Semester

Spring

Date of Graduation

2022

Document Type

Dissertation

Degree Type

PhD

College

Eberly College of Arts and Sciences

Department

Forensic and Investigative Science

Committee Chair

Luis Arroyo

Committee Co-Chair

Sadie Bergeron

Committee Member

Sadie Bergeron

Committee Member

Patrick Callery

Committee Member

Tatiana Trejos

Abstract

Analytical methods aiming for the detection of novel psychoactive substances are continuously revised due to their utility in the seized drug and toxicology realms. One method frequently employed for the preliminary identification of illicit materials is portable Raman spectroscopy. Even when a substance in possession of an offender is identified, conclusive evidence that it may have been consumed requires additional confirmatory work and further toxicological evaluation of a biological specimen. Many times, the substance consumed may not be detected in the analyzed specimen due to its extensive metabolism. It is therefore challenging to rule out the identity of the drug ingested if metabolic studies have not been performed on a particular substance. This research aims to evaluate portable Raman as a quick, safe, non-destructive method for drug analysis using the instrument’s built-in algorithms and in-house machine and deep learning algorithms. Furthermore, metabolic and toxicologic studies using zebrafish and human liver microsomes are used to elucidate selected opioids.

In the first part of this research, a portable Raman instrument—TacticID was validated according to the United Nations Office on Drugs and Crime guidelines using 14 drugs and 15 cutting agents commonly encountered in seized drugs. Analysis was performed through glass and plastic packaging. In-house binary mixtures (n = 64) at the following ratios—1:4, 1:7, 1:10, and 1:20 were evaluated and the results compared to direct analysis in real-time mass spectrometry (DART-MS). Whereas Raman performed better at detecting diluents which consisted of the majority in the mixtures, DART-MS resulted in higher identification for easily ionizable drugs which were present in lower percentages. To compliment the weaknesses in each technique, both methods were combined, resulting in 96% accuracy. However, analysis of 15 authentic adjudicated cases resulted in 83% accuracy using the combined methods, demonstrating the usefulness of these methods as preliminary tests over traditional subjective techniques such as color tests.

In instances where a portable Raman instrument is used for drug screening, its accuracy as a single technique is crucial. In this study, the correct identification of the instrument detecting both drug and diluent in binary mixtures was 19%. Therefore, machine learning methods were explored as alternatives to the instrument’s built-in hit quality index algorithm. The findings in this research demonstrated that neural networks and convolutional neural networks were superior to the other algorithms, increasing the correct identification of both compounds to 65 and 64%, respectively. This work demonstrated how the contribution of machine learning can help improve the accuracy of analytical instruments outputs thereby increasing confidence in compounds reported.

In the second part of this research, zebrafish which share 70% of gene similarity to humans, were used as a toxicity model to provide information about drug effects on a living system. Fentanyl was selected as a model drug and zebrafish (0 – 96 hours post fertilization) were dosed at 0.01 – 100 µM. Major dose dependent phenotypic effects included pericardial malformations, spine, and yolk extension malformation, all of which inhibited the normal growth and development of the larvae. Additionally, the metabolism of fentanyl and valerylfentanyl were elucidated using zebrafish. Therefore, this work provided insight into the zebrafish model as an alternative to human toxicity and metabolism. The knowledge gained through this research will be used to understand the mechanisms by which these toxic and metabolic effects are observed.

Download

Share

COinS