ARTICLE ABSTRACTPurpose: There is evidence that therapeutic cancer vaccines can lengthen survival for some patients with cancer, but responses vary widely from one person to another. Methods to predict clinical outcomes will advance the field and provide new insights into critical determinants of in vivo efficacy.Experimental Design: This retrospective study included 141 subjects from phase II trials of PROSTVAC-VF, a poxvirus-based cancer vaccine currently in phase III clinical trials for advanced prostate cancer. A glycan microarray was used to profile prevaccination antiglycan antibody populations in sera as potential biomarkers for PROSTVAC-VF. The screen for predictive biomarkers identified antiglycan antibodies that consistently stratified subjects into groups with different Kaplan–Meier survival estimates. Because of the potential for overfitting, a permutation test was used to estimate the false discovery rate.Results: Prevaccination antibody levels to blood group A trisaccharide (BG-Atri) were found to have a statistically significant correlation with survival. Long-term survival was approximately doubled in subjects with abundant anti-BG-Atri immunoglobulin M (IgM) relative to subjects with little or no preexisting IgM for BG-Atri. This survival correlation was specific to vaccine treatment, as no correlation was observed in control patients immunized with wild-type poxviruses lacking the key tumor antigen, prostate-specific antigen (PSA). Moreover, anti-BG-Atri IgM levels were not correlated with general measures of disease severity, such as PSA levels, Gleason score, or Halabi predicted survival.Conclusion: In addition to reporting a new potentially predictive biomarker for PROSTVAC-VF, this study highlights the use of glycan microarray technology for improving our understanding of vaccine immunology. Clin Cancer Res; 19(5); 1290–9. ©2012 AACR.