A Prospective Comparison Analysis of Blood Biomarkers for the Diagnosis and Prediction of Sinusoidal Obstruction Syndrome

Background: Sinusoidal obstruction syndrome (SOS) remains a significant, potentially lethal complication after allogeneic hematopoietic cell transplantation (allo-HCT). Although a liver biopsy is required to diagnose SOS accurately, it is not considered a mandatory evaluation due to its invasiveness. Therefore, the Seattle and Baltimore criteria with minor revisions have been widely used. However, the diagnostic accuracy of those criteria is insufficient. A noninvasive and more accurate diagnostic strategy is necessary. A number of studies have reported several candidate blood biomarkers for the diagnosis and prediction of SOS. However, which biomarkers or combination thereof are most useful for the diagnosis and prediction of SOS is unclear. We explored the best diagnostic and predictive biomarkers/combination among previously reported biomarkers for SOS using a stringent definition based on a liver biopsy.

Methods: We performed this single-center prospective observational study in patients who received allo-HCT from April 2014 to February 2019. Seven biomarkers (PAI-1, P3P, ferritin, total bilirubin [T-bil], direct bilirubin [D-bil], brain natriuretic peptide [BNP], and protein C activity) were examined at pre-conditioning, at days 5 and 30, and at the onset of CTCAE grade ≥2 liver disorder after allo-HCT. We described how to diagnose definitive SOS (Fig. 1). A logistic regression (LR) model and the area under the receiver operating characteristic curves (AUC) were used to compare the seven biomarkers in the diagnosis and prediction of definitive SOS. The sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio for the SOS diagnosis and prediction were also calculated by the best cut-off values, using the Youden index. Results of statistical tests with a p < 0.05 were considered significant.

Results: A total of 180 patients were included. The median age was 48 (range: 16-68) years old. Forty-eight patients developed CTCAE grade ≥2 liver disorders. Of these, 10 were diagnosed with definitive SOS. The results of LR and AUC analyses of the SOS diagnosis and prediction are shown in the Table. PAI-1, P3P, ferritin, T-bil, and D-bil were found to be significant diagnostic markers for SOS. Among these, PAI-1 showed the highest AUC (0.85; 95% confidence interval [CI], 0.67-1.00). Furthermore, PAI-1, P3P, ferritin, T-bil, D-bil, and BNP were significant predictors for SOS. Among these biomarkers, P3P showed the highest AUC (0.82; 95% CI, 0.67-0.97). To perform further comparisons using multivariable models in SOS prediction, we first constructed a base model including the times of allo-HCT, disease status, and conditioning intensity. The AUC of the base model was 0.66 (95% CI, 0.48-0.84). After adding P3P to the base model, the AUC significantly improved to 0.88 (95% CI, 0.76-1.00) (p = 0.049). In the kinetics analysis of biomarkers, notably, PAI-1 and P3P increased over the peri-transplant period only in patients with definitive SOS (Fig. 2). In contrast, those values in patients with liver disorders other than SOS or without liver disorders did not show significant kinetic characteristics in the allo-HCT period.

Discussion: SOS is attributed to toxic injury of the sinusoidal endothelial cells. PAI-1 is a known endothelial factor, released when the endothelial cells are damaged. This could be why PAI-1 was considered useful for the SOS diagnosis. P3P has been shown to be a sensitive biomarker for liver fibrosis. Furthermore, fibrous alterations in the hepatic remodeling process are well-known significant features for patients with SOS. Thus, liver fibrosis may pathophysiologically be a risk factor for SOS, and P3P may allow clinicians to detect SOS-high-risk patients with high accuracy, even at the time of allo-HCT when preclinical liver fibrosis can exist.

Conclusion: We demonstrated that PAI-1 and P3P were the most useful biomarkers for the diagnosis and prediction of SOS, respectively.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal