second-line or later nivolumab treatment.Trial RegistrationUMIN000040462

Abstract PO5-05-12: Loco-regional Treatment in De Novo Bone Only Metastatic Breast Cancer; Prospective, Multi-Institutional Real-World Data, BOMETIN, Protocol MF14-1a

Abstract Introduction: The impact of loco-regional treatment (LRT) on survival in de novo bone-only metastatic breast cancer (BC) is controversial. The aim of this study is to assess the effect of LRT on survival utilizing international, prospectively acquired data in this cohort of patients. Materials and Methods: Patients with de novo metastatic BC with bone-only metastases were divided into two groups: those receiving systemic therapy only (ST) and those undergoing LRT. Patients who received LRT were divided into two groups: those who received ST after LRT (LRT+ST arm) and those who received ST prior to LRT (ST+LRT arm). Solitary or multiple bone metastases were classified, and factors associated with disease progression were analyzed. Results: There were a total of 744 patients with de novo bone-only metastatic BC treated at each of the participating institutions between 2014 and 2022, with 372 (50%) participants in each arm. Median follow-up was 48 months (32-66, 25-75%). Patients in the LRT group were significantly younger than the ST group [50 (42, 60) vs 55 (44, 66), p=0.0001]. There were no significant differences in grade, Her2 neu and triple negative status, receipt of hormonal therapy and intervention to metastatic sites. During follow-up, 58% (n=217) of patients in ST arm and 32% (n=120) of patients in LRT arm died (p&amp;lt; 0.001). Local progression was observed in 20% of the patients in the ST arm whereas it was 9% in the LRT arm (p=0.0001). Systemic progression occurred more in ST arm; 66% (n=244) compared to 41% (n=152) of patients in LRT group (p&amp;lt; 0.001). The Hazard of death was 64% lower in LRT group than in ST group (HR: 0.36, 95% CI: 0.29-0.45), p&amp;lt;0.0001). The burden of metastatic disease was significantly different between groups with the solitary bone metastasis rate higher in LRT group than the ST only group (50% vs 24%, p&amp;lt; 0.001). However, the LRT group had better overall survival for both solitary (HR: 0.38, 95% Cl: 0.26-0.55) and multiple (HR: 0.38, 95% Cl: 0.29-0.51) bone metastases patients. Within the LRT group, survival rates were similar whether the breast surgery was performed before or after ST.Multivariate Cox analysis showed that LRT and ER/PR positivity significantly decrease the hazard of death (p&amp;lt; 0.05). Conclusion: Analysis of this large multi-institutional patient cohort provides further evidence that LRT improves overall survival and lowers loco-regional recurrence in patients with de novo bone-only metastatic BC. In breast cancer patients with bone-only metastases at presentation, the decision for LRT should be made through a multidisciplinary approach with consideration of surgical therapy at the primary tumor. Citation Format: Atilla Soran, Serdar Ozbas, Lutfi Dogan, Didem Can Trabulus, Jamila El-Azrhi, Kazim Senol, Berk Goktepe, Shruti Zaveri, Salyna Meas, Umut Demirci, Hasan Karanlik, Aykut Soyder, Ahmet Dağ, Ahmet Bilici, Mutlu Dogan, Mehmet Ali Nahit Sendur, Hande Koksal, Mehmet Ali Gulcelik, Neslihan Cabıoğlu, Levent Yeniay, N. Zafer Utkan, Nuri Karadurmus, Gul Daglar, Turgay Simsek, Birol Yildiz, Cihan Uras, Mustafa Tukenmez, Cihangir Ozaslan, Niyazi Karaman, Arda Isik, Berkay Demirors, Efe Sezgin, Vahit Ozmen, Anthony Lucci. Loco-regional Treatment in De Novo Bone Only Metastatic Breast Cancer; Prospective, Multi-Institutional Real-World Data, BOMETIN, Protocol MF14-1a [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-05-12.

Leveraging radiomics and machine learning to differentiate radiation necrosis from recurrence in patients with brain metastases.

Radiation necrosis (RN) can be difficult to radiographically discern from tumor progression after stereotactic radiosurgery (SRS). The objective of this study was to investigate the utility of radiomics and machine learning (ML) to differentiate RN from recurrence in patients with brain metastases treated with SRS. Patients with brain metastases treated with SRS who developed either RN or tumor reccurence were retrospectively identified. Image preprocessing and radiomic feature extraction were performed using ANTsPy and PyRadiomics, yielding 105 features from MRI T1-weighted post-contrast (T1c), T2, and fluid-attenuated inversion recovery (FLAIR) images. Univariate analysis assessed significance of individual features. Multivariable analysis employed various classifiers on features identified as most discriminative through feature selection. ML models were evaluated through cross-validation, selecting the best model based on area under the receiver operating characteristic (ROC) curve (AUC). Specificity, sensitivity, and F1 score were computed. Sixty-six lesions from 55 patients were identified. On univariate analysis, 27 features from the T1c sequence were statistically significant, while no features were significant from the T2 or FLAIR sequences. For clinical variables, only immunotherapy use after SRS was significant. Multivariable analysis of features from the T1c sequence yielded an AUC of 76.2% (standard deviation [SD] ± 12.7%), with specificity and sensitivity of 75.5% (± 13.4%) and 62.3% (± 19.6%) in differentiating radionecrosis from recurrence. Radiomics with ML may assist the diagnostic ability of distinguishing RN from tumor recurrence after SRS. Further work is needed to validate this in a larger multi-institutional cohort and prospectively evaluate it's utility in patient care.

