CAN BASELINE PET‐CT FEATURES PREDICT OUTCOMES IN ADVANCED HODGKIN LYMPHOMA? A PROSPECTIVE EVALUATION OF UK PATIENTS IN THE RATHL TRIAL (CRUK/07/033)

Abstract

Introduction: PET‐CT after 2 ABVD (PET2) was used to guide treatment in advanced Hodgkin Lymphoma (HL) in the RATHL trial. Patients (pts) with PET2 negative (‐) scans were randomised to continue ABVD or AVD; pts with PET2 positive (+) scans were escalated to BEACOPP. This study aim was to evaluate whether baseline PET features of metabolic tumour volume (MTV), total lesion glycolysis (TLG) and number of extranodal sites could predict prognosis and PET2 response. Methods: Baseline total MTV and TLG and MTV and TLG of the bulkiest lesion (bulk) were measured in the first 100 pts using i) standardised uptake value (SUV) ≥ 2.5, ii) uptake ≥140% of mean liver uptake iii) ≥ 41% of maximum tumour SUV. Baseline total/bulk MTV and TLG using SUV ≥ 2.5 and extranodal sites were measured in all UK pts (n = 848). Results: MTV/TLG using SUV ≥ 2.5 and TLG compared to liver were associated with PFS and progression or death from HL (HL event) in the first 100 pts but the 41% method was not. MTV/TLG were then measured using SUV ≥ 2.5 in all UK pts, split into training and validation sets of 571 and 277. Pts with PET2+ scans had significantly higher total and bulk MTV and TLG than pts with PET2‐ scans; all p < 0.0002. Cox and logistic regression were used to assess association of MTV/ TLG and other baseline factors with PFS and HL events by 3 yr. In univariable analysis (UV) age, stage, B‐symptoms and TLG (total and bulk) were associated with PFS, but MTV and PET extranodal sites were not. Age, B‐symptoms and total TLG were significant in multivariable (MV) analysis. Stage, B‐symptoms and TLG (total and bulk) were associated with increased risk of a HL event by 3y in UV analysis but total TLG was the only significant variable in the stepwise selected MV model. A threshold of 3318 g (optimal by Youden's index) was used to divide pts into high and low total TLG groups. HL event rate at 3y was 12.8% for all pts with low TLG vs. 23.9% for all pts with high TLG; HR 2.2 (95%CI: 1.5‐3.4), p < 0.001. After a negative PET2, the rate of progression or death from HL was 21.5% vs 10.9% for high and low TLG respectively at 3y. The groups diverged further at 5y with rate of HL events of 31.0% and 13.1% for high and low TLG.Similar results were obtained in the validation set, suggesting that the threshold derived from the training set was reliable. Conclusion: In advanced HL, baseline TLG and MTV are significantly associated with PET2 response. TLG is a strong independent risk factor for prognosis, and may be useful for selecting patients likely to benefit from more intensive earlier therapy. A MV model including TLG may assess risk better than current clinical parameters but further work is needed. Such a model may be especially useful in pts with negative PET2 scans, in whom the overall 3 yr‐PFS of 85% was lower than anticipated.