Anatomical risk stratification for major portal vein complications in dual portal vein living donor liver transplantation: a retrospective cohort study Authors: Hyun Hwa Choi, Jae-Yoon Kim, Jiyoung Kim, Jaewon Lee, Su young Hong, YoungRok Choi, Kwang-Woong Lee, Suk Kyun Hong Published: 2026 ABSTRACT PurposeRight lobe living donor liver transplantation (LDLT) with dual portal veins (PVs) remains technically challenging. This study aimed to identify independent risk factors for PV complications.MethodsWe retrospectively analyzed 111 recipients of dual PV LDLT between 2011 and 2020. Recipient characteristics, anatomical geometry, and surgical factors were evaluated. Outcomes were overall PV complications and major PV complications (Clavien-Dindo grade ≥III). Logistic regression was performed.ResultsPV complications developed in 41 patients (36.9%), including 16 major events (14.4%). Univariate analysis revealed associations with right posterior PV (RPPV) diameter, axial angle, and coronal angle. On multivariate analysis, larger RPPV diameter (odds ratio [OR], 1.79; P = 0.041) and wider axial angle (OR, 1.08; P = 0.015) were independent predictors of major PV complications. Reconstruction method was not significant. Patients with overall major Clavien-Dindo grade ≥III complications had inferior 100-month survival (80% vs. 100%; P = 0.014, log-rank test).ConclusionIn dual PV LDLT, anatomical geometry—specifically RPPV diameter and axial angle—independently predicts major PV complications, whereas surgical technique does not. Preoperative 3-dimensional imaging and anatomical risk stratification should inform donor selection and surgical planning. INTRODUCTION Living donor liver transplantation (LDLT) has become essential to address the persistent shortage of deceased donor organs [123]. Among its technical challenges, right lobe grafts with dual portal vein (PV) orifices represent one of the most demanding anatomical variations [456]. Reconstruction of dual PVs typically requires direct venoplasty, Y-graft interposition, or patch grafting, all of which are technically feasible and have shown comparable survival outcomes [678910]. However, our previous comparative analysis and other reports suggest that the choice of reconstruction technique alone does not account for the development of PV complications. Instead, anatomical geometry and donor–recipient characteristics may play a more decisive role, influencing the risks of anastomotic stenosis, thrombosis, and long-term graft dysfunction [681112]. This study, therefore, aimed to determine the independent predictors of PV complications in LDLT using dual PV grafts, focusing on the impact of anatomical factors rather than surgical technique. By clarifying these risk factors, we sought to provide practical insights for preoperative planning and improved outcomes in this technically complex setting. METHODS Ethics statement This retrospective analysis was reviewed and approved by the Institutional Review Board of Seoul National University Hospital (No. 2307-161-1452). The research adhered to the ethical principles outlined in the Declaration of Helsinki (revised 2013). Because only existing clinical data were used without patient identifiers, the requirement for individual informed consent was waived. Study design and patient selection Between January 2011 and December 2020, a total of 1,028 LDLTs were performed at Seoul National University Hospital, a high-volume tertiary referral center. For this study, we excluded patients who underwent non-right lobe graft transplantation, those with a single PV orifice, and those who required retransplantation. Consequently, 111 recipients who received right lobe grafts with dual PV orifices were included in the analysis. In our earlier study, we analyzed outcomes according to the method of dual PV reconstruction. In that cohort, patients were stratified as follows: 62 (55.9%) underwent direct venoplasty, 39 (35.1%) received Y-graft interposition, and 10 (9.0%) underwent patch graft reconstruction [13]. The median follow-up duration was 73.7 months (range, 5.4–155.4 months), allowing for a robust evaluation of long-term outcomes [13]. