Objectives

Compared different Standardised Uptake Value (SUV) formulations in paediatric oncology patients.
Determined which SUV correction method shows the least dependency on body weight.
Demonstrated that SUV corrected for Body Surface Area (BSA) has the least dependency on body weight, providing more consistent assessments of metabolic activity in children.

Methodology

Retrospective, single-center, cross-sectional study.
Analyzed [18F]FDG PET/CT scans of 461 paediatric oncology patients.
Used EARL1 reconstructions.
Measured SUVmean in liver and blood pool (aorta) using manually placed Regions of Interest (ROIs).
Calculated SUV corrected for body weight (SUVBW), Lean Body Mass (LBM) according to James (SUVLBMJames) and Janmahasatian (SUVLBMJanma), and BSA according to DuBois (SUVBSADuBois) and Haycock (SUVBSAHaycock).
Used coefficient of determination (r2) from linear regression to assess correlation between SUV and body weight.

SUVBW exhibited the strongest correlation with body weight: r2 = 0.65 (liver) and r2 = 0.50 (blood pool).
SUVBSADuBois and SUVBSAHaycock had the weakest correlation: r2 = 0.09 (liver) and r2 = 0.06 (blood pool).
SUVLBMJames and SUVLBMJanma had moderate correlations: r2 = 0.51 and 0.47 (liver), and r2 = 0.44 and 0.42 (blood pool), respectively.
Relative SUV dependence on body weight (a/b in the linear regression equation) was lowest for SUVBSADuBois and SUVBSAHaycock (a/b = 0.003 for both liver and blood pool).

The study is limited by its retrospective design and the inability to analyze changes in SUV measurements over time within individual patients. This is a crucial point for treatment response assessment.
Although the study mentions mitigating the influence of chemotherapy by also measuring SUV in the blood pool, the impact of prior chemotherapy on liver SUV measurements isn't fully quantified or statistically adjusted for. A more rigorous control for chemotherapy exposure (e.g., time since last chemotherapy, type of chemotherapy) would strengthen the findings.
The study did not account for potential confounders like glucose levels and tumor load, which are known to influence [18F]FDG distribution. Future studies should include these variables in the analysis.
The conclusion recommends against using SUVBW, which is well-supported. However, the suggestion to use SUVBSA, while supported by the data, could be strengthened by acknowledging the practical challenges of implementing BSA-based corrections in clinical settings and comparing its feasibility with LBM-based corrections (especially considering PERCIST guidelines).
The discussion mentions potential future use of dual-energy CT or machine learning, but this is speculative and not directly related to the study's core findings.