Abstract
Total metabolic tumor volume (TMTV) is a promising quantitative biomarker for therapy assessment and prognosis in Hodgkin Lymphoma affected patients that allows prediction of patient outcome. The aim of this study was to evaluate the TMTV reproducibility between different sources of variability in tumor delimitation such as SUV-based thresholds (2.5, 41% and 50%) and software tools (Beth Israel plugin (BI) and LIFEx). Effect of contouring procedure both including single and multiple regions of interest was also studied in patients with multiple lesions, and optimal cut-offs for each studied method were displayed to compare the effect on prognosis. Strong alikeness in TMTV was found for 2.5 under software choice. Best accuracy in contouring compared to visual assessment of the disease was found for BI multiple ROI and LIFEx single ROI drawing. Similar cut-offs were found between both software for all considered thresholds, but best resemblance and highest cut-off due to an overestimation of the TMTV was found for 2.5 SUV. Our findings suggest that optimal reproducibility in TMTV is found for SUV > 2.5 threshold under choice of contouring methodology or software tool, meaning that overestimation of the TMTV threshold using 2.5 looks to be preferable than underestimation with 41% and 50%.
Highlights
Finding a common unified method is necessary for forthcoming studies which could allow comparisons between institutions leading to optimal patient management within medical centers[5]
Receiver Operating Characteristics (ROC) curves were drawn using the same threshold to show the differences between both Total metabolic tumor volume (TMTV) cut-offs and to contrast HL progression. Comparison between both PET machines Discovery ST (D-ST) and Discovery IQ (D-IQ) showed no statistical significance in TMTV values neither for BI nor LIFEx (p = 0.26 and p = 0.27 for 2.5, p = 0.07 and p = 0.16 for 41%, and p = 0.09 and p = 0.23 for 50%, respectively)
Statistical significance could be observed between TMTV obtained using different thresholds (TMTV2.5, TMTV41%, and TMTV50%) within the same software (p < 10–20), as well as when comparing both software applying the same threshold for 41% and 50% (p < 0.02)
Summary
Finding a common unified method is necessary for forthcoming studies which could allow comparisons between institutions leading to optimal patient management within medical centers[5]. Threshold choice within the software has been proven to influence the TMTV, though no significant differences between predicted prognosis have been found[6]. Contrasting the effect of software packages and different thresholds on the TMTV estimation could help predict future patients outcome, basing their survival foreshadow on the parameters given by the chosen methodology[7]. Contouring was performed using different software tools (BI and LIFEx6,9) and TMTV discrimination was achieved using SUV based thresholds (2.5 SUV, 41% SUVmax, and 50% SUVmax). Both potential sources of bias were analyzed to compare TMTV reproducibility. Research focused on the effect of evaluated parameters on the prognosis of the disease to predict patient outcome
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
