Abbreviated Name

SELDI-2

Lead Investigator

Semmes, John — Old Dominion University

Coordinating Investigator

Zheng, Yingye — Fred Hutchinson Cancer Center

Involved Investigators

• Chesnut, Gregory — Center for Prostate Disease Research Uniformed Services University of the Health Sciences and the Walter Reed National Military Medical Center
• Semmes, John — Old Dominion University
• — Johns Hopkins University Department of Urology
• Lin UW, Daniel W. — University of Washington
• Leach, Robin J — University of Texas Health Science Center at San Antonio
• Hutchinson, Amy — Frederick National Laboratory for Cancer Research
• Srivastava, Sudhir — National Cancer Institute
• Bigbee, William L. — University of Pittsburgh Cancer Institute
• Grizzle, William E. — University of Alabama at Birmingham
• Zheng, Yingye — Fred Hutchinson Cancer Center

Abstract

This project –A Comprehensive Program for the Validation of Prostate Cancer Early Detection with Novel Protein Identification Techniques -- is divided into three phases. The goal of Phase I was to assess the reproducibility and portability of Surface-Enhanced Laser Desorption and Ionization time-of-flight mass spectrometry (SELDI-TOF-MS) using protein profiles generated from serum. Phase I was recently successfully completed at six institutions using a single source of pooled sera.

Aims

The overall goal of Phase II is to develop and evaluate an algorithm for classifying cases and controls using protein profiles produced from SELDI-TOF-MS using serum collected from prostate cancer cases and non-cancer controls. Objective 1 - Identification of biorepositories Objective 2 - Generate protein profiles Objective 3 - Construction of classifiers and development of algorithm. Objective 4 – Validate the classifier constructed in Objective 3. If Objective 3, Aim 1, is successfully met

Analytic Method

The DMCC will apply the boosting decision tree,11 the boosting logistic regression13, and other classification methods to identify a classification algorithm that has minimum Cross-Validation error. The performance of this classifier will be assessed with the following the intended clinical uses in mind: 1)For men with PSA > 4, the population by current practice undergone biopsies with 25% Positive Predicted Value (PPV) for positive biopsies, we could judge that the new test has clinical potential if on this population sensitivity is significantly better than 85% and specificity is significantly better than 50% (null hypothesis corresponding to PPV=36%). This question is answered by comparing the biopsy-negative non-cancer controls with each of two cancer groups (Gleason Score < 7, and 7 or greater), and combined. In order for the Phase II validation study has adequate power (85%), the target observed specificity for this classifier in training sample should be at least 65% at sensitivity fixed at 95% on ROC curve for all prostate cancers (Aim 1: two case groups combined and compared to controls). Aim 2 (Gleason 7 or greater prostate cancers compared to control and low grade prostate cancer groups combined) is considered as secondary because using Gleason Score 7 or greater as the cutoff point for high grade is more for feasibility rather reason (there are not enough prostate cancers with Gleason Score 8 or 9). 2) If the criterion for Aim 1 is satisfied, validation part of Phase II will proceed. If the observed sensitivity and specificity are significantly better than the values specified by null hypotheses but less than the target values, the study group will evaluate whether the observed performance still has potential clinical benefit, and if it does, whether to increase sample size for the validation part of Phase II.

Comments

No comments available.

Outcome

A number of challenges arose during the validation study. One was the time required to collect the needed samples. Although at the study onset, it was anticipated that 6 months would be required for samples collection, due to (a) dissimilar rates of each sample type (e.g., cancer, biopsy negative control, inflammation, etc.), (b) sample-matching challenges, (c) requirements for adequate numbers of high-grade and lower grade tumors, and (d) the rigorous requirements for a range of samples across PSA levels, the sample acquisition ultimately required two years. Another challenge was a broken source on the SELDI-TOF-MS at UAB. This occurred concomitant with the destruction and shearing of the fan that spread metal fragments through the system, leading to unstable spectral patterns, it required several months to identify that the SELDI-TOF-MS source was the cause of the instability. Ultimately, daily QC procedures using a reference pooled serum sample allowed the team to recognize these issues and correct them prior to the initiation of the analysis stage. The final challenge occurring during the analysis of samples was a defective peptide calibration standard used for the first week of the study at EVMS. This calibration sample was derived from a defective manufacturer lot such that only 5 of the 7 peaks could be identified. This suboptimal standard was used for about 1/3 of the samples run at EVMS. Subsequently, this lack of uniform 7 peak standard data was solved by extrapolation of data across the study and was unnecessary when the wavelet analytical approach was employed.

Secure Outcome

No secure outcome available.

Publications

• SELDI-TOF MS whole serum proteomic profiling with IMAC surface does not reliably detect prostate cancer.

Biomarkers

• Proteomics pattern

Data Collections

• SELDI Data - Phase II

Start Date

Apr 19 2004

Estimated Finish Date

Sep 29 2005

Finish Date

Feb 1 2007

Protocol ID

110

Protocol Type

Validation

Fields of Research

• Proteomics

Collaborative Group

Prostate and Urologic Cancers Research Group

Cancer Types

• Malignant neoplasm of prostate

Phased Status

2