Moving a proteomic diagnostic from a clinical need to postmarketing validation

Abstract

Abstract Unlike some tumor types, the majority of the common solid tumors appear not to be driven by single dominant targetable pathways, and even those that are display variable behavior that can't be completely explained by these mutations. Diseases such as lung cancer are much more complex and heterogeneous, with many distinct and overlapping subsets of tumors within the class, and each subset will demand an in depth analysis to define the optimal therapeutic approach. These groups are starting to be defined by multiple technologies, exemplified by those tumors with mutant EGFR or activated ALK, who achieve substantial clinical benefit from minimally toxic targeted therapy. Even for these subsets, multiple resistance mechanisms have emerged requiring different salvage strategies. DNA sequence analysis will likely yield other small subgroups with direct therapeutic implications, and expression arrays are beginning to identify others, but analysis of the proteome has many theoretical advantages, for a complete knowledge of the proteome would encompass all known mechanisms of functional dysregulation associated with the development of cancer, including DNA mutations, rearrangements, transcriptional alterations and promoter methylation, but also post-translational modifications. This depth of information is still far from reality, however, and the true information content of today's technologies leaves a lot to be desired. Proteomic-based biomarker development has additional hurdles not found in the development of genomic ones. Analyte lability, chemical reactivity, low abundance of interesting proteins, and lack of a simple method for amplifying specific signals are among these, leading to many sources of systematic bias. However, clinically useful biomarkers will have to be robust enough to withstand routine clinical handling procedures, so we initiated studies using the simple, inexpensive, and rapid technology of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI MS) we studied unfractionated, pretreatment sera to identify NSCLC patients with improved survival after treatment with the EGFR TKIs gefitinib and erlotinib. Mass spectra, independently acquired at two institutions, gave highly concordant results, and were used to generate an algorithm predictive of time to progression and survival. This prediction algorithm was then validated in a blinded manner in two independent cohorts of NSCLC patients treated with EGFR TKIs. This classification algorithm did not predict outcome in three independent cohorts of patients who did not receive treatment with EGFR TKIs. Thus, if upheld in prospective clinical trials, this analysis of pre-treatment peripheral blood might be useful in selecting therapy for advanced non-small cell lung cancer patients. We are currently in the process of testing this signature in sample sets from past randomized clinical trials, and a prospective trial is underway. We are currently using much more informative shotgun proteomic analyses to define novel diagnostic biomarkers as well as therapeutic targets.