Kang YJ, et al. J Breast Cancer. 2021 Dec;24:e55. https://doi.org/10.4048/jbc.2021.24.e55

Background: Manual review of electronic health records for clinical research is labor-intensive and prone to reviewer-dependent variations. Large language models (LLMs) offer potential for automated clinical data extraction; however, their feasibility in surgical oncology remains underexplored. Objective: This study aimed to evaluate the feasibility and accuracy of LLM-based processing compared with manual physician review for extracting clinical data from breast cancer records. Methods: We conducted a retrospective comparative study analyzing breast cancer records from 5 academic hospitals (January 2019-December 2019). Two data extraction pathways were compared: (1) manual physician review with direct electronic health record access (group 1: 1366/3100, 44.06%) and (2) LLM-based processing using Claude 3.5 Sonnet (Anthropic) on deidentified data automatically extracted through a clinical data warehouse platform (group 2: 1734/3100, 55.94%). The automated extraction system provided prestructured, deidentified data sheets organized by clinical domains, which were then processed by the LLM. The LLM prompt was developed through a 3-phase iterative process over 2 days. Primary outcomes included missing value rates, extraction accuracy, and concordance between groups. Secondary outcomes included comparison with the Korean Breast Cancer Society national registry data, processing time, and resource use. Validation involved 50 stratified random samples per group (900 data points each), assessed by 4 breast surgical oncologists. Statistical analysis included chi-square tests, 2-tailed t tests, Cohen κ, and intraclass correlation coefficients. The accuracy threshold was set at 90%. Results: The LLM achieved 90.8% (817) accuracy in validation analysis. Missing data patterns differed between groups: group 2 showed better lymph node documentation (missing: 152/1734, % vs 294/1366, %) but higher missing rates for cancer staging (211/1734, 12.17% vs 43/1366, 3.15%). Both groups demonstrated similar breast-conserving surgery rates (1107/1734, 84% vs 868/1366, %). Processing efficiency differed substantially: LLM processing required 12 days with 2 physicians versus 7 months with 5 physicians for manual review, representing a 91% reduction in physician hours (96 h vs 1025 h). The LLM group captured significantly more survival events (41 vs 11; P =.002). Stage distribution in the LLM group aligned better with national registry data (Cramér V=0.03 vs 0.07). Application programming interface costs totaled US $260 for 1734 cases (US $0.15 per case). Conclusions: LLM-based curation of automatically extracted, deidentified clinical data demonstrated comparable effectiveness to manual physician review while reducing processing time by 95% and physician hours by 91%. This 2-step approach—automated data extraction followed by LLM curation—addresses both privacy concerns and efficiency needs. Despite limitations in integrating multiple clinical events, this methodology offers a scalable solution for clinical data extraction in oncology research. The 90.8% accuracy rate and superior capture of survival events suggest that combining automated data extraction systems with LLM processing can accelerate retrospective clinical research while maintaining data quality and patient privacy.

Previous randomized trials, performed decades ago, showed no survival benefit of intensive screening for distant metastasis in breast cancer. However, recent improvements in targeted therapies and diagnostic accuracy of imaging have again raised the question of the clinical benefit of screening for distant metastasis. Therefore, we investigated the association between the use of modern imaging and survival of patients with breast cancer who eventually developed distant metastasis. We retrospectively reviewed data of 398 patients who developed distant metastasis after their initial curative treatment between January 2000 and December 2015. Patients in the less-intensive surveillance group (LSG) had significantly longer relapse-free survival than did patients in the intensive surveillance group (ISG) (8.7 vs. 22.8 months; p = 0.002). While the ISG showed worse overall survival than the LSG did (50.2 vs. 59.9 months; p = 0.015), the difference was insignificant after adjusting for other prognostic factors. Among the 225 asymptomatic patients whose metastases were detected on imaging, the intensity of screening did not affect overall survival. A small subgroup of patients showed poor survival outcomes when they underwent intensive screening. Patients with HR-/HER2 + tumors and patients who developed lung metastasis in the LSG had better overall survival than those in the ISG did. Highly intensive screening for distant metastasis in disease-free patients with breast cancer was not associated with significant survival benefits, despite the recent improvements in therapeutic options and diagnostic techniques.

Objective To investigate the prevalence and risk factors of axillary web syndrome (AWS) in Korean patients.Methods This retrospective study included a total of 189 women who underwent breast cancer surgery and received physical therapy between September 2019 and August 2020. We analyzed AWS and the correlation between the patients’ demographics, underlying disease, type of surgery and chemotherapy or radiation therapy, and lymphedema.Results The prevalence of AWS was found to be 30.6%. In the univariable analysis, age, chemotherapy, and hypertension were related to AWS. Finally, the multivariable logistic regression revealed that chemotherapy (odds ratio [OR]=2.84; 95% confidence interval [CI], 1.46–5.53) and HTN (OR=2.72; 95% CI, 1.18–6.30) were the strongest risk factors of AWS.Conclusion To the best of our knowledge, this was the first study that explored the risk factors of AWS in a Korean population after breast cancer surgery. As almost one-third of patients suffer from AWS after breast cancer surgery, it is essential to closely monitor the development of AWS in patients with hypertension or undergoing chemotherapy.

Although skin- or nipple-sparing mastectomy has been popular in the treatment of breast cancer, the radical excision of breast tissue is unavoidable in certain circumstances. However, the ability of an acellular dermal matrix (ADM) to expand remains questionable, and this situation may further hinder tissue expansion. From October 2017 to January 2020, patients who underwent immediate breast reconstruction with tissue expander placement using ADM whose initial fill volume was less than 50 mL were retrospectively reviewed. The primary outcomes were the number of visits and number of days required to complete the expansion, and the secondary outcomes were the amount of postoperative expansions, expander fill ratio and expander volume. Between the prepectoral group (n = 26) and subpectoral group (n = 39), the mean number of days (81.46 days versus 88.64 days, p = 0.365) and mean number of visits (5.08 versus 5.69, p = 0.91) required to complete expansion exhibited no significant differences. Additionally, there were no significant differences in the mean amount of postoperative expansion (314.23 mL versus 315.38 mL, p = 0.950), the mean final volume (353.08 mL versus 339.62 mL, p = 0.481) or the mean final volume ratio (0.89 versus 0.86, p = 0.35) between the two groups. Therefore, we suggest that prepectoral tissue expander placement after conventional mastectomy can be a valid option.