Development and validation of a nomogram based on perioperative risk factors for the prediction of early thrombotic complications following pediatric liver transplantation

To the Editor: Despite dramatic improvements in the prognosis of pediatric liver transplantation (LT), postoperative complications are still a leading cause of mortality in grafts and patients. Child recipients were shown to have a higher risk of thrombotic complications than in adult recipients, due to differences in etiology, narrowed hepatic vessels due to hypoplasia, more meticulous surgical techniques, and other factors.[1] Thrombotic complications include hepatic artery thrombosis (HAT), portal vein thrombosis (PVT), and venous outflow tract obstruction (VOTO) including inferior vena cava obstruction and hepatic venous outflow obstruction. Currently, there is still a 3–10% incidence of PVT and a 5–15% incidence of HAT, and VOTO varies between 1% and 13%.[2] Clinicians tend to overemphasize antithrombotic therapy to reduce the risk of thrombosis after pediatric LT, which leads to a higher risk of bleeding and warrants better antithrombotic management. Regarding the difficult problem of balancing thrombotic events and bleeding events, clinicians urgently need to perceive the risk of postoperative thrombosis to assist in optimizing antithrombotic regimens and there is no available quantitative risk assessment tool to intuitively identify high-risk children conveniently in clinical practice. In this retrospective research, we analyzed the relationship between perioperative parameters and postoperative thrombosis and constructed a prediction model to intuitively evaluate the risk of postoperative thrombotic complications in pediatric LT. A total of 219 pediatric recipients aged ≤16 years who received LT for end-stage liver disease from the Third Affiliated Hospital of Sun Yat-sen University between 2013 and 2022 were screened and 202 patients were finally included in this analysis. The study protocol was approved by the Ethics Committee of the Third Affiliated Hospital of Sun Yat-sen University (No. II2023-225-01) with a waiver of informed consent and adhered to the 1975 Declaration of Helsinki. An early posttransplant thrombotic complication was defined as a hepatic thrombotic event that occurred within three months of transplantation (absent during surgery, but confirmed on posttransplant examinations by either Doppler-ultrasound imaging, computed tomography, or through surgery exploration), including HAT, PVT, and VOTO. Patient survival was analyzed up to November 2022 or until failure or death of the graft and patient, or until loss to follow-up. After randomizing the entire dataset into training and validation cohorts, 152 patients were included in the training cohort and 50 patients were included in the validation cohort. Univariate and multivariate analyses were performed on thrombotic complications, PVT and HAT, respectively. A stepwise backward selection process on the basis of Akaike information criterion (AIC) analysis was used to select variables that finally entered the nomogram. The performance of the nomogram in the training and validation cohorts was evaluated mainly from three perspectives: receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA). Detailed descriptions of the data and statistical methods are included in the Supplementary Methods, https://links.lww.com/CM9/C4. The antithrombotic regimen implemented in our center is as follows: in the case of ensuring complete hemostasis during surgery, therapy begins immediately after surgery or on the first day after surgery, which includes heparin sodium 20 U·kg−1·h−1 to maintain activated partial thromboplastin time (APTT) for 60–80 s, bridging warfarin on days 4–7, and maintaining international normalized ratio (INR) 1.5–2.0 until 3 months after transplantation. In the early postoperative three-month period, 36 children developed thrombotic complications: PVT in 18 (8.9%), HAT in 13 (6.4%), and VOTO in 6 (2.9%). According to Kaplan–Meier survival analysis, early thrombotic complications had a significant impact on the mortality of child recipients. Further survival analysis for different types of thrombosis showed worse prognosis in patients with PVT and HAT [Supplementary Figure 1, https://links.lww.com/CM9/C4]. These data are consistent with clinical reality and early postoperative thrombosis prompted clinicians to emphasize antithrombotic therapy. A good perception for the identification of high-risk children suggested by the prediction model may assist clinicians in adjusting antithrombotic regimens to reduce the risk of thrombotic complications and avoid bleeding complications caused by excessive antithrombotic therapy for children with low risk of thrombosis. For detailed perioperative clinical characteristics and the demographics of donors and recipients, a total of 44 perioperative variables were summarized in the whole cohort, training cohort, and validation cohort [Supplementary Table 1, https://links.lww.com/CM9/C4]. Variables were assessed in univariate and multivariate analyses, and the portal vein diameter of donor side (PV donor) (odds ratio [OR] = 0.25, 95% confidence interval [CI] 0.06–0.97), operation time (OR = 7.52, 95% CI: 1.86–33.29), and cold ischemia time (CIT) (OR = 5.09, 95% CI: 1.55–18.97) were identified as independent prognostic factors for early postoperative thrombotic complications [Supplementary Table 2 and Supplementary Figure 2, https://links.lww.com/CM9/C4]. For PVT, operation time (OR = 5.41, 95% CI: 1.84–15.94), CIT (OR = 4.59, 95% CI: 1.39–15.23) and portal vein diameter of recipient side (PV recipient) (>4.50 mm) (OR = 0.22, 95% CI: 0.08–0.64) were filtered out as independent predictors. With regard to HAT, operation time (OR = 14.00, 95% CI: 2.96–66.29) and hepatic artery diameter of donor side (HA donor) (>2.00 mm) (OR = 0.17, 95% CI: 0.04–0.71) were the most