From May 2020 to September 2024, 58 patients with non-metastatic muscle-invasive urothelial bladder cancer (MIBC), diagnosed via transurethral resection of bladder tumor (TURBT) and/or re-resection (Re-TURBT), were referred for neoadjuvant chemotherapy (NAC) followed by radical cystectomy (RC) and eventually extended pelvic lymph node dissection (ePLND). All participants underwent multi-parametric MRI (mpMRI) using a 3 Tesla scanner (Siemens Healthineers, MAGNETOM Vida; GE Healthcare, Dis- covery 750) for the assessment of VI-RADS and nacVI-RADS score, both 2–4 weeks prior to staging TURBT and after completing the final NAC cycle.
The NacVI-RADS score was evaluated in comparison to final histopathological findings and tumor regression grade (TRG) derived from RC specimens. TURBT and RC pathological samples were analyzed by an experienced uro-pathologist with 15 years of exprience in the field. The AI-powered software was applied to the first CT images to calculate the Skeletal Muscle Index (SMI-L3) in order to assess sarcopenia. For each patient, at the primary MRI, the lesion with the highest VI-RADS score was considered as the index. A VI-RADS cut-off score of ≥3 to define MIBC was assumed. CT images before neoadjuvant chemotherapy were acquired on a multidetector CT scanner (Somatom Sensation 64; Siemens Healthineers, Erlangen, Germany). These CT scans, routinely performed as part of the standard staging protocol for urothelial tumors prior to initiating neoadjuvant therapy, were utilized for body composition analysis.
The AI-powered software was applied to the first CT images to calculate the Skeletal Muscle Index (SMI-L3) in order to assess sarcopenia.
The Quantib body composition® software (Rotterdam, Netherlands) was used to measure body composition quantitatively. This software analyzed CT images of patients taken during staging CT examinations stored in our institutional picture archiving and communication system (PACS) selecting just the non-contrast phase. The software focused on the L3 vertebral body level and automatically segmented the images to determine the areas of the abdominal, psoas, and long spine muscles. Finally, the software generated in a few minutes a form displaying the relevant values, as shown in Figure 1. Manual corrections were not made, and patients with grossly incorrect segmentations were excluded from the study. There was no specific definition for what constituted an incorrect segmentation; all segmentations were considered either of high quality or had only a small portion of the muscle volume accurately delineated.
In line with the EAU guidelines, patients received 3-to-4 cycles of cisplatin-based combination chemotherapy consisting of Cisplatin plus Gemcitabine prior to cystectomy.
Based on CT-AI powered software the SMA (Skeletal Muscle Area) was obtained by summing the muscle areas of the psoas muscle, abdominal muscle, and long spine muscle. SMI (Skeletal Muscle Index) was obtained by normalizing the SMA by the patients’ squared height and represents a represents a more appropriate tool to define sarcopenia than the psoas muscle volume. According to the literature, we defined sarcopenia as SMI <50 cm2/m2 for men and SMI <35 cm2/m2 for women, which was the cohort’s mean cut-off in our series9.
The assessment of the follow-up MRI, after the NAC cycles, was performed using a nacVI-RADS algorithm for evaluation of response to therapy (Neoadjuvant Chemotherapy VI-RADS—nacVI-RADS). NacVIRADS categorically define complete Radiological Response (RaR), based on prior VI-RADS score, presence of residual disease, tumor size, and infiltration of the muscularis propria.
Complete RaR was labeled in case of absence of intravesical lesions after NAC completion (nacVI-RADS 1–2). Partial RaR was defned as any radiological downstaging with evidence of residual disease, with or without evidence of muscle invasiveness (nacVI-RADS 3–4). No RaR was defned as lack of radiological downstaging and/or upstaging, regardless prior VI-RADS scoring (nacVI-RADS 5).