Citations

Background: Generative AI and synthetic media have enabled realistic human Embodied Conversational Agents (ECAs) or avatars. A subset of this technology replicates faces and voices to create realistic likenesses. When combined with avatars, these methods enable the creation of “digital twins” of physicians, offering patients scalable, 24/7 clinical communication outside the immediate clinical environment. This study evaluated surgical patient perceptions of an AI-generated surgeon avatar for postoperative education. Methods: We conducted a pilot feasibility study with 30 plastic surgery patients at Mayo Clinic, USA (July–August 2025). A bespoke interactive surgeon avatar was developed in Python using the HeyGen IV model to reproduce the surgeon’s likeness. Patients interacted with the avatar through natural voice queries, which were mapped to predetermined, pre-recorded video responses covering ten common postoperative topics. Patient perceptions were assessed using validated scales of usability, engagement, trust, eeriness, and realism, supplemented by qualitative feedback. Results: The avatar system reliably answered 297 of 300 patient queries (99%). Usability was excellent (mean System Usability Scale score = 87.7 ± 11.5) and engagement high (mean 4.27 ± 0.23). Trust was the highest-rated domain, with all participants (100%) finding the avatar trustworthy and its information believable. Eeriness was minimal (mean = 1.57 ± 0.48), and 96.7% found the avatar visually pleasing. Most participants (86.6%) recognized the avatar as their surgeon, although many still identified it as artificial; voice resemblance was less convincing (70%). Interestingly, participants with prior exposure to deepfakes demonstrated consistently higher acceptance, rating usability, trust, and engagement 5–10% higher than those without prior exposure. Qualitative feedback highlighted clarity, efficiency, and convenience, while noting limitations in realism and conversational scope. Conclusions: The AI-generated physician avatar achieved high patient acceptance without triggering uncanny valley effects. Transparency about the synthetic nature of the technology enhanced, rather than diminished, trust. Familiarity with the physician and institutional credibility likely played a key role in the high trust scores observed. When implemented transparently and with appropriate safeguards, synthetic physician avatars may offer a scalable solution for postoperative education while preserving trust in clinical relationships.

PURPOSE Previous studies suggest that SPOP mutations result in increased sensitivity to androgen receptor pathway inhibitors (ARPIs) but are limited by lack of granular data. We hypothesize that SPOP mutations are associated with improved outcomes across the spectrum of advanced prostate cancer (PC). METHODS Using the real-world clinicogenomic Prostate Cancer Precision Medicine Multi-Institutional Collaborative Effort database, we analyzed outcomes based on SPOP mutation status. The primary end point was overall survival (OS) from the diagnosis of metastatic disease. Secondary end points included real-world progression-free survival (PFS) and PSA90 response. RESULTS Among 2,097 patients with metastatic PC, 5.5% (N = 115) had SPOP -mutated tumors. Compared with SPOP wild-type, patients with SPOP mutations were older at diagnosis (median 65 v 63 years; P = .001) and had higher (≥8) Gleason sum (68% v 54%; P = .02), higher incidence of lung metastasis (17% v 6%; P = .001), and increased frequency of intraductal features (13% v 5%; P < .001). SPOP mutations were associated with improved PSA90 response with first ARPI exposure in the castrate-resistant setting (60% v 40.6%; OR, 2.20 [95% CI, 1.15 to 4.19]; P = .02) but not in the castrate-sensitive setting (78.9% v 71.5%; OR, 1.49 [95% CI, 0.66 to 3.37]; P = .43). Median PFS with first ARPI did not differ in SPOP- mutated versus wild-type group in both the castrate-sensitive (24.2 v 19.2 months; hazard ratio [HR], 0.80 [95% CI, 0.49 to 1.32]; P = .39) and castrate-resistant settings (15.0 v 12.4 months; HR, 0.81 [95% CI, 0.58 to 1.12]; P = .20). In multivariable analysis, SPOP mutations were associated with improved OS (HR, 0.65, P = .02). CONCLUSION SPOP mutations were associated with longer OS despite the presence of aggressive clinical features. The distinct clinical and molecular features of SPOP -mutated PC support its consideration as a unique molecular subtype with prognostic implications.

