(662e) Pharmacometabonomics Approach for Early Prediction of Chemotherapy Induced Peripheral Neuropathy | AIChE

(662e) Pharmacometabonomics Approach for Early Prediction of Chemotherapy Induced Peripheral Neuropathy

Authors 

Verma, P. - Presenter, Purdue University
Ramkrishna, D., Purdue University
Renbarger, J., Indiana University School of Medicine
Skiles, J., Indiana University School of Medicine
Cooper, B., Purdue University
Introduction: Vincristine is a core chemotherapeutic drug administered to pediatric Acute Lymphoblastic Leukemia (ALL) patients. In a subgroup of patient population, it can lead to Vincristine Induced Peripheral Neuropathy (VIPN), which can be severe and painful with long-term effects1, and is considered as the dose-limiting toxicity. Although vincristine has been used in chemotherapy for more than 50 years now, predictors and mechanism of VIPN induction remain unclear. Historically, an empirical cap of 2 mg is set for vincristine dosage. Due to this, a cohort of patients not susceptible to VIPN receives sub-therapeutic treatment, while another cohort may experience severe neuropathy. Thus, it is of interest to find predictors/biomarkers which can discriminate between these cohorts based on VIPN severity, before initiating chemotherapy.

Methods and Results: Here, pharmacometabonomics approach is employed to find metabolites as biomarkers to predict VIPN. Untargeted metabolite profiling using tandem mass spectrometry (mass spectrometry/mass spectrometry (MS/MS)) was performed on blood samples of pediatric ALL patients undergoing chemotherapy with vincristine as the primary drug. Samples were collected at three time points: day 8, day 29 and around 6 months during the treatment. These patients were classified as being susceptible to high or low neuropathy based on Total Neuropathy Score (TNS) calculated frequently throughout the course of the treatment. Following metabolite profiling, machine learning algorithms were used to determine a small set of metabolites highly predictive for VIPN, at those time points. Firstly, missing metabolite peaks were estimated using a variant of k-nearest neighbors imputation algorithm2, taking into account the possibility that a peak might be missing because that metabolite was not present in the patient sample, or because it was not detected by the instrument. Then, support vector machines along with recursive feature elimination algorithm was used to find that small set. Cross-validation models built with these selected metabolites as predictors had an area under Receiver Operating Characteristics (ROC) curve of approximately 0.95 at all the time points. Furthermore, rigorous manual validation of these metabolite peaks was performed to ensure that they were integrated correctly by the mass spectrometry instrument. Moreover, their molecular structures were identified using the MS/MS spectra and online databases such as HMDB and METLIN. This information was used to find relevant biological pathways, which substantiated significance of these metabolites as potential biomarkers.

Conclusion: A small set of metabolites was found that could accurately predict neuropathy in pediatric ALL patients treated with vincristine. A predictive model built with these metabolites will aid in dosage decision making for clinicians, and improve treatment efficacy for patients.

References:

  1. Gidding, C. E. M., et al. "Vincristine revisited." Critical reviews in oncology/hematology 29.3 (1999): 267-287.
  2. Armitage, Emily Grace, et al. "Missing value imputation strategies for metabolomics data." Electrophoresis 36.24 (2015): 3050-3060.