临床病理数据预测乳腺癌复发风险
2015年《新英格兰医学杂志》发表的TAILORx研究结果证实，21基因复发风险评分可以预测早期乳腺癌术后复发风险以及术后辅助化疗能否获益，复发风险评分0～10分可以避免术后辅助化疗，复发风险评分≥26分推荐进行术后辅助化疗。不过，21基因检测费用大约需要4000美元，即使对于美国患者而言也太贵，况且尚未普及，未必可行，亦非必要。2017年，美国田纳西大学根据TAILORx研究结果发表了一种列线图（二维函数坐标图），利用病理报告提供的乳腺癌临床病理数据，不需进行21基因检测即可预测复发风险评分结果。不过，当时该列线图无法对复发风险评分11～25分患者进行预测。2018年《新英格兰医学杂志》发表了TAILORx研究结果更新，于是美国田纳西大学也对该列线图进行了更新。
2019年5月28日，欧洲乳腺癌专科医师学会《乳腺》在线发表美国田纳西大学的研究报告，根据最近发表的TAILORx临床研究结果，对临床病理数据列线图预测复发风险评分进行检验，并对复发风险评分进行分层。
该研究根据全国癌症数据库2010～2014年接受过21基因检测且肿瘤大小6～50毫米的65754例雌激素受体阳性HER2阴性淋巴结阴性乳腺癌患者数据对列线图进行更新。通过逻辑回归模型分析，对五个临床病理数据（患者年龄、肿瘤大小、肿瘤分级、孕激素受体状态、乳腺癌组织学类型）进行评定，以预测复发风险评分低（0～25）或高（26～100）。最后，根据全国癌症数据库2015年接受过21基因检测的18585例患者数据对预测结果进行验证。
结果，肿瘤分级和孕激素受体状态是预测复发风险评分高低的最强因素，其次为组织学类型、肿瘤大小、年龄。真假阳性率曲线表明，该列线图对于复发风险评分，无论高低，预测作用都很强，总体准确率高达92.7%，一致性指数高达0.81（95%置信区间：0.80～0.81）。
因此，该研究根据84339例患者数据，更新并验证了21基因复发风险评分预测列线图，田纳西大学将继续为无法负担、无法获得或没有必要进行21基因检测的乳腺癌患者在线免费提供该列线图。
http://utgsm.shinyapps.io/OncotypeDXCalculator
相关阅读
Breast. 2019 May 28;46:116-125. [Epub ahead of print]
Nomogram update based on TAILORx clinical trial results - Oncotype DX breast cancer recurrence score can be predicted using clinicopathologic data.
Amila Orucevic, John L. Bell, Megan King, Alison P. McNabb, Robert E. Heidel.
The University of Tennessee Medical Center, Knoxville, TN, USA; The University of Tennessee Health Science Center, Memphis, TN, USA.
HIGHLIGHTS
A nomogram/calculator can predict Oncotype DX results without performing the test.
A nomogram/calculator uses clinicopathologic data from pathology reports.
An updated nomogram was built from 84,339 patients' data (National Cancer Data Base).
A nomogram correctly assigns 92.7% of cases to the Oncotype DX lowor high-risk category at calculated probability range.85%-100%.
An updated nomogram/calculator is available at http://utgsm.shinyapps.io/OncotypeDXCalculator.
OBJECTIVES: Oncotype DX (ODX), 21-gene breast cancer (BC) assay, predicts risk of recurrence and benefits of addition of chemotherapy to hormonal therapy for early-stage BC. We previously published a nomogram/calculator that could predict ODX results without performing the test by using clinicopathologic characteristics of BC available from pathology reports. Patients with intermediate-risk (11-25) ODXRS (RS) were excluded from that nomogram. This update tests the predictive value of clinicopathologic variables for forecasting the ODXRS while including intermediate-risk-ODXRS patients and stratifying ODXRS based on recently published TAILORx clinical trial results (0-25=low-risk, 26-100=high-risk-ODXRS; intermediate-risk-ODXRS belongs to the low-risk category).
MATERIAL AND METHODS: The nomogram was built on 65,754 ODX-tested ER+/HER2-/lymph-node-negative patients with 6-50 mm tumor, captured by the National Cancer Data Base (NCDB) from 2010 to 2014. Five clinicopathologic variables (age, tumor size, grade, progesterone-receptor status (PR) and BC-histologic type) were assessed with logistic regression to predict for a low-risk (0-25) or a high-risk (26-100) ODXRS. Results were validated on a separate 18,585 ODX-tested cohort from 2015.
RESULTS: Grade and PR were the highest significant predictors of both low-risk and high-risk-ODXRS, followed by histologic type, tumor size and age. The Receiver Operator Characteristic (ROC) curve showed strong statistical model for both low-risk and high-risk-ODXRS prediction outcomes (c-index=0.81).
CONCLUSIONS: An updated nomogram is now developed/validated on the entire population of ODX-tested patients (84,339) captured by the NCDB. The nomogram/calculator, available on-line at the UTMCK/Shiny website (http://utgsm.shinyapps.io/OncotypeDXCalculator), will continue serving as a surrogate for BC patients for which ODX testing is not affordable, available or necessary.
KEYWORDS: Invasive breast cancer, Oncotype DX, Recurrence score prediction, Clinicopathologic variables, Breast cancer nomograms, On-line nomogram, Calculator for prediction of Oncotype DX recurrence score
DOI: 10.1016/j.breast.2019.05.006