Pancreatic Reference Set Application: Brian Haab-Van Andel (2012)

Pancreatic Ref Set App-Haab (2012)
Haab, BrianVan Andel Research Institute
Feng, ZidingFred Hutchinson Cancer Research Center
The panel consists of two distinct assays for CA 19-9 and three different glycoforms of the protein MUC5AC CA 19-9
No design specified.
G.I. and Other Associated Cancers Research Group

New markers are greatly needed for the detection and diagnosis of pancreatic cancer. Patients at high risk for developing pancreatic cancer (for, example because of genetic predisposition or health status) can be screened by endoscopy or a related imaging procedure, but these methods are expensive and burdensome to the patient. Blood-based markers would facilitate regular screening. In addition, patients with known abnormalities of the pancreas (for example, as observed incidentally from an abdominal scan) need to determine whether they have cancer or not. The great majority of patients with pancreatic findings by CT do not have conditions that require treatment, yet nearly all patients undergo invasive and burdensome procedures as a consequence of the CT. Again, a blood-based marker could alleviate this situation and potentially add accuracy to the diagnosis. In preliminary work we showed the potential for highly-accurate discrimination of pancreatic cancer from pancreatitis and healthy control subjects using a panel of protein and glycan markers in the serum. We used an antibody array platform in which we can obtain sensitive, reproducible measurements of protein abundance and glycosylation status in low sample volumes. The detection of the glycosylation status is important for the high accuracy of the test because the glycans attached to the marker proteins are altered in cancer patients. Based on the good performance in these early studies, we now want to validate the performance in rigorously controlled, blinded sample sets. The reference set developed by the EDRN will enable a definitive characterization of our marker performance. In addition, we can make an accurate comparison to other markers that will be applied to the same set and determine whether disparate markers could be used together for added benefit.

Antibody-Lectin Sandwich Arrays: We will use the Antibody-Lectin Sandwich Array (ALSA, Fig. 1) to acquire marker measurements on the reference set samples. A five-marker panel for pancreatic cancer: Our technology allows us to look at the levels of several specific glycans on proteins captured on antibody arrays. This approach can give better performance than standard methods for certain proteins that display disease-associated glycan changes. The multiplexed technology allows us to screen through many candidate markers using low sample volumes. CA 19-9 determination: CA 19-9 is the current best single marker for pancreatic cancer, so it will be important to determine the CA 19- 9 levels in the reference set to allow comparison to the new markers. The typical clinical assays required several hundred microliters of sample, which is considerable consumption of the valuable reference specimens. Our microarray assay requires only a few microliters of sample per replicate, and we previously have optimized and validated the CA 19-9 assay on the antibody microarray platform. Therefore another goal of our work with the reference set will be to obtain high quality measurements of CA 19-9.
Antibody-Lectin Sandwich Arrays We will use the Antibody-Lectin Sandwich Array (ALSA, Fig. 1) to acquire marker measurements on the reference set samples. In the antibody-lectin sandwich array method (ALSA, Fig. 1) [1], immobilized antibodies on microscope slides capture proteins out of biological solutions, and detection antibodies or lectins are used to probe the levels of the proteins or the glycans attached to the proteins (the example in Fig. 1 shows the detection of the CA 19-9 carbohydrate antigen on individual proteins [2]). The ability to measure glycans on specific proteins is valuable for developing effective biomarkers, as we have found that the detection of a glycan on a specific protein can perform better as a biomarker than the detection of the protein alone [1, 3-7]. Because some patients who do not have elevations of a certain protein do have glycan alterations on that protein, the measurement of both types of information gives better overall performance. The ALSA platform lends itself well to biomarker development and validation, since only 6 μl of diluted sample is used on each array, and many samples can be run in high throughput. The data analysis plan and the statistical analysis of the sample size were described previously in the setaside fund requests. The data and our calls of cases and controls will be sent to the DMCC for analysis, who will determine the performance of our biomarker panel. We will test the hypothesis that the new marker panel does not perform better than CA 19-9 alone (null hypothesis). Rejection of this hypothesis within the 95% confidence interval will give strong evidence that the new panel is significantly better than CA 19-9, the current best marker for pancreatic cancer. (Sample size calculations previously were provided to show that our anticipated performance can be assessed with at least 0.8 power using this reference set.) Such a result would provide a strong impetus for broad validation of the marker. The DMCC also will compare performance with other biomarkers panels applied by other groups and investigate the potential value of combining these diverse markers. The use of a common set of samples for all the markers enables this analysis. It will be extremely valuable to see which of the candidate markers performs best and to see how they fit together. If they have complementary performance, they could be used together to achieve better results than any individual marker. For example, if a subgroup of patients is detected by panel A, and a different, non-overlapping group is detected by panel B, the panels could be used together for improved performance, provide false-positive detection did not increase correspondingly. Validation studies will be planned for promising markers coming out of these analyses.

No datasets are currently associated with this protocol.

Version 5.1.0