Abstract Background: Prolonged cytopenias are a well-recognized complication following CAR-T therapy, leading to increased morbidity and healthcare utilization. Autologous SCB offers a potential strategy to promote hematologic recovery. We conducted a retrospective multi-institutional analysis across 18 centers comparing outcomes of SCB versus supportive care alone in RRMM patients who experienced prolonged cytopenias &gt;30 days after CAR-T infusion. Methods: Patients were included if they received SCB within 1 year after commercial CAR-T between 6/2021 and 3/2024. To identify matched controls, we reviewed other CAR-T recipients from the same consortium and selected those with an absolute neutrophil count (ANC) of &lt;0.72x109/L and platelet count of &lt;21K at day 30 post-infusion-thresholds representing the 75th percentile of cytopenia severity in the SCB cohort. These criteria ensured control patients had comparable hematologic toxicity but without SCB. Hematologic recovery in the SCB cohort was assessed using the CIBMTR engraftment criteria. For the non-SCB (nSCB) group, cell counts were analyzed cross-sectionally at day 90 post-CAR-T and compared to the SCB cohort with Kruskal Wallis tests. Progression-free survival (PFS) and overall survival (OS) were evaluated using Kaplan Meier methods. Results: Of the 590 CAR-T recipients, 91 patients (15.4%) developed prolonged cytopenias; 39 patients received SCB and 52 did not. SCB patients more frequently had ISS stage III disease (23.1% vs. 11.5%), exrtamedullary disease (59% vs. 46.2%) and penta-refractory status (43.6% vs. 34.6%). The median number of prior lines of therapy was 6 in both cohorts. Thirteen patients (33.3%) in the SCB and 11 (21.2%) in the nSCB received high intensity infusional chemotherapy (DCEP/PACE/HyperCVAD). Pre-CAR-T cytopenias tended to be less frequent in the SCB vs. nSCB cohort: ANC &lt;1000 in 5.1% vs. 13.5%, Hgb &lt;8 in 23.1% vs. 21.2%, and platelets &lt;50K in 28.2% vs. 40.4%. Median plasmacytosis in the bone marrow prior to CAR-T in the SCB group was 35% and 47.5% in the nSCB group. In the SCB cohort, 2 patients harbored DNMT3A mutations, one of whom had CCUS. Ide-cel was administered in 64.8% and cilta-cel in 35.2% of all patients. The rates and grades of cytokine release syndrome did not differ between groups. Six patients (15.4%) had HLH in the SCB group vs. none in the control group. All (100%) SCB patients received growth factor support; 58% also received TPO agonists. In the nSCB group, 43 patients (82.7%) received growth factor, and 16 patients (30.8%) received TPO agonists. Out of 24 pre-SCB bone marrow biopsies available, only 1 patient had relapsed MM while the remainder had hypocellular marrows. Median CD34+ cell dose was 2.93 million/kg (range: 1.76-23.55) administered at a median of 53 days post-CAR-T (range 24-265). All but one patient (97.4%) in the SCB group achieved hematologic recovery; median time to recovery was 24 days (range 9-87). No new toxicities attributable to SCB were observed. On day 90 post CAR-T infusion, SCB patients had higher median ANC (2.12 vs. 1.59x109/L, p=0.277), Hgb (10.6 vs. 8.7g/dL, p=0.002), and platelet counts (135 vs. 35K/L (p&lt;0.001). Minimal residual disease negativity was achieved in 82.4% (SCB) vs. 83.3% (nSCB). Infections occured in 53.8% and 48.1% of SCB and nSCB cohorts. All serious infections occurred prior to SCB. There were 12 deaths in the SCB group (7 from disease progression, 2 from infection, 1 from neurotoxicity (NT), and 1 unknown) and 27 in the nSCB group (22 from progression, 3 from infection, 1 from HLH, 1 from NT). After a median follow-up period of 12.6 months in the SCB cohort and 11.6 months in the nSCB arm, the mPFS was 11.0 months and 8.2 months, respectively. The mPFS was 10.6 months (SCB) and 5.8 months (nSCB) (p=0.216), and 11.0 months (SCB) and 10.9 months (nSCB), for ide-cel and cilta-cel treated-patients, respectively. Median OS was not reached vs.12.3 months in the SCB vs. nSCB groups, respectively. Conclusion: Overall, SCB was safe and effective in promoting rapid hematologic recovery in nearly all patients. Compared to controls, SCB recipeints had significantly higher Hgb and platelet levels by day 90 post-CAR-T infusion. SCB was associated wih a near doubling of PFS in patients who received ide-cel, though interpretation is limited by non-randomized design and small sample size. Further study is needed to optimize paient selection, timing, and integration of SCB into CAR-T care.

