Prediction of VOD Using Biomarkers of Endothelial Injury

Abstract

Veno-occlusive disease (VOD) of the liver occurs with a frequency of 5–15% after myeloablative conditioning and allogeneic stem cell transplantation (SCT). While risk factors for VOD are well known, predicting the occurrence of VOD in individuals remains challenging. Since the primary mechanism of injury in VOD is conditioning-related damage to hepatic sinusoidal endothelial cells and hepatocytes, we measured soluble biomarkers of endothelial injury in the peri-transplant period to determine if they correlated with the occurrence of VOD. Methods: 59 patients received cyclophosphamide (1800 mg/m 2 x 2) and TBI (14 Gy) as conditioning therapy, and tacrolimus with sirolimus (Sir+) or methotrexate (Sir-) as GVHD prophylaxis. Only patients with HLA-matched donors were included and selected for analysis based on the occurrence of VOD (VOD+ n=18, VOD- n=41), diagnosed by clinical, radiologic and pathologic criteria. Banked samples collected after conditioning but prior to SCT (day -1) and weekly after SCT (day 7, 14, 21) were thawed and analyzed by ELISA using commercially available kits and quantified using a VersaMax plate reader. Von Willebrand Factor (vWF) and thrombomodulin were assayed in plasma, and E-selectin and soluble intercellular adhesion molecule-1 (ICAM) were assayed in serum. Assays were performed in duplicate and results are the mean of two assays. Not all patients had every time point analyzed due to missing specimens. The within-sample results were compared using the 2-sided Wilcoxon rank-sum test, and the Bonferroni method was used to adjust for multiple comparisons (p value for significance=0.0125). Results: Comparing patients who did and did not develop VOD, patients with VOD had significantly elevated levels of vWF suggestive of endothelial damage at all timepoints prior to the development of VOD in comparison with controls (p≤0.0118, Figure). Thrombomodulin levels were predictive of VOD at all post-SCT timepoints (p≤0.0013, Figure). ICAM levels were significantly elevated up to Day 21 (p≤0.0028). E-selectin was less useful and statistically significant increases in levels were not observed. Since sirolimus has effects on endothelial function that may contribute to VOD through mechanisms different than the conditioning regimen, we stratified the analyses by sirolimus exposure, comparing Sir+VOD+ patients with Sir+VOD- controls. vWF levels in Sir+VOD+ patients were predictive for VOD at all timepoints when compared with controls (p≤0.003). Thrombomodulin levels were informative against controls for all post-SCT timepoints (p≤0.003). ICAM was informative as well (p≤0.003 pre-SCT, day 7, 14), while E-selectin levels were uninformative. The discriminative value of elevated serum and plasma biomarkers was limited to Sir+ patients in this small dataset, since other than some vWF timepoints, biomarkers could not distinguish Sir- patients who developed VOD patients from control groups without VOD. There were no differences in biomarkers among VOD- patients, suggesting that in the absence of VOD, markers of endothelial injury are not elevated even when sirolimus is used. Conclusions: Plasma vWF and thrombomodulin and serum ICAM elevations before and early after SCT can be used to predict the occurrence of VOD. These assays are most useful in patients receiving sirolimus. This analysis demonstrates the contribution of sirolimus to endothelial injury and VOD after SCT, and may help select patients in whom prophylactic or pre-emptive strategies against endothelial damage and VOD may be useful.