Bayar Mohamed Amine, Antoun Sami, Lanoy Emilie
a Service de Biostatistique et d'Epidémiologie, Gustave-Roussy , Université Paris-Saclay , Villejuif , France.
b Université Paris-Saclay, Univ. Paris-Sud, UVSQ, CESP, INSERM , Villejuif , France.
Expert Rev Anticancer Ther. 2017 Apr;17(4):311-318. doi: 10.1080/14737140.2017.1298446. Epub 2017 Mar 2.
The term 'morphomics' stands for the markers of body composition in muscle and adipose tissues. in recent years, as part of clinical cancer research, several associations between morphomics and outcome or toxicity were found in different treatment settings leading to a growing interest. we aim to review statistical approaches used to evaluate these markers and suggest practical statistical recommendations. Area covered: We identified statistical methods used recently to take into account properties of morphomics measurements. We also reviewed adjustment methods on major confounding factors such as gender and approaches to model morphomic data, especially mixed models for repeated measures. Finally, we focused on methods for determining a cut-off for a morphomic marker that could be used in clinical practice and how to assess its robustness. Expert commentary: From our review, we proposed 13 key points to strengthen analyses and reporting of clinical research assessing associations between morphomics and outcome or toxicity.
“形态组学”一词代表肌肉和脂肪组织中身体成分的标志物。近年来，作为临床癌症研究的一部分，在不同治疗环境中发现了形态组学与预后或毒性之间的若干关联，这引发了越来越多的关注。我们旨在回顾用于评估这些标志物的统计方法，并提出实用的统计建议。涵盖领域：我们确定了最近用于考虑形态组学测量特性的统计方法。我们还回顾了对主要混杂因素（如性别）的调整方法以及对形态组学数据建模的方法，特别是重复测量的混合模型。最后，我们重点关注确定可用于临床实践的形态组学标志物临界值的方法以及如何评估其稳健性。专家评论：通过我们的综述，我们提出了13个关键点，以加强评估形态组学与预后或毒性之间关联的临床研究的分析和报告。
