202 Automated assessment of tumor infiltrating lymphocytes informs mortality in thin melanoma
Journal of Investigative Dermatology(2023)
摘要
While thin cutaneous melanomas (≤1.0 mm) demonstrate excellent individual prognosis, they are highly prevalent and contribute to one-quarter of aggregate melanoma deaths. We investigated whether an automated tumor-infiltrating lymphocyte (TIL) classification algorithm could predict mortality in thin melanoma beyond existing pathologic examination. This nested case-case study was sampled from a retrospective cohort of 27,660 patients with newly diagnosed thin melanoma in Queensland, Australia. Random matching of fatal cases (patients who died from melanoma) to non-fatal cases was performed by age, sex, year of diagnosis, tumor thickness, and follow-up duration. Subsequently, haematoxylin-eosin slides from 85 fatal cases and 85 paired non-fatal cases were analyzed through a QuPath-based cell classification algorithm (named NN192). We calculated the metric eTIL% (TILs/TILs+Tumor Cells) for each slide. Adjusted conditional logistic regression was employed to determine odds ratios (ORs) for melanoma-specific mortality. Thin melanomas in the lowest eTIL% quartile demonstrated higher mortality [OR: 3.47; 95% confidence interval (CI): 1.64-7.35; p = 0.001] than tumors in the remaining quartiles, including after adjustment for anatomical location, ulceration, and mitoses [OR: 3.31; 95% CI: 1.52-7.20; p = 0.003]. Pathologist TIL grading (absent/non-brisk/brisk) was not prognostic in this cohort. Overall, automated TIL analysis through the NN192 algorithm identified an immune-cold subset of thin melanomas that independently displayed higher melanoma-specific mortality. Pending prospective validation, this classifier has the potential to stratify clinical management of early-stage melanoma according to patients’ risk of disease progression.
更多查看译文
关键词
thin melanoma,tumor infiltrating,lymphocytes
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要