influential predictors associated with HAT [Supplementary Tables 3 and 4, https://links.lww.com/CM9/C4]. To construct a prediction model for early postoperative thrombotic complications, an AIC-based stepwise regression was applied based on the results of the univariate logistic regression analysis, and the model contained CIT, operation time, HA donor and hepatic artery diameter of the recipient side (HA recipient), portal vein diameter of anastomosis (PV anastomosis), and PV donor [Figure 1A]. The model was built by incorporating the six variables and is shown as a nomogram. The ROC curve for the training and validation cohorts is presented [Figure 1B,C], with an area under the curve (AUC) of 0.886 in the training cohort and an AUC of 0.859 in the validation cohort, which showed a favorable discrimination. Calibration plots and the DCA curves also had a favorable performance in the two cohorts [Figure 1D–G]. The optimal cutoff value was 0.18 obtained by the Youden index principle, corresponding to 210 points on the nomogram. In the training cohort, this corresponds to a sensitivity of 82% and a specificity of 81%. In the validation cohort, the sensitivity was 75% and the specificity was 90%. Therefore, we propose that when the score exceeds 210 points and the risk is higher than 18%, it can be considered as a child with high risk of thrombotic complications.Figure 1: (A) Nomogram for assessing the risk of early postoperative thrombotic complications in child recipients. (B) ROC curve for evaluating discrimination performance in the training cohort; AUC represented by blue line: 0.886. (C) ROC curve in the validation cohort; AUC represented by blue line: 0.859. (D,E) Calibration curve for training cohort and validation cohort. (F,G) DCA of the nomogram in the training cohort and the validation cohort. AUC: Area under the curve; CIT: Cold ischemia time; DCA: Decision curve analysis; HA donor: Hepatic artery diameter of donor side; HA recipient: Hepatic artery diameter of recipient side; PV anastomosis: Portal vein diameter of anastomosis; PV donor: Portal vein diameter of donor side. ROC: Receiver operating characteristic.In our model, CIT >360 min and operation time >540 min were primary predisposing factors not only for thrombotic complications but also for PVT and HAT. Traditionally, a CIT exceeding 12 h was considered a critical threshold, above which CIT would lead to an increased risk of poor-quality organs.[3] Our research showed a more limited CIT, suggesting that prolonged CIT has a pronounced impact on the formation of thrombi, which deserves surgeons' attention. A longer operation time is a complicated problem caused by many factors such as a complex intraoperative situation, more intraoperative blood loss, difficult technical operation, and other unpredictable causes. Nevertheless, it is meaningful to realize that a shorter operation time and CIT are beneficial for reducing the risk of postoperative complications because they can be consciously regulated. Clinicians may need to provide more intensive care and more emphasis on anti-thrombotic therapy in children with these two factors. One of the most important factors leading to thrombotic complications is related to the condition of hepatic vessels. For child recipients, it is more prevalent to confront a diameter discrepancy between donor and recipient vessels, which induces a host of attendant problems such as more difficult suture procedures, additional vascular grafts, low blood flow, intraoperative clots, and more frequent angulation of vascular anastomoses, thereby further increasing the risk of postoperative thrombosis. Portal vein diameter <4 mm and hepatic artery diameter <2 mm were once suggested to be associated with PVT and HAT in previous researchers' results.[4,5] Nonetheless, a more detailed analysis involving the donor and recipient sides of hepatic vessels and the ratio between them has rarely been presented. In our analysis, more detailed vessel diameter data were included (PV diameter of donor side, recipient side, and anastomosis; HA diameter of donor side and recipient side). According to the final results, whether it was the overall thrombotic complications or individual PVT and HAT, the diameters of the portal vein and hepatic artery had a significant impact on the formation of the thrombus. It must be acknowledged that this study has its limitations. First, although our research proposed a novel quantitative assessment tool for the risk of postoperative thrombotic complications, only the prediction model of thrombotic complications was developed this time due to the limited number of cases of pediatric LT in our center. In the next step, we will further try to conduct a multicenter retrospective study to include more cases to construct a prediction model for different thrombotic complications. Second, risk factors for VOTO were not presented due to its small number of events (only six children developed VOTO and all of them survived after receiving symptomatic treatment). In summary, the current study identified risk factors associated with early thrombotic complications after pediatric LT. Furthermore, a prediction nomogram with desirable accuracy and good clinical utility was successfully developed to identify child recipients with high risk of thrombotic complications. Further research is required to validate these selected factors and confirm the feasibility of the nomogram. Acknowledgements None. Funding This work was supported by grants from the National Natural Science Foundation of China (Nos. 81900597, 81970567), the Natural Science Foundation of Guangdong Province, China (Nos. 2019A1515011698, 2021A1515012136, 2021A1515010571, 2018A030313043), Science and Technology Program of Guangzhou City (No. 201803040005), National Bioengineering Research Center Cultivation Platform (No. WW201905), and the Major Talent Project Cultivation Plan Project (No. P02093).