Differential privacy preserved federated learning for prognostic modeling in COVID-19 patients using large multi-institutional chest CT dataset.

Notwithstanding the encouraging results of previous studies reporting on the efficiency of deep learning (DL) in COVID-19 prognostication, clinical adoption of the developed methodology still needs to be improved. To overcome this limitation, we set out to predict the prognosis of a large multi-institutional cohort of patients with COVID-19 using a DL-based model. This study aimed to evaluate the performance of deep privacy-preserving federated learning (DPFL) in predicting COVID-19 outcomes using chest CT images. After applying inclusion and exclusion criteria, 3055 patients from 19 centers, including 1599 alive and 1456 deceased, were enrolled in this study. Data from all centers were split (randomly with stratification respective to each center and class) into a training/validation set (70%/10%) and a hold-out test set (20%). For the DL model, feature extraction was performed on 2D slices, and averaging was performed at the final layer to construct a 3D model for each scan. The DensNet model was used for feature extraction. The model was developed using centralized and FL approaches. For FL, we employed DPFL approaches. Membership inference attack was also evaluated in the FL strategy. For model evaluation, different metrics were reported in the hold-out test sets. In addition, models trained in two scenarios, centralized and FL, were compared using the DeLong test for statistical differences. The centralized model achieved an accuracy of 0.76, while the DPFL model had an accuracy of 0.75. Both the centralized and DPFL models achieved a specificity of 0.77. The centralized model achieved a sensitivity of 0.74, while the DPFL model had a sensitivity of 0.73. A mean AUC of 0.82 and 0.81 with 95% confidence intervals of (95% CI: 0.79-0.85) and (95% CI: 0.77-0.84) were achieved by the centralized model and the DPFL model, respectively. The DeLong test did not prove statistically significant differences between the two models (p-value=0.98). The AUC values for the inference attacks fluctuate between 0.49 and 0.51, with an average of 0.50±0.003 and 95% CI for the mean AUC of 0.500 to 0.501. The performance of the proposed model was comparable to centralized models while operating on large and heterogeneous multi-institutional datasets. In addition, the model was resistant to inference attacks, ensuring the privacy of shared data during the training process.

(181) Prior Radiation Therapy is Associated with Increased Risk of Intra-Operative Complications in Patients Undergoing Primary Inflatable Penile Prosthesis Placement: Results from a Large Multi-Institutional Collaborative

Abstract Introduction Radiation therapy (RT) can predispose the onset of corporal fibrosis and microvasculopathy. The impact of RT on intra-operative complications is not well described in the literature. Objective In this study, we assess intra-operative and post-operative outcomes among patients with or without prior history of RT following primary Inflatable Penile Prosthesis (IPP) implantation. Methods We performed a multicenter, retrospective analysis of men undergoing primary IPP placement between January 2015 and December 2022. A total of 3530 patients were identified, of which 3319 had no prior history of RT and 211 patients had history of RT. After 1:1 propensity score matching for all confounding variables, 211 patients with a history of RT were compared to 211 nonradiated patients for intra-operative and post-operative outcomes. Chi-square and Mann-Whitney tests were used for statistical analysis for categorial and continuous variables respectively. For multivariable analysis, we used a forward stepwise model and included significant variables found on pre-operative characteristics of patients. Kaplan-Meier time to event analysis were compared with Mantel-Cox log-rank test. Results Patients with history of RT encountered more intra-operative complications (4.3% vs 0.9%, p=0.032); specifically, these men experienced more distal crossover events (1.9% vs 0, p=0.044). On multivariable analysis, history of prior RT (OR 5.29, 95% CI 1.07-26.3, p=0.041) and diabetes mellitus (OR 10.5, 95% CI 2.58-43.5, p=0.01) were associated with increased risk of intra-operative complications. History of obesity (OR 0.11, 95% CI 0.01-0.91, p=0.040) reduced risk of intra-operative complications. There was no significant difference in early or late post-operative non-infectious complications between both groups and there was also no significant difference in risk of post-operative complications within 24 months on Kaplan Meier analysis (log rank p=0.9). Conclusions In this large multi-institutional cohort, history of RT is an independent predictor of intra-operative complications but not post-operative complications during primary IPP placement. These findings may help inform patient-specific counseling while guiding surgical planning to prevent untoward morbidity following surgery. Disclosure No.