Application of whole genome amplification for the investigation of genomic mutations in Hodgkin’s lymphoma AND comparing and contrasting different approaches to identifying predictive biomarker combinations

Ramagiri, Pradeep (2012). Application of whole genome amplification for the investigation of genomic mutations in Hodgkin’s lymphoma AND comparing and contrasting different approaches to identifying predictive biomarker combinations. University of Birmingham. M.Res.

Preview

Ramagiri_12_MRes.pdf
PDF - Accepted Version
Download (2MB)

Abstract

Tumour cells of classical Hodgkin’s lymphoma (cHL)comprise only a fraction (1% or less) of the total cellular infiltrate that poses a great difficulty in the investigation of underlying genetic events in cHL development. This hurdle can be overcome by exome sequencing of a small number (~50) of isolated HL cells. This study used 6 HL cell lines (HDLM-2, KM-H2, L-428, L540, L-591, and L-1236) and 4 HL patients’ samples to look at the mutations in the TNFAIP3 gene. We observed same nucleotide changes in the cell lines KM-H2 (deletion of intron 2– exon 6 region), and L-1236 (G491A) as reported in a previous study (16).

Distinguishing malignant Adrenocortical carcinoma ACC from benign Adrencortical Adenoma (ACA) has been a major problem. In this study we attempt to design a biomarker tool based on steroid metabolic excretion data containing 32 distinct adrenal derived steroid secretions obtained from 102 ACA patients and 45 ACC patients. Variable selection was performed by forward selection, stepwise selection, and all subset combination methods using logistic regression. However, investigation into these variable subsets revealed ‘overfitting’. Models with 4 and 5-variable subset combinations have relatively high prediction capability with a significant sensitivity and specificity (AUC=99.3% and AUC=99.8%, respectively).

Type of Work:

Thesis (Masters by Research > M.Res.)

Award Type:

Masters by Research > M.Res.

Supervisor(s):