Detection of cancer tissue and plasma/serum biomarkers will be another important focus of the AFFINOMICS programme. An aim of the project is to develop and employ technologies to screen for potential protein biomarkers with differential concentration in diseases compared to control groups. Platforms to be used include a fluorescent bead-based assay using a single antibody to each protein target, the sensitivity of which is presently around ng/ml, depending on the quality and specificity of each antibody, and multiplexed microarrays of antibodies or other binders as capture molecules. Analysis will take place in three stages: (i) discovery, in which all antibodies are used to analyse representative samples of a particular disease; (ii) validation, in which at least a few hundred representative samples of the same disease have been collected and finally (iii) qualification, in which several thousand samples from biobank collections are analysed to further validate the potential biomarker. The validiation phase is used to remove false positives identified in the discovery phase due to the small sample sizes.
Candidate markers will be selected using open access information, such as known cancer biomarkers, or those derived from signatures identified by partners (ULUND, DKFZ, VTT), histochemical studies (KTH), the Plasma Proteome Institute and the ICBC. Partner VTT has compiled and annotated published gene expression data into an extensive in silico transcriptomics (IST) database, containing mRNA expression levels of most human genes across 9,783 Affymetrix gene expression array experiments representing 43 normal human tissue types, 68 cancer types, and 64 other diseases (www.genesapiens.org). This database has been mined for novel disease-specific gene expression patterns, resulting in the identification of several promising biomarker candidates for the diagnosis and prognosis of specific cancer types and subtypes. For example, based on the initial finding from the database search, the transcription factor Sox11 was recently shown to be a novel prognostic factor for improved recurrence-free survival in epithelial ovarian cancer (co-authored by partners VTT and ULUND). The IST database will be utilised here (i) to prioritise targets for binder generation and (ii) to guide the selection of relevant clinical and biobank sample collections for binder validation studies. DKFZ also designed about 900 polyclonal antibodies from prior transcriptional data. Some 150 relevant tumour-derived proteins will be used as AFFINOMICS targets, 15-30 for each of the renal, prostate, breast, pancreatic, ovarian and brain cancers to complement host response serum biomarkers.
As well as tissue biomarkers for prognosis, targets will include serum markers for early diagnosis and follow-up. It has recently been shown by ULUND that the immune compartment is an early responder to different cancer indications and thus comprises a novel source of cancer response serum biomarkers with very high specificity and sensitivity when used to discriminate patients vs. normal individuals. Thus by covering around 110 different response biomarkers, a high coverage of relevant markers is obtained. The patient serum proteome includes proteins derived both from tumour leakage and host response. In proof of principle studies, using antibody microarrays in pancreatic and breast carcinomas, ULUND and DKFZ have observed ‘signatures’ of several proteins that discriminate healthy individuals from cancer patients with high sensitivity and specificity (>96%). The concept that serum contains much more information than originally anticipated has been validated in follow-up studies using serum/plasma derived from breast and pancreatic carcinoma patients. The output of a novel and large set of binders from the present proposal will allow the serum biomarker studies to be significantly expanded, allowing not only discovery, but pre-validation/validation studies to facilitate translation into the clinic. This novel avenue of proteomic research with diagnostic implications is critically dependent on rapid access to a large binder collection to provide the increased data resolution required for clinical therapy selection and monitoring disease progression, and thus have the potential to bring this proposal far beyond the current state of the art.