Abstract Background Patients (pts) with multiple myeloma and CNS involvement (CNS MM) have poor outcomes, and we recently showed that CAR-T has activity and is reasonable in this population (Gaballa et al., 2025). We sought to evaluate the safety and efficacy of Bispecific T-cell engagers (BsAbs) in CNS MM pts, with specific interest in evaluating ICANS. Methods We conducted a retrospective study of CNS MM pts treated with BsAbs via the MM immunotherapy consortium. Primary endpoints were toxicity rates, with a specific focus on ICANS. Secondary endpoints were systemic responses, CNS responses, and survival outcomes. CNS MM was defined as involvement of the brain parenchyma, spinal cord tissue, leptomeningeal disease, or CSF positivity. CNS response was determined based on CSF and imaging, as complete CNS response, partial CNS response, and non-response/progression. Results The study included 24 pts from 13 US institutions treated with teclistamab (33%), talquetamab (50%), elranatamab (17%), and were used as monotherapy in 83% while 17% received them in combination with either daratumumab or IMiDs. The median age was 64.5 (range 41-82), 50% female, 79% white, and 17% black. High-risk cytogenetics were identified in 39%, 4% had prior primary PCL, and 21% had secondary PCL (2 active within 30 days of BsAb initiation). All pts were triple-refractory, 33% were penta-refractory, and 39% had non-CNS EMD. Pts received a median of 5 prior lines of therapy (IQR 4-7), 71% prior transplant, and 21% prior CAR-T. Prior BCMA therapy was received in 83% of the talquetamab group and 17% of the BCMA BsAbs group (teclistamab and elranatamab). CNS disease was determined based on both imaging and CSF findings in 58%, imaging only in 33%, and CSF only in 8%. Site of CNS disease included 42% brain/cranial nerves, 13% spinal cord, or 46% with both. CNS disease was diagnosed prior to BsAb in 19 pts, with a median interval of 46 days (IQR 7-234), and 3 pts were identified within 2 weeks after BsAb initiation. Two pts identified at days 68 and 121 were included in the safety evaluation but excluded from the efficacy assessments. One pt received teclistamab and intrathecal chemotherapy (IT chemo) as a bridge to CAR-T and achieved a complete CNS response before CAR-T. CNS-directed therapy consisted of: 30% radiotherapy (RT), 30% IT chemo, 13% RT + IT chemo, 9% RT + surgery, and 4% steroids. CRS grade (G) 1-2 occurred in 50%, with no G3-4 CRS. ICANS G 1, 2, and 3 occurred in 13%, 21%, and 4%, respectively, with no G 4 ICANS. One pt developed delayed parkinsonism, and another developed leukoencephalopathy; both had CNS MM diagnosed pre-BsAbs and achieved a CNS response (1 complete, 1 partial). Infections occurred in 67% of pts, with 42% being severe, and 1 infection-related death. The systemic ORR of 18 evaluable pts was 67%, including 22% CR, 17% VGPR, and 28% PR. The overall CNS response of 19 evaluable pts was 58%, including 37% complete CNS response and 21% partial CNS response. Of 11 pts achieving a CNS response, 82% didn't have subsequent CNS progression. In 9 pts achieving CNS response after BsAb initiation, the median time to CNS response was 0.9 months (range 0.03-2.8), with an unreached median duration of CNS response. With a median follow-up of 11 months, the median PFS was 5.0 months (95% CI 2.5-NA) and OS was 12.2 months (95% CI 7.7-NA). No significant difference in median PFS (10.6 vs 4.6 months, p=0.11) or OS (20.2 vs 9.3 months, p=0.37) was observed between BCMA BsAbs (n=11) and talquetamab (n=11). Due to small sample limitation, no significant difference