Early Detection Research Network
Team Project

Liver Full Reference Set Application: David Lubman - Univ of Michigan (2011)

Liver Full Ref Set App: Lubman (2011)
336
Feng, ZidingFred Hutchinson Cancer Research Center
combinations of HGF and CD14 and AFP and HGF.
No design specified.
Glycomics
[u'G.I. and Other Associated Cancers Research Group']

In this work we will perform the next step in the biomarker development and validation. This step will be the Phase 2 validation of glycoproteins that have passed Phase 1 blinded validation using ELISA kits based on target glycoproteins selected based on our previous work. This will be done in a large Phase 2 sample set obtained in a multicenter study funded by the EDRN. The assays will be performed in our research lab located in the Center for Cancer Proteomics in the University of Michigan Medical Center. This study will include patients in whom serum was stored for future validation and includes samples from early HCC (n = 158), advanced cases (n=214) and cirrhotic controls (n = 417). These samples will be supplied by the EDRN (per Dr. Jo Ann Rinaudo) and will be analyzed in a blinded fashion by Dr. Feng from the Fred Hutchinson Cancer Center. This phase 2 study was designed to have above 90% power at one-sided 5% type-I error for comparing the joint sensitivity and specificity for differentiating early stage HCC from cirrhotic patients between AFP and a new marker. Sample sizes of 200 for early stage HCC and 400 for cirrhotics were required to achieve the stated power (14). We will select our candidates for this larger phase validation set based on the results of previous work. These will include HGF and CD14 and the results of these assays will be used to evaluate the performance of each of these markers and combinations of HGF and CD14 and AFP and HGF. It is expected that each assay will be repeated three times for each marker and will also be performed for AFP as the standard for comparison. 250 uL of each sample is requested for analysis.

hase 2 validation of glycoproteins that have passed Phase 1 blinded validation using ELISA kits based on target glycoproteins selected based on our previous work.
The data for each sample for the ELISA assays will be sent to the Data Management and Coordinating Center(DMCC) of the EDRN for analysis. The DMCC has worked with us on analysis of the data in the 50:50 reference set of the phase 1 validation. They will continue to help us on blinding, randomization, study design, data analysis and interpretation for the validation study. They will also use similar methods to analyze the data in this phase 2 trial as used in the previous work. A logistic regression model was used to estimate the classification percentile values at different cutoff points for both single and combined markers. Based on empirical distributions of the resulting sensitivity and specificity, AUROC relevant summary indices can be computed by descriptive statistics and resampling processes. The performance of each marker was determined by generating a ROC curve and determining the AUROC and the optimal sensitivity and specificity for each marker and also the likelihood ratios. The area under the ROC (AUROC) curves were calculated and compared. In addition, they compared the results to that of the standard biomarker, AFP and studied how the new glycosylated markers might combine with AFP.

No datasets are currently associated with this protocol.