For tailored therapy selection and result optimization in diffuse large B-cell lymphoma (DLBCL), it was critical to identify novel predictive biomarkers. Although the usefulness of metabolic tumor volume (MTV) in DLBCL patients receiving loncastuximab tesirine had not been determined, it was a significant prognostic factor. For a study, researchers sought to determine how the use of quantitative PET/CT data affected the treatment response and survival of patients receiving this medication.

They examined screening PET/CT images of patients enrolled in the LOTIS-2 trial for post-hoc analysis. Using the 41% of SUVmax threshold, the metabolic volumes of all individual lesions were added to obtain the MTV. Total lesion glycolysis (TLG) was calculated for each individual lesion as the sum of MTV and SUVmean. Using Hermes Affinity Viewer software, a nuclear medicine (NM) radiologist estimated SUVmax, MTV, and TLG on screening PET/CT scans. A second NM reader and an automated deep-learning algorithm independently verified the first reader’s estimated values. They investigated whether these quantitative measures, which were presented as continuous variables, could forecast treatment response, event-free survival (EFS), and overall survival (OS). In order to identify cutpoints (c) as markers for risk stratification, they also adopted an outcome-oriented methodology. To assess the prediction effectiveness of the selected cutpoints, internal validation based on bootstrap was carried out. The difference between the AUC of the bootstrap sample and the AUC of the original dataset was used to determine optimism. It was thought that the optimism-corrected AUCs (ocAUC) would adjust for model overfitting. The major objective of the study was to determine how quantitative PET/CT measures would affect the prediction of complete metabolic response (CMR), which was the primary endpoint, and survival, which was the secondary endpoint. Due to their similar survival rates, they categorized patients with no metabolic response (NMR), disease progression (PD), and not evaluable (NE).

Out of 118, the 145 individuals who were recruited had PET/CT pictures that could be reviewed at the time of analysis. To calculate CMR, they first looked into the predictive power of PET/CT measures. Patients that achieved CMR had considerably decreased SUVmax, MTV, and TLG values. They found that MTV values had a good interobserver agreement. As continuous variables, log2(SUVmax), log2(MTV), and log2(TLG) were predictive of failure to achieve CMR (1-unit increase odds ratio [OR]= 1.06, 95%CI 1.01-1.10, P=0.008, AUC=0.666; OR=1.52, 95%CI 1.22-1.89, P=0.002, AUC=0.744; and OR=1.49, 95%CI 1.23-1.80, P<.0001, AUC=0.758; respectively). They identified the cutpoints of SUVmax ≥17 (OR=3.94, 95%CI 1.62-9.58, P=0.002, ocAUC= 0.653), MTV ≥43ml (OR=6.82, 95%CI 2.51-18.51, P=0.0002, ocAUC=0.716) and TLG ≥434 (OR=6.84, 95%CI=2.61-17.89, P<.001, ocAUC=0.720) as predictors of failure to achieve CMR. The relationship between PET/CT measurements and EFS and OS was then evaluated. According to treatment response, the median EFS and OS (months) were: CMR= not attained (NR) (95%CI 14.2-NR) & NR (95%CI 16.2-NR); PMR= 3.4 (95%CI 2.8-7.4) & 11.2 (95%CI 7.1-16.4); NMR/PD= 1.4 (95%CI 1.3-2) & 5.8 (95%CI 2.6-6.9); and NMR/PD= 1.4 (95%CI 1.3-2) & 5.8(95%CI 2.6-6.9); (P<.0001) respectively. log2(MTV) (hazard ratio (HR)=1.28, 95%CI 1.15-1.42, P<.0001 and HR=1.38, 95%CI 1.22-1.55, P<.0001) & log2(TLG) (HR=1.21, 95%CI 1.11-1.32, P<.0001 and HR=1.29, 95%CI 1.17-1.42, P<.0001) predicted shorter EFS and OS, respectively. Only a shorter OS was predicted by log2(SUVmax) (HR=1.02, 95%CI 1.00-1.03, P=0.033). SUVmax≥18 ((EFS: HR=1.65, 95%CI 1.06-2.55, P=0.027 & OS: HR=1.7, 95%CI 1.06-2.75, P=0.029), MTV ≥68ml (EFS: HR=3.02, 95%CI 1.94-4.7, P<.0001 & OS: HR=3.26, 95%CI 2.05-5.19, P<.0001) & TLG ≥479 (EFS: HR=2.34, 95%CI 1.52-3.6, P=.0001 & OS: HR=2.46, 95%CI 1.54-3.91, P=.0001).

In the analysis, they showed how PET/CT data could predict the development of rel/ref DLBCL while still having the ability to classify risks using both continuous and categorical variables. MTV is an imaging biomarker that can be extracted from standard PET/CT scans and is used to help doctors choose the patients who will respond best to loncastuximab tesirine.