BMC Medical Imaging
Development and internal evaluation of an integrated [18F]FDG PET/CT-clinicopathological nomogram for predicting progression-free survival in MALT lymphoma
By Yanmei Han, Yajing Zhang, Ruolin Wu and 3 others.
Abstract
Mucosa-associated lymphoid tissue (MALT) lymphoma is an indolent B-cell lymphoma, yet a subset of patients develop early disease progression or relapse. Early identification of prognostic factors is crucial for clinical risk stratification, but existing prognostic systems do not incorporate metabolic imaging parameters. This study aimed to develop and internally evaluate a prognostic model integrating [ 18 F]fluorodeoxyglucose ([ 18 F]FDG) positron emission tomography/computed tomography (PET/CT) with clinicopathological variables to predict progression-free survival (PFS) in patients with MALT lymphoma. In this retrospective study, 98 patients with histologically confirmed MALT lymphoma who underwent baseline [ 18 F]FDG PET/CT between January 2020 and December 2024 were analyzed. The least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation was used for variable selection, followed by multivariable Cox regression to develop a model for predicting progression-free survival (PFS) in patients with MALT lymphoma. Internal validation was performed using a full-pipeline bootstrap procedure with 500 resamples. Time-dependent receiver operating characteristic (time-ROC) curves and the concordance index (C-index) were estimated to evaluate the model’s predictive performance. Exploratory risk re-stratification was performed among patients in the MALT-IPI intermediate-risk group. Among 17 candidate predictors and 28 PFS events, LASSO regression selected five variables: age, primary disease site, SUVmax, Ki-67 index, and ECOG performance status. These variables were incorporated into the integrated prognostic model. The apparent C-index of the five-variable model was 0.814 (95% CI, 0.738–0.891). Full-pipeline bootstrap validation showed an optimism-corrected C-index of 0.740 (95% uncertainty interval, 0.662–0.757). Area under the time-ROC curve values were 0.766 at 1 year, 0.872 at 2 years, and 0.883 at 3 years. In the MALT-IPI intermediate-risk subgroup, the integrated model further separated patients into lower- and higher-risk groups with different observed PFS. We developed and internally evaluated a prognostic model integrating age, primary disease site, SUVmax, Ki-67 index, and ECOG performance status for predicting PFS in patients with MALT lymphoma. The model showed moderate optimism-corrected discrimination and exploratory risk separation within the MALT-IPI intermediate-risk subgroup.