was observed in the median PFS (4.6 vs 8.2 months, p=0.48) or OS (20.2 vs 9.9 months, p=0.73) between BsAb monotherapy (n=19) and BsAb combination therapy (n=3). One pt who received teclistamab as a bridge to CAR-T achieved a complete CNS response and remains progression-free to date, demonstrating the best long-term outcome with a PFS of 674 days. Conclusions Our study demonstrates that BsAb use in pts with CNS MM is safe without increased incidence of ICANS. The use of BsAbs in combination with a multimodal CNS-directed approach led to an overall CNS response rate of 58% and an unreached median duration of CNS response. Previous studies (Gaballa et al., 2025) reported a 100% CNS response rate with CAR-T. In this study, 1 pt treated with BsAb as a bridge to CAR-T had the longest PFS, warranting further research to evaluate this combination's efficacy in this very high-risk population.

Background: During postoperative recovery, patients and their caregivers often lack crucial information, leading to numerous repetitive inquiries that burden healthcare providers. Traditional discharge materials, including paper handouts and patient portals, are often static, overwhelming, or underutilized, leading to patient overwhelm and contributing to unnecessary ER visits and overall healthcare overutilization. Conversational chatbots offer a solution, but Natural Language Processing (NLP) systems are often inflexible and limited in understanding, while powerful Large Language Models (LLMs) are prone to generating “hallucinations”. Objective: To combine the deterministic framework of traditional NLP with the probabilistic capabilities of LLMs, we developed the AI Virtual Assistant (AIVA) Platform. This system utilizes a retrieval-augmented generation (RAG) architecture, integrating Gemini 2.0 Flash with a medically verified knowledge base via Google Vertex AI, to safely deliver dynamic, patient-facing postoperative guidance grounded in validated clinical content. Methods: The AIVA Platform was evaluated through 750 simulated patient interactions derived from 250 unique postoperative queries across 20 high-frequency recovery domains. Three blinded physician reviewers assessed formal system performance, evaluating classification metrics (accuracy, precision, recall, F1-score), relevance (SSI Index), completeness, and consistency (5-point Likert scale). Safety guardrails were tested with 120 out-of-scope queries and 30 emergency escalation scenarios. Additionally, groundedness, fluency, and readability were assessed using automated LLM metrics. Results: The system achieved 98.4% classification accuracy (precision 1.0, recall 0.98, F1-score 0.9899). Physician reviews showed high completeness (4.83/5), consistency (4.49/5), and relevance (SSI Index 2.68/3). Safety guardrails successfully identified 100% of out-of-scope and escalation scenarios. Groundedness evaluations demonstrated strong context precision (0.951), recall (0.910), and faithfulness (0.956), with 95.6% verification agreement. While fluency and semantic alignment were high (BERTScore F1 0.9013, ROUGE-1 0.8377), readability was 11th-grade level (Flesch–Kincaid 46.34). Conclusion: The simulated testing demonstrated strong technical accuracy, safety, and clinical relevance in simulated postoperative care. Its architecture effectively balances flexibility and safety, addressing key limitations of standalone NLP and LLMs. While readability remains a challenge, these findings establish a solid foundation, demonstrating readiness for clinical trials and real-world testing within surgical care pathways.

