Diagnostic performance of contrast-enhanced CT-scan in sinusoidal obstruction syndrome induced by chemotherapy of colorectal liver metastases: Radio-pathological correlation

Purpose Sinusoidal obstruction syndrome (SOS) is a likely side effect of colorectal liver metastases (CRLM) chemotherapy. This study aimed to assess computed tomography scan (CT-scan) performance for SOS diagnosis for patients receiving neoadjuvant chemotherapy (NC) prior to CRLM surgery, comparing obtained results with pathological gold standard. Methods Preoperative CT-scans of 67 patients who had received a NC prior to liver resection for CRLM from 2011 to 2016 were retrospectively analysed. Positive diagnosis and severity of SOS were established after consensual review of the slides by three pathologists. Preoperative CT-scans were separately interpreted by two radiologists and evocative signs of SOS were sought, defined according to a literature review and operators experience. In order to identify SOS predictors, univariate analysis and multivariate logistic regression were used to study CT-scan signs and pathological results correlation. Results Twenty-nine patient (43%) had an SOS, 22 (33%) were low-grade and 7 (10%) were high-grade. All patient had received a median of 6 cures (3–27) containing Oxaliplatin for 53 (79%) of them. In univariate analysis, hepatic heterogeneity (p < 0.001), puddle-like or micronodular appearance (p < 0.001), peripheral distribution of heterogeneity (p = 0.085), clover-like sign (p = 0.02), splenomegaly (p = 0.0026), spleen volume increase ≥30% (p = 0.04) or splenic length increase ≥15% (p = 0.04), as well as the subjective impression of the observer (P < 0.001) were significantly associated with SOS diagnosis. In multivariate analysis, clover-like sign (OR 1.87, 95% CI 1.18–2.95, p = 0.0081), increase in spleen volume ≥30% (OR 1.29, 95% CI 1.01–1.64, p = 0.04), and the peripheral distribution of heterogeneity (OR 1.53, 95% CI 1.21–1.94, p < 0.001) were independent SOS predictors. The area under the ROC curve was 0.804. The inter-observer agreement for SOS diagnosis was moderate (Kappa = 0.546). Conclusion CT-scan can detect suggestive signs of SOS in patients receiving chemotherapy for CRLM. By integrating clinical and biological information into CT-scan data, it may be fruitful to create a positive diagnostic and severity score for chemotherapy-induced SOS.