A grading system based on multiparametric MRI-predicted tumor perfusion, invasiveness and texture (GOLDS-PIT) for surgical difficulty assessment in pituitary macroadenomas: A prospective single-centre cohort study.

Abstract

To develop and validate a novel grading system (Grading system Of 4-Level assessing Difficulty of Surgery based on Perfusion, Invasiveness and Texture, GOLDS-PIT) for assessing surgical difficulty based on multiparametric MRI-predicted perfusion, invasiveness, and texture in patients with pituitary macroadenomas (PMAs). Preoperative multi-parametric MRI were retrospectively collected from 240 patients (Cohort A) with PMAs to construct a radiomics model predicting tumor perfusion (RP-model) and morphological-radiomic model predicting texture (MRT-model). The GOLDS-PIT based on multiparametric MRI-predicted tumor perfusion, invasiveness and texture was then developed with cohort A and validated with another 162 cases of internal cohort B. The AUC of the RP- and MRT- model in the validation cohort was 0.906 and 0.908 respectively. A sixteen-score GOLDS-PIT stratified surgical difficulty into low, medium, high and extremely high categories. In validation cohort, combined approaches were selected in more patients in extremely high and high difficulty groups significantly higher than low and moderate difficulty groups (33% vs 1%, p<0.001). These groups had the lower gross total resection rate (84% vs 46%, p<0.001) , higher severe complication rate (50% vs 19%, p<0.001). and higher surgical resource consumption: intraoperative blood loss[600(363-1425)mL vs 300(200-400)mL, p<0.001], operative duration [5.9(4.0-7.0)h vs 3.0(2.5-4.0)h, p<0.001], and hospitalization cost [¥96,433 vs ¥67,264, p<0.001]. We developed and validated a novel surgical difficulty grading system for PMAs, which can assist surgeons in surgical planning and anticipating surgical outcomes, providing a practical reference for determining appropriate resection strategies.