Retrieval-augmented generation (RAG) quality depends on how source documents are segmented before indexing; fixed-length chunks can split concepts or add noise, reducing precision. We evaluated whether proposition, semantic, and adaptive chunking improve accuracy and relevance for safer clinical decision support. Using a curated domain knowledge base with Gemini 1.0 Pro, we built four otherwise identical RAG pipelines that differed only in the chunking strategy: adaptive length, proposition, semantic, and a fixed token-dependent baseline. Thirty common postoperative rhinoplasty questions were submitted to each pipeline. Outcomes included medical accuracy and clinical relevance (3-point Likert scale) and retrieval precision, recall, and F1; group differences were tested with ANOVA and Tukey post hoc analyses. Adaptive chunking achieved the highest accuracy—87% (Likert 2.37 ± 0.72) versus baseline 50% (1.63 ± 0.72; p = 0.001)—and the highest relevance (93%, 2.90 ± 0.40). Retrieval metrics were strongest with adaptive (precision 0.50, recall 0.88, F1 0.64) versus baseline (0.17, 0.40, 0.24). Proposition and semantic strategies improved all metrics relative to baseline, though less than adaptive. Aligning chunks to logical topic boundaries yielded more accurate, relevant answers without modifying the language model, offering a model-agnostic, data-source-neutral lever to enhance the safety and utility of LLM-based clinical decision support.

Introduction: Peripheral Arterial Disease (PAD) is a prevalent condition associated with considerable morbidity and mortality, a burden that is further intensified by the frequent co-occurrence of hypertension. The coexistence of these conditions not only amplifies cardiovascular risks but also complicates clinical management, thereby contributing to elevated mortality rates. Although their correlation is well understood, mortality trends in United States have not been studied. Methods: This study utilized International Classification of Diseases, 10th Revision (ICD-10) codes to extract age-adjusted mortality data for individuals with PAD and hypertension from the CDC WONDER (Center for Disease Control and Prevention Wide-Ranging Online Data for Epidemiologic Research) cohort. The overall sample and subgroups defined by demographics (age, sex, race/ethnicity) and region (state, census and urban-rural status) were analyzed. Trends in mortality were quantified by computing associated annual percentage changes (APC) using Joinpoint Regression Program Version 4.9.0.0 trend analysis software. Results: A total of 540954 deaths were observed, of which 260517 were men (48.2%). Overall, the age-adjusted mortality rates (AAMR) exhibited a sharp increase from 1999 to 2001 (APC:25.3; 95% CI: -0.36,56.2), followed by a sustained decrease until 2017 (APC: -2.16; 95% CI: -11.1, -1.52), after which a notable upward trend was observed (APC:9.38; 95% CI: -0.38,24.1). A similar AAMR trend was seen on gender-based subgroup analysis with male patients showing the highest AAMR 13.5 (95% CI: 13.1,13.7) in 2020. Amongst all racial groups, Black or African American showed the highest 14.9 (95% CI: 14.8-15) while Asian or Pacific Islander showed the lowest 5.8 (95% CI: 5.7-5.9) AAMR over 2 decades. The West region had the highest overall age-adjusted mortality rate (AAMR) of 10.2 (95% CI: 10.2-10.3). Among individual states, Vermont displayed the highest AAMR, reaching 15.8 (95% CI: 15.1-16.5). Conclusion: Healthcare disparities are evident. The unprecedented recent increase in mortality rates, following years of successful decline, necessitates a thorough evaluation of healthcare policies and PAD treatment advancements to identify the causative factors behind this trend reversal.

Abstract Disclosure: H. Ahmad: None. M. Kakakhel: None. A. Ullah: None. M. Saeed: None. J. Ikram: None. A. Qadeer: None. M. Khawar: None. H. Jawad: None. S. Haider: None. R. Haider: None. F. Ahmed: None. R.W. Kirchoff: None. Background: Type 1 diabetes mellitus (T1DM) is an autoimmune disease characterized by the destruction of insulin-producing cells in the pancreas. It is a global health concern with regional variation in incidence and prevalence. This study aims to analyze the regional incidence and prevalence of T1DM among adolescents and young adults (15-39 years) from 1990 to 2021 in 26 regions. Methods: Global Burden of Disease (GBD) data for 2021 were analyzed to assess regional incidence and prevalence of T1DM among adolescents and young adults. The Annual Average Percentage Change (AAPC) was calculated to evaluate trends from 1990 to 2021, with statistical significance determined using p-values. Results: In 2021, middle socio-demographic index (SDI) has the highest incidence (54541) as well as prevalence (1.86 million), followed by Low-middle SDI (I: 48416, P: 1.80 million) and South Asia (I: 46310, P: 1.73 million). Oceania has the lowest incidence (356) as well as prevalence (12885), followed by Andean Latin America (I: 911, P: 31954) and Caribbean regions (I: 1214, P: 54610), respectively. The largest increase in AAPC of incidence was showed by Southern Latin America (2.23, CI: 2.11-2.30), High-income Asia Pacific (2.22, CI: 2.18-2.26), and Western Europe (1.80, CI: 1.73-1.84) regions, while Western Europe (1.72, CI: 1.66-1.75), Eastern Europe (1.64, CI: 1.63-1.66), and Australasia (1.38, CI: 1.36-1.40) showed largest increase in AAPC of prevalence. The Caribbean region had the greatest decline in AAPC of prevalence (-0.11, CI: -0.08 to -0.12) followed by Eastern Sub-Saharan Africa (-0.07, CI: -0.06 to -0.09) and Central Latin America (-0.04, CI: -0.02 to -0.07). There was no region with decline in incidence. All AAPC values were statistically significant (p &lt; 0.000001). Conclusion: Middle SDI region exhibited the highest incidence and prevalence, highlighting the vulnerability of these populations, while Oceania and Andean Latin America bore the least burden. Notably, several regions, particularly in Europe, experienced substantial increases in both incidence and prevalence, emphasizing the need for targeted interventions. Conversely, some regions, such as the Caribbean, showed encouraging declines in prevalence, suggesting that effective prevention and control strategies can be implemented. These findings underscore the importance of tailored approaches to address the evolving burden of T1DM. Presentation: Saturday, July 12, 2025

Abstract Disclosure: M. Kakakhel: None. S. Cheema: None. R. Haider: None. R. Noor: None. S. N.l: None. S.M. Irshad: None. A. Ullah: None. H. Jawad: None. A. Nawaz: None. M. Khawar: None. A. Qadeer: None. H. Ahmad: None. R. Alcaraz: None. F. Ahmed: None. Background: Bone density and structure-related disorders, such as osteoporosis and related fractures, significantly contribute to morbidity and mortality, particularly among aging populations. Monitoring temporal trends and geographic variations in mortality is essential to inform targeted public health interventions. In this study, we aimed to examine temporal and geographic variations in mortality associated with bone density and structure-related disorders in the United States from 1999 to 2022. Methods: Using ICD-10 codes M80-M85, this retrospective ecological study utilized mortality data extracted from the CDC WONDER database for the years 1999-2022. Age-adjusted mortality rates (AAMRs) were calculated and stratified by demographic variables (age, sex, and race/ethnicity) and geographic regions. Temporal trends were analyzed using Joinpoint regression to estimate annual percent changes (APCs). Geographic variations were visualized using choropleth maps, and subgroup analyses were performed to explore disparities. Statistical significance was set at p &lt; 0.05. Results: Age-adjusted mortality rates (AAMRs) for bone density and structure-related disorders