Utility of Serum Biomarker Profiles in Predicting the Treatment Outcome in Chron

[Guest guest]

Utility of Serum Biomarker Profiles in Predicting the Treatment

Outcome in Chronic Hepatitis C Virus Patients

Reported by Jules Levin

DDW May 17-22, 2008, San Diego

Krishna S Kasturi1, R sen3, Audrey N Nguyen1, Ned Snyder2

1. Internal Medicine, University of Texas Medical Branch, Galveston,

TX, USA, 2. Gastroenterology & Hepatology, University of Texas

Medical Branch, Galveston, TX, USA, 3. Pathology, University of Texas

Medical Branch, Galveston, TX, USA

Background & Aim: ProteinChip technology (SELDI-TOF) has been used as

a profiling technique to identify potential biomarkers which may

represent proteins, protease-cleavage products, or novel peptides.

The purpose of this study was to use this technology to investigate

the differences in the pre-treatment serum profiles of patients who

were later classified as HCV treatment responders (R1) and non-

responders/ relapsers (R2).

Materials and Methods: 59 patients that had completed HCV combination

therapy were identified and classified as R1 or R2. Blood samples

prior to treatment were prefractionated using Equalizer® bead protein

biomarker discovery technology and SELDI ProteinChip® array

profiling. The fractions were analyzed on the PCS 4000 time of flight

mass spectrometer using CM10, IMAC (pretreated with Cu) and Q10

chips. Pooled normal human plasma was analyzed concurrently as

controls and to establish criteria for peak magnitude significance.

Data of the proteomic spectra was analyzed by Ciphergen Express Data

Manager software with Pattern Track and Two-way Hierarchical

Clustering algorithm. Advanced statistical analyses (logistic

regression, correlations, and case summaries) were done using SPSS

v.11.0 for Windows.

Results: Sample characteristics: Males=39, Females=12; Caucasian=31,

Blacks=9, Hispanics=10, Asian=1; Mean age 45 years (range 28 - 58);

Stage of fibrosis 0 (n=4), 1 (n=11), 2 (n=16), 3 (n=12), and 4 (n=8);

R1 (n=27), R2 (n=24).

Biomarker panel: 83 peaks of <20 KDa significantly differentiated

(p<0.05) R1 from R2. Upon further analysis (logistic regression), we

were able to identify 4 peaks of < 20 KDa (panel of 4 potential

biomarkers) that allowed the separation of R1 and R2 (sensitivity

81.5%, specificity 90.6%, PPV 88%, NPV 85%). Inclusion of other

routinely obtained biochemical variables did not have a favorable

influence on this panel. A predictive model was also created using

age, gender, race, HCVRNA and genotype of the study subjects

(sensitivity 81.5%, specificity 81.3%, PPV 78.5%, NPV 83.8%).

Conclusion: A predictive model using a biomarker panel was superior

in differentiating the serum profiles of treatment responders from

non-responders/ relapsers. Additional work to identify these putative

biomarkers along with the development of assays to validate this

study, and the application of this research data in a larger sample

will be necessary to evaluate the usefulness of serum biomarker

profiles in predicting response to HCV combination therapy. We would

like to acknowledge the technical support from Ciphergen Biosystems,

Inc., Fremont, CA, for this project.