Towards qualification of FDG PET as biomarker of response to neo-adjuvant therapy: A meta-analysis.
e22128 Background: For FDG PET to qualify as a biomarker of response to cancer therapy beyond proof of principle, the evidence should be aggregated. Methods: A systematic literature search (Jan 1995-May 2011) for studies in solid extracerebral tumors with distinct cohorts of n≥10 patients, PET before and after cytotoxic neoadjuvant systemic therapy (with or without radiotherapy), dichotomized histopathological (PA) response. For model development we selected studies reporting individual or dichotomous aggregated level data, and performed bivariate SROC meta-regression. Results: 52 studies were identified, comprising data from 1,710 patients: rectal (32%), esophageal (21%), sarcoma (20%), breast (13%) and other cancers (14%). Most PET measures were SUVmax (65%) and other SUV measures (32%). Treatment type ChT (38%) and ChRT (62%); 46% were PA responders. The meta-regression indicated a weak increase in specificity for chemotherapy studies over chemoradiotherapy studies (p=0.08), however this disappeared after adjusting for publication bias. High baseline uptake rates improved diagnostic sensitivity (p=0.01). The SROC-AUC after adjusting for publication bias was 0.77. Maximal sensitivity and specificity was achieved with a PET reduction of 60%. Conclusions: There is a moderate association between FDG-PET change and histopathological response. This relation appears to be a function of baseline uptake, which may (at least in part) relate to repeatability issues.