decreased overall from 1999 to 2019, with a subsequent rise observed in 2020 and 2022. Joinpoint regression revealed a significant negative trend from 2007-2016 (APC = -7.46%, p &lt; 0.000001), followed by a positive reversal from 2019-2022 (APC = 3.78%, p = 0.013). Both males and females experienced declines in mortality, but females had a slightly higher APC of -3.10% compared to -2.82% for males. Asian/Pacific Islander groups exhibited the most pronounced decrease in mortality (APC = -11.92%, p &lt; 0.000001) from 2009-2015, with a positive rebound in recent years. Black or African American populations had the smallest decline (APC = -1.35%). The Southern U.S. experienced the steepest decline in mortality from 2010-2016 (APC = -10.92%, p &lt; 0.000001), followed by a strong positive trend from 2019-2022 (APC = 6.61%, p &lt; 0.000001). Conversely, the West and Northeast regions showed milder trends with less pronounced reversals. Older age groups (65-74, 75-84) showed significant decreases in mortality, particularly the 65-74 group, which reversed to a positive trend (APC = 6.97%) from 2019-2022. The highest number of deaths occurred at the decedent’s home (3,330 in 2022). Subgroup analyses indicated significant temporal shifts in the mortality patterns across different racial/ethnic groups and geographic regions. Hispanic/Latino and Asian/Pacific Islander populations exhibited an increase in mortality rates post-2019, particularly in 2020. Conclusion: From 1999 to 2022, mortality from bone density and structure-related disorders in the U.S. exhibited significant temporal and geographic variations. These findings highlight the need for targeted public health strategies. Presentation: Monday, July 14, 2025

Abstract Disclosure: A. Ullah: None. R. Haider: None. M. Khawar: None. M. Abid: None. M. Hadeed Khawar: None. M. Shaukat: None. M. Asad Cheema: None. A. Komel: None. M. Ahmed Khan: None. R. Noureen: None. A. Batool: None. M. Mubashir: None. A. Qadeer: None. R. W. Kirchoff: None. Background: The role of dissecting the right paraesophageal lymph node (RPELN) in managing papillary thyroid carcinoma (PTC) remains a subject of debate. This meta-analysis examines both established and novel risk factors associated with RPELN metastasis. Methods: Relevant studies published up to December 2024 were retrieved using PubMed, Embase, Ovid, and Scopus search engines. Data were extracted and evaluated using the Newcastle-Ottawa scale. Statistical analysis was performed using Review Manager 5.4, with heterogeneity assessed through the Q-test and I²-test. Sensitivity analysis was conducted using the leave-one-out method, and the Egger regression test was used to generate a funnel plot for publication bias assessment. Results: Out of 2,444 articles screened, 26 studies involving 16,427 patients were included in the meta-analysis. The overall RPELN metastasis rate was 12.98% (2,133/16,427). The pooled analysis revealed significant associations between RPELN metastasis and the following factors: age under 55 years (P &lt; 0.00001), male gender (P &lt; 0.00001), tumor size &gt;1 cm (P &lt; 0.00001), multifocality (P &lt; 0.00001), capsular invasion (P &lt; 0.00001), vascular invasion (P &lt; 0.00001), extra-thyroidal extension (P &lt; 0.0001), central lymph node metastasis (P &lt; 0.00001), lateral lymph node metastasis (P &lt; 0.00001), Hashimoto’s thyroiditis (P = 0.002), micro-calcifications (P = 0.01), and altered echogenicity (P = 0.03). Conversely, no significant associations were observed for tumor location in the right lobe (P = 0.07), superior pole (P = 0.08), middle pole (P = 1.00), inferior pole (P = 0.28), or tumor margins (P = 0.93) (all P &gt; 0.05). Conclusion: Factors such as age &lt;55, male sex, tumor size &gt;1 cm, multifocality, capsular invasion, vascular invasion, extra-thyroidal extension, central lymph node metastasis, lateral lymph node metastasis, and Hashimoto’s thyroiditis are significantly associated with RPELN metastasis. These should be carefully considered during RPELN dissection in PTC management. Presentation: Sunday, July 13, 2025