Bose Ashmita, Dittrich Peter, Gorecki Jerzy
Institute of Physical Chemistry, Polish Academy of Sciences, Warsaw, Poland.
Department of Mathematics and Computer Science, Friedrich Schiller University Jena, Jena, Germany.
Front Chem. 2022 Jul 8;10:901918. doi: 10.3389/fchem.2022.901918. eCollection 2022.
It can be expected that medical treatments in the future will be individually tailored for each patient. Here we present a step towards personally addressed drug therapy. We consider multiple myeloma treatment with drugs: bortezomib and dexamethasone. It has been observed that these drugs are effective for some patients and do not help others. We describe a network of chemical oscillators that can help to differentiate between non-responsive and responsive patients. In our numerical simulations, we consider a network of 3 interacting oscillators described with the Oregonator model. The input information is the gene expression value for one of 15 genes measured for patients with multiple myeloma. The single-gene networks optimized on a training set containing outcomes of 239 therapies, 169 using bortezomib and 70 using dexamethasone, show up to 71% accuracy in differentiating between non-responsive and responsive patients. If the results of single-gene networks are combined into the concilium with the majority voting strategy, then the accuracy of predicting the patient's response to the therapy increases to ∼ 85%.
可以预期，未来的医学治疗将针对每个患者进行个性化定制。在此，我们朝着个性化药物治疗迈出了一步。我们考虑使用硼替佐米和地塞米松这两种药物进行多发性骨髓瘤治疗。据观察，这些药物对一些患者有效，而对另一些患者则无效。我们描述了一个化学振荡器网络，它有助于区分无反应患者和有反应患者。在我们的数值模拟中，我们考虑了一个由3个相互作用的振荡器组成的网络，该网络用俄勒冈振子模型描述。输入信息是对多发性骨髓瘤患者测量的15个基因之一的基因表达值。在一个包含239种治疗结果的训练集上进行优化的单基因网络，其中169种使用硼替佐米，70种使用地塞米松，在区分无反应患者和有反应患者方面显示出高达71%的准确率。如果将单基因网络的结果采用多数投票策略合并到评议会中，那么预测患者对治疗反应的准确率将提高到约85%。
