Photon-counting computed tomography and magnetic resonance imaging in rectal cancer: reassessment of normalized iodine concentration and apparent diffusion coefficient in an expanded cohort

Abstract

Abstract Objectives To reassess pretreatment photon-counting computed tomography (PCCT)-derived normalized iodine concentration (NIC) in an expanded rectal-cancer cohort, using magnetic resonance imaging (MRI)-derived apparent diffusion coefficient (ADC) as a within-cohort reference marker for histopathologic treatment response. Methods This single-center retrospective study included 63 patients with rectal adenocarcinoma who underwent pretreatment pelvic MRI and contrast-enhanced PCCT, followed by neoadjuvant treatment and surgical resection. Dworak tumor regression grades 0–1 were classified as poor response ( n = 22) and grades 2–4 as moderate/good response ( n = 41). MRI and PCCT were performed a median of 3 days apart (IQR, 1–4.5). The previously reported NIC threshold of ≥ 0.36 was reassessed without re-optimization. Twenty-one of 63 patients overlapped with the preceding pilot cohort; therefore, the present study is an expanded reassessment rather than an independent external validation. Results Twenty-two patients showed poor response and 41 moderate/good response. ADC interobserver agreement was excellent (ICC, 0.959; 95% confidence interval [CI], 0.90–0.98). In the 61 patients with paired numeric NIC measurements from both observers, NIC interobserver agreement was also excellent (ICC, 0.905; bootstrap 95% CI, 0.858–0.936). Mean ADC was lower in poor responders than in moderate/good responders (0.876 ± 0.186 vs. 1.035 ± 0.208 × 10⁻³ mm²/s; p = 0.003; AUC, 0.738; 95% CI, 0.608–0.867). NIC did not differ between groups (median, 0.313 vs. 0.370; p = 0.91; AUC, 0.509; 95% CI, 0.362–0.656). Applying the previously published NIC threshold ≥ 0.36 yielded sensitivity 53.7%, specificity 63.6%, and accuracy 57.1% for Dworak grades 2–4. Adding NIC to ADC did not improve model fit (likelihood-ratio p = 0.601). Conclusion In this exploratory expanded cohort, lower pretreatment ADC was associated with Dworak grades 0–1. Single-phase portal-venous NIC measured with the described two-dimensional ROI approach did not reproduce the high performance reported in the pilot study and showed no evidence of improved model fit beyond ADC in this cohort. These findings require independent prospective validation.