Bilkent University

Objective: This study analyzed demographic, clinical, and pathological features, treatment responses, and survival outcomes in non-metastatic esophageal SCC patients. Methods: A retrospective review of 74 patients was conducted. Performance status was assessed using ECOG, treatment response via RECIST 1.1, and staging via AJCC 7th edition TNM. Results: The male-to-female ratio was 0.95, with a mean age of 50.6. Tumors were mainly distal (58%), and most patients (68.7%) presented at stage 2 (25%) or 3 (43.7%). Surgery was performed in 64.9% (47.3% surgery only, 35.1% CRT only, 17.6% CRT + surgery). Median overall and disease-free survival were 38 and 24 months, respectively. Female gender, middle/lower tumor location, fewer metastatic lymph nodes, and CRT + surgery correlated with better survival (p < 0.05). Conclusion: Surgery was the most common treatment, but CRT + surgery yielded superior survival. Local recurrence was more frequent than distant metastasis. A combined approach is crucial for optimal outcomes.

Abstract We study black-box vector optimization with Gaussian process bandits, where there is an incomplete order relation on objective vectors described by a polyhedral convex cone. Existing black-box vector optimization approaches either suffer from high sample complexity or lack theoretical guarantees. We propose Vector Optimization with Gaussian Process (VOGP), an adaptive elimination algorithm that identifies Pareto optimal solutions sample efficiently by exploiting the smoothness of the objective function. We establish theoretical guarantees, deriving information gain-based and kernel-specific sample complexity bounds. Finally, we conduct a thorough empirical evaluation of VOGP and compare it with the state-of-the-art multi-objective and vector optimization algorithms on several real-world and synthetic datasets, emphasizing VOGP’s efficiency ( e.g. , $$\sim 18\times$$ lower sample complexity on average). We also provide heuristic adaptations of VOGP for cases where the design space is continuous and where the Gaussian process model lacks access to the true kernel hyperparameters. This work opens a new frontier in sample-efficient multi-objective black-box optimization by incorporating preference structures while maintaining theoretical guarantees and practical efficiency.

Background/Objectives: Systemic inflammatory markers have recently gained attention as prognostic indicators in various acute conditions. However, their predictive value in non-variceal upper gastrointestinal bleeding (UGIB) remains uncertain. This study aimed to evaluate the prognostic performance of the Systemic Inflammation Response Index (SIRI) and the Aggregate Index of Systemic Inflammation (AISI) for in-hospital mortality among patients with non-variceal UGIB and to compare them with established clinical scoring systems. Methods: This retrospective cohort study included 531 adult patients admitted with non-variceal UGIB between April 2023 and February 2025. Demographic, clinical, and laboratory data were collected at presentation. Inflammatory indices (SIRI, AISI, AISI/Hb) and established risk scores (Glasgow-Blatchford, Rockall, AIMS-65, and ABC) were calculated. The primary outcome was all-cause in-hospital mortality. Discriminatory ability was assessed using receiver operating characteristic (ROC) curve analysis, and independent predictors were identified by multivariable logistic regression. Results: The overall in-hospital mortality rate was 4.7% (25/531). Non-survivors were older and had lower systolic blood pressure, higher serum urea, and elevated inflammatory indices. Among biomarkers, SIRI (AUC = 0.773, 95% CI: 0.737-0.809) and AISI (AUC = 0.709, 95% CI: 0.670-0.747) showed good discriminatory ability, comparable to AIMS-65 (AUC = 0.765) and ABC (AUC = 0.786). In multivariable models, SIRI (OR = 1.10, p = 0.011) and AISI (OR = 1.04 per 100 units, p = 0.003) remained independent predictors of mortality after adjustment for age, systolic blood pressure, hemoglobin, serum urea, and albumin. Conclusions: SIRI and AISI are independent predictors of in-hospital mortality in patients with non-variceal UGIB, demonstrating comparable prognostic performance to conventional risk scores. These readily available inflammatory indices may serve as simple and cost-effective adjuncts for early risk stratification in clinical practice.

AimIMpower133 was designed with 4 cycles of chemoimmunotherapy with atezolizumab, unlike many other studies permitted 6-cycle induction. This study aimed to compare the efficacy and safety of a 4-cycle regimen with an extended 6-cycle regimen of carboplatin, etoposide, and atezolizumab in the first-line treatment of extensive-stage small-cell lung cancer (ES-SCLC).MethodsThis retrospective multicenter study was conducted across 13 oncology centers in Turkey. A total of 181 patients with ES-SCLC who received first-line treatment with chemotherapy plus atezolizumab were analyzed and grouped into those receiving 4 cycles (n = 101) and 6 cycles (n = 80) of treatment.ResultsThe median follow-up time was 20.9 months. The objective response rates were similar between the 4- and 6-cycle groups (79.2% vs 78.8%, P = .940). The disease control rates were also comparable (84.2% vs 86.3%, P = .940). The median progression-free survival (PFS) was 6.3 months in the 4-cycle and 8.1 months in the 6-cycle group (P = .357). The median overall survival (OS) was 16.6 months for the 4-cycle and 13.6 months for the 6-cycle group (P = .583). Intracranial progression was higher in the 4-cycle group (50.0% vs 25.5%, P = .007). Immune-related adverse events were similar between the groups.ConclusionThis is one of the largest real-world datasets comparing the efficacy and safety of 4 versus 6 cycles of induction chemoimmunotherapy. Although there were no significant improvements in PFS or OS, the 6-cycle regimen was associated with a lower observed rate of intracranial progression, predominantly among patients without baseline brain metastases. Further studies are needed to identify this subject in future.

