哨兵节点
结果(博弈论)
黑色素瘤
医学
节点(物理)
计算机科学
内科学
癌症研究
数学
物理
癌症
乳腺癌
数理经济学
量子力学
作者
Anja Ulmer,Vanessa Pfefferle,Vincent Walter,Massimo Granai,Ulrike Keim,Falko Fend,Mihály Sulyok,Hans Bösmüller
标识
DOI:10.1016/j.ejca.2022.06.054
摘要
Introduction Sentinel node biopsy is a key procedure to predict prognosis in melanoma. In a prospective study we compare reporting on melanoma cell densities in cytospin preparations with semiquantitative histopathology for predicting outcome. Patients and methods Sentinel nodes from 900 melanoma patients were bisected. One half of each node was disaggregated mechanically. The melanoma cell density (number of HMB45 positive cells per million lymphocytes with at least one cell showing morphological features of a melanoma cell) was recorded after examining two cytospins. For the second half the maximum diameter of metastasis was determined after haematoxylin and eosin (H&E) and immunohistological staining of three slides. Results Cytospins were positive for melanoma in 218 of 900 patients (24%). Routine pathology was positive in 111 of 900 (12%) patients. A more extensive pathological workup in cytospin-only positive patients led to a revised diagnosis (from negative to positive) in 23 of 101 patients (22.7%). We found a moderate but significant correlation between melanoma cell densities (determined in cytospins) and the maximum diameter of metastasis (determined by pathology) (rho = 0.6284, p < 0.001). At a median follow-up of 37 months (IQR 25–53 months) melanoma cell density (cytospins) (p < 0.001), thickness of melanoma (p = 0.008) and ulceration status (p = 0.026) were significant predictors for melanoma specific survival by multivariable testing and were all confirmed as key predictive factors by the random forest model. Maximum diameter of metastases, age and sex were not significant by multivariable testing (all p > 0.05). Conclusion Recording melanoma cell densities by examining two cytospins accurately predicts melanoma outcome and outperforms semiquantitative histopathology.
科研通智能强力驱动
Strongly Powered by AbleSci AI