(P27) Rate of Change in Quantitative PET/CT Metrics Prior to Definitive Chemoradiotherapy for Cervical Cancer Does not Add Prognostic Information Over Single Time Point Metrics

Abstract

Quantitative metrics from 18-F fluorodeoxyglucose positron emission tomography with computed tomography (PET/CT), such as pre-treatment maximum standard uptake value (SUVmax), metabolic tumor volume (MTV), and total lesion glycolysis (TLG), have been shown to prognosticate recurrence and/or survival from cervical cancer. The change in SUV through treatment and persistent PET avidity on follow-up imaging also predict outcome; however, the prognostic value of temporally changing metrics prior to and early in treatment is unknown. This study aimed to evaluate the prognostic value of the rate of change in quantitative metrics from serial PET/CT scans prior to and through definitive chemoradiotherapy, to identify patients at increased risk of disease recurrence. We identified thirty-three patients with FIGO IB1-IVB cervical cancer treated with definitive chemoradiation and brachytherapy boost and who underwent PET/CT imaging at at least two of the following: diagnosis/staging, simulation, and/or mid-treatment. Patient charts were retrospectively reviewed under Insitutional Review Board approval to collect patient demographics, treatment characteristics, and clinical outcomes. SUVmax, MTV (via gradient-based segmentation), and TLG were measured on PET/CT scans from at least 2 time points. The changes in each metric between each pair of time points was calculated to define the SUV velocity (SUVV), metabolic tumor growth rate (MTGR), and change in TLG (dTLG/dt). Single predictor Cox proportional hazards models were generated with Benjamini-Hochberg false discovery rate adjustment. Generalized linear mixed models evaluated temporally repeated measures (SUVmax, MTV, TLG) and logistic regression evaluated interval rates (SUVV, MTGR, and dTLG/dT) as predictors of recurrence. Median follow up was 12 months (range, 0-39) with 2 local failures (6.1%), 7 distant failures (21.2%), and 3 deaths (9.1%, all after a recurrence). Median (range) time to local failure and distant failure was 21.5 months (15-28) and 3 months (1-10), respectively. Of the 33 patients, 25 (75.8%), 25 (75.8%), and 30 (90.9%) had scans at diagnosis/staging, simulation, and mid-treatment, respectively. For all patients, median (range) SUVmax, MTV, and TLG at simulation were 16.9 U (5.6-41.4), 35.1 ml (1.7-298.2), and 270.2 U*ml (6.9-2979.2). Median (range) SUVV, MTGR, and dTLG/dt between diagnosis and simulation were 0.03 U/d (-1.91-0.18), 0.11 ml/d (-11.32-1.03), and 0.22 U*ml/d (-240.25-8.97), respectively. Single predictor Cox models showed increased hazard of recurrence for larger cervix GTV as measured on PET/CT (HR 1.01, 95% CI 1.00-1.02, p<0.01) and MTV at simulation (HR 1.02, 95% CI 1.00-1.03, p=0.03), however these were not significant after false discovery adjustment. The continuous change in SUV, MTV, and TLG over time and SUVV, MTGR, dTLG/dT between any two time points, were not prognostic/predictive of recurrence. As previously reported, pre-treatment MTV is prognostic of recurrence. The rate of change in SUV, MTV, and TLG prior to treatment or through the initiation of chemoradiotherapy do not add additional prognostic information. However, future studies with larger sample size, more consistent time points, and longer follow-up are warranted. While serial PET/CT may not add to single time point metrics, repeat PET/CT may be important to replicate treatment position, or to plan for lymph node boosts or brachytherapy.