Genomic data sharing has become an essential component of biomedical research, enabling large-scale collaborations and accelerating discoveries in human genetics. To balance the need for accessibility with privacy concerns, several controlled-access mechanisms have been proposed, including genomic beacons. Genomic beacons answer simple presence/absence queries about specific genetic variants. However, prior work has demonstrated that beacons remain vulnerable to genome reconstruction attacks, where an adversary can recover large portions of participants’ genomes using summary statistics. Building on insights from prior reconstruction attacks, we introduce an approach that unifies SNP correlation and allele frequency alignment objectives within a single-stage joint optimization framework. Unlike earlier two-stage methods that alternately minimized correlation and frequency losses, our joint formulation improves both reconstruction accuracy and computational effeciency, achieving an average F1-score of 71.4%, a 1.4 percentage point improvement over the state-of-the-art method and substantially outperforming the baseline approach (45%), while reducing the runtime for reconstructing 2000 SNPs across 100 individuals from 10 to 7.4 h, marking a 26% decrease in computational cost. These results underscore the pressing need for more robust privacy-preserving mechanisms in genomic beacon protocols.

IntroductionRecurrent laryngeal nerve (RLN) paralysis is a distressing complication of thyroidectomy, and many methods have been investigated for prevention.ObjectiveThe aim of our study is to investigate to what extent intraoperative neuromonitoring (IONM) predicts postoperative RLN paralysis and what factors may be causing it.MethodsPatients who underwent thyroidectomy in our clinic between January 2014 and December 2020 were included. Patient diagnoses, demographic characteristics, laboratory and imaging results, surgical information, and IONM results were evaluated.ResultsData from 287 patients and 523 RLNs were evaluated. Twenty-two (4.2%) of the RLNs were paralytic. The rate of paralysis was significantly higher in total thyroidectomies than in lobectomies (p = 0.046), permanent paralysis rate was higher in left RLNs. Malignancy was statistically associated with permanent paralysis (p = 0.017). The rate of overall and permanent paralysis was higher in patients with negative results of first, second response, first, and second laryngeal twitch examinations. The means of the first and second responses of the paralyzed RLNs were lower (p = 0.016,p < 0.01, respectively). The sensitivity of the IONM increased when the new thresholds of the first and second responses were determined to be 155 and 170 µv, respectively, by ROC analysis.ConclusionsThis study demonstrates that voltage variation and magnitude obtained from IONM can be predictive and sensitivity can be increased by using a threshold higher than 100 µV and patients in this range may be at risk for RLN paralysis. Our results may guide future studies of IONM and risk scoring systems to be developed for RLN paralysis.

To evaluate the biomechanical and tomographic characteristics of keratoconic corneas undergoing accelerated cross-linking (CXL) for over one year and to assess the impact of cone localization on postoperative biomechanical behavior. Prospective-comparative. This study included 68 eyes of 68 patients with progressive keratoconus. Based on cone location, the patients were divided into two groups: group 1 (central cone; 34 eyes) and group 2 (paracentral cone; 34 eyes). Scheimpflug imaging was used to assess the corneal thickness, anterior and posterior Kmean, Kmax, and astigmatism. The Ocular Response Analyzer (ORA) measures corneal compensated intraocular pressure (IOPcc), Goldmann-correlated pressure (IOPg), corneal hysteresis (CH), and corneal resistance factor (CRF). All the baseline measurements were repeated at 1, 6, and 12 months. Demarcation line (DL) depth was assessed at 1 month using anterior segment optical coherence tomography. Best corrected visual acuity was recorded at each visit. At 1year, anterior Kmean, posterior Kmean and Kmax showed significant reductions (p < 0.01). No significant differences in CH or CRF changes were found between the central and peripheral cones (p = 0.22, p = 0.33). Visual improvement and reduction in spherical equivalent were greater in the central cone group (p = 0.03, p = 0.03). DL depth correlated positively with CH and CRF at 1 month (p = 0.02, p = 0.008), but not at 1 year. Accelerated CXL improves corneal shape, but may not produce detectable changes in ORA-derived biomechanics. Regional stiffening may not be captured by global indices, and the cone location appears to be more influential on visual recovery than biomechanical parameters.