The development of composite circulating biomarker models for use in anticancer drug clinical development

Abstract

The development of informative composite circulating biomarkers predicting cancer presence or therapy response is clinically attractive but optimal approaches to modeling are as yet unclear. This study investigated multidimensional relationships within an example panel of serum insulin-like growth factor (IGF) peptides using logistic regression (LR), fractional polynomial (FP), regression, artificial neural networks (ANNs) and support vector machines (SVMs) to derive predictive models for colorectal cancer (CRC). Two phase 2 biomarker validation analyses were performed: controls were ambulant adults (n 5 722); cases were: (i) CRC patients (n 5 100) and (ii) patients with acromegaly (n 5 52), the latter as ''positive'' discriminators. Serum IGF-I, IGF-II, IGF binding protein (IGFBP)-2 and -3 were measured. Discriminatory characteristics were compared within and between models. For the LR, FP and ANN models, and to a lesser extent SVMs, the addition of covariates at several steps improved discrimination characteristics. The optimum biomarker combination discriminating CRC vs. controls was achieved using ANN models [sensitivity, 94%; specificity, 90%; accuracy, 0.975 (95% CIs: 0.948 1.000)]. ANN modeling significantly outperformed LR, FP and SVM in terms of discrimination (p < 0.0001) and calibration. The acromegaly analysis demonstrated expected high performance characteristics in the ANN model [accuracy, 0.993 (95% CIs: 0.977, 1.000)]. Curved decision surfaces generated from the ANNs revealed the potential clinical utility. This example demonstrated improved discriminatory characteristics within the composite biomarker ANN model and a final model that outperformed the three other models. This modeling approach forms the basis to evaluate composite biomarkers as pharmacological and predictive biomarkers in future clinical trials.Biomarkers are increasingly important tools in oncology, and in particular to facilitate drug development.

DACH1: Its Role as a Classifier of Long Term Good Prognosis in Luminal Breast Cancer

“…In breast cancer hierarchical clustering analysis of gene expression array data has proven useful in providing broad molecular classification [3], but other techniques are required to identify biomarkers defining membership to various subgroups. Subsequently, computer algorithms incorporating a multilayer perceptron based Artificial Neural Network (ANN) method [16] have been adopted to identify cancer-relevant biomarkers to assist in clinical decision-making [17], [18]. Previously ANN has been used to identify a panel of protein biomarkers [19] capable of classifying breast cancer patients parallel to that achieved using gene expression profiling [3].…”

Prognostic relevance and performance characteristics of serum IGFBP‐2 and PAPP‐A in women with breast cancer: a long‐term Danish cohort study

“…With RFS as the endpoint, we repeated the above performance characteristics analyses using the SANN and SSVM models. We speculated that these machine‐driven approaches might better capture the MDR of the IGF‐related peptides . The optimal models for SANN and SSVM, in the testing sets, yielded AUCs of 0.665 and 0.690 for RFS as endpoint (Table )—in other words, there was no material improvement in performance over the Cox model.…”

“…We have previously capitalized on the modeling of multidimensional relationships of the IGF system using machine‐driven approaches, such as artificial neural networks (ANN) , and demonstrated considerable improvements (over and above conventional regression models) in performance characteristics (and thus, potential clinical utility) after measuring multiple IGF‐related biomarkers in the detection of colorectal cancer.…”

A Machine‐Learning Approach to Identify a Prognostic Cytokine Signature That Is Associated With Nivolumab Clearance in Patients With Advanced Melanoma

“…To achieve the most desirable prognostic performance, the combination of multiple features into a composite biomarker may be more robust and effective than an individual parameter . The machine‐learning modeling approach was utilized as an efficient way to integrate multiple biologic factors and compute them into a composite biomarker signature for outcome association .…”

“…To achieve the most desirable prognostic performance, the combination of multiple features into a composite biomarker may be more robust and effective than an individual parameter. 35 The machine-learning modeling approach was utilized as an efficient way to integrate multiple biologic factors and compute them into a composite biomarker signature for outcome association. 19 The machine-learning model developed for this analysis demonstrated good performance characteristics, with high concordance between the predicted nivolumab clearance from the derived cytokine profiles vs. the actual clearance level.…”