A systematic review and meta-analysis of the radiological features predicting lateral pelvic lymph node positivity in rectal cancers

Lateral pelvic lymph node (LPLN) metastasis is a major contributor to the local recurrence in rectal cancer. Management of LPLN metastasis remains controversial. In the neoadjuvant chemoradiotherapy (nCRT) setting, pre-treatment nodal size remains the recommended sole criteria for surgical resection of LPLN despite increasing evidence of its poor sensitivity and specificity. We therefore aimed to review the literature for radiological diagnostic features which are predictive of LPLN involvement on histopathological assessment. A systematic literature search was conducted across MEDLINE, Embase, Clinicaltrials.gov and Cochrane databases. A variety of radiological features were studied: size, size in combination with preoperative therapy, pretreatment SA diameter, morphology and miscellaneous – radiomics, uptake scans and nomograms. A meta-analysis was then conducted for pre-nCRT short axis (SA) diameter; results not included in the meta-analysis were covered with a narrative review. Quality assessment was carried out using QUADAS-2 and GRACE tools. A total of 63 papers were included in the final report. Pre-neoadjuvant SA diameter of 7 mm was a significant cut-off. A forest plot from these 5 papers produced a pooled odds ratio of 10.75 [95% CI 7.97, 14.49] favouring nodal involvement, although heterogeneity was substantial (I 2 : 68%). Radiomics and machine learning significantly outperformed traditional radiological features with high sensitivity and specificity highlighting potential for future diagnostic development. A range of radiological features have been studied in the literature; these features remain poor predictors of LPLN positivity in terms of sensitivity and specificity. Results are heterogeneous and further research including into radiomics and artificial intelligence aided models is required.