Applying PET to Broaden the Diagnostic Utility of the Clinically Validated CA19.9 Serum Biomarker for Oncology

For good reason, discovering biomarkers that can be assayed from biologic fluids has long been regarded as a holy grail for medical diagnostics. Indeed, several decades of systematic research have identified many secreted molecules differentially regulated (most often upregulated) in the context of malignant cancers that are now routinely measured in humans to screen for disease onset, develop prognoses, and monitor tumor response or recurrence. Their rapid commercialization, favorable economics, and simple experimental outputs (lending itself to standardization for multicenter trials) have engendered the widespread use of many analytic platforms to measure serum biomarker levels (e.g., enzyme-linked immunosorbent assays). The resulting vast body of epidemiologic data has consistently reinforced the notion that, although exciting progress has been made, we have yet to find a single, smoking-gun serum biomarker that can be effectively applied to address all of the above-mentioned clinical issues for a given cancer. Some of the most successful serum biomarkers in oncology make this point the most convincingly. For instance, whereas the highly restricted tissue expression of prostate specific antigen (a kallikrein uniquely produced and secreted from the prostate and upregulated in prostate cancer) has made it an exceptional tool for detecting residual tumor burden or recurrence after radical prostatectomy, its expression by benign pathologies of the prostate (e.g., prostatitis) limits its value in the context of early diagnosis of aggressive cancer (1,2). Likewise, whereas serum levels of CA125 (a peptide proteolytically cleaved from the transmembrane protein MUC16) are a faithful indication of changing tumor burden in ovarian cancer patients receiving systemic therapy, frequent over-expression of MUC16 in benign disorders (e.g., endometriosis, abdominal inflammation) or hormonally triggered fluctuations in ambient serum levels (e.g., menstruation, pregnancy) have limited the value of serum CA125 levels as a primary screening tool, particularly for asymptomatic premenopausal women (3,4). Although these data argue strongly for large-scale screening efforts to identify novel circulating biomarkers with better specificity for cancer, it is also reasonable to speculate that simply changing the manner in which a partially flawed serum biomarker is assayed could rescue its diagnostic utility in some contexts. Particularly for cases in which the biomarker’s specificity is corrupted by elevated secretion from other nonmalignant pathologies in distant host tissues, a tool that can discriminate the tissue sites of circulating biomarker production could more clearly distinguish benign from malignant disease. These considerations led us to hypothesize that a radiotracer capable of targeting the antigen at its tissue of origin could be a realistic supplement to antibody-based serum biomarker measurements. Because we anticipated that the limitations of preclinical disease modeling would prevent us from elegantly testing this hypothesis in small animals, we developed an experimental plan to facilitate rapid clinical translation by placing mutual emphasis on choosing a biomarker that fits the aforementioned epidemiologic criteria and one for which clinical grade reagents already existed. In this regard, CA19.9, or Sialyl Lewis A antigen, immediately emerged as a logical candidate for a radiotracer development program. Indeed, its shortcomings as a screening and diagnostic tool for pancreatic ductal adenocarcinoma (PDAC; the disease for which serum CA19.9 is most commonly measured) largely fit the aforementioned scenario, because epidemiologic data have shown that its specificity for PDAC is predominantly attenuated by antigen secretion from benign pathologies among distant organs (e.g., jaundice in the liver) and less so by the common benign disorders of the pancreas (e.g., pancreatitis) that an imaging tool would likely not be able to distinguish from malignant tissue (5–7). Moreover, several antibody development programs have been initiated (8–11), and one fully human monoclonal antibody (mAb), 5B1, was recently discovered. The recombinant mAb 5B1 potently binds an extracellular epitope on the Sialyl Lewis A protein, and the naked mAb was recently shown to confer powerful antitumor effects when systemically administered in preclinical cancer models (11). In considering radiolabeling strategies, we referred to the rapidly expanding body of preclinical and clinical data pointing to the quality of imaging data conferred by the radionuclide 89Zr (12–15). Consequently, we hypothesized that coupling 89Zr to 5B1 would result in a high-quality radiotracer with potential for very near-term clinical translation.