Tumor Treatment Response Based on Visual and Quantitative Changes in Global Tumor Glycolysis Using PET-FDG Imaging. The Visual Response Score and the Change in Total Lesion Glycolysis

Clin Positron Imaging. 1999 May;2(3):159-171. doi: 10.1016/s1095-0397(99)00016-3.

Abstract

"Functional" tumor treatment response parameters have been developed to measure treatment induced biochemical changes in the entire tumor mass, using positron emission tomography (PET) and [F-18] fludeoxyglucose (FDG). These new parameters are intended to measure global changes in tumor glycolysis. The response parameters are determined by comparing the pre- and posttreatment PET-FDG images either visually from the change in image appearance in the region of the tumor, or quantitatively based on features of the calibrated digital PET image. The visually assessed parameters are expressed as a visual response score (VRS), or visual response index (VRI), as the estimated percent response of the tumor. Visual Response Score (VRS) is recorded on a 5 point response scale (0-4): 0: no response or progression; 1: 1-33%; 2: >33%-66%; 3: >66%-99%; and 4: >99%, estimated response, respectively. The quantitative changes are expressed as total lesion glycolysis TLG or as the change in TLG during treatment, also called deltaTLG or Larson-Ginsberg Index (LGI), expressed as percent response. The volume of the lesion is determined from the PET-FDG images by an adaptive thresholding technique. This response index is computed as, deltaTLG (LGI) = {[(SUV(ave))(1) * (Vol)(1) - (SUV(ave))(2) * (Vol)(2)]/[(SUV(ave))(1) * (Vol)(1)]} * 100. Where "1" and "2" denote the pre- and posttreatment PET-FDG, scans respectively. Pre- and posttreatment PET-FDG scans were performed on a group of 41 locally advanced lung (2), rectal (17), esophageal (16) and gastric (6) cancers. These patients were treated before surgery with neoadjuvant chemo-radiation. Four experienced PET readers determined individual VRS and VRI blinded to each other as well as to the clinical history. Consensus VRS was obtained based on a discussion. The interobserver variability captured by intraclass correlation coefficient was 89.7%. In addition, reader reliability was assessed for the categorized VRS using Kendall's coefficient of concordance for ordinal data and was found to be equal to 85% This provided assurance that these response parameters were highly reproducible. The correlation of deltaTLG with % change in SUV(ave) and % change in SUV(max), as widely used parameters of response, were 0.73 and 0.78 (P <.0001) respectively. The corresponding correlation of VRI were 0.63 and 0.64 (P <.0001) respectively. Both deltaTLG and VRI showed greater mean changes than SUV maximum or average (59.7% and 76% vs. 46.9% and 46.8%). We conclude that VRS and deltaTLG are substantially correlated with other response parameters and are highly reproducible. As global measures of metabolic response, VRS, VRI and deltaTLG (LGI) should provide complementary information to more commonly used PET response parameters like the metabolic rate of FDG (MRFDG), or the standardized uptake value (SUV), that are calculated as normalized per gram of tumor. These findings set the stage for validation studies of the VRS and deltaTLG as objective measures of clinical treatment response, through comparison to the appropriate gold standards of posttreatment histopathology, recurrence free survival, and disease specific survival in well characterized populations of patients with locally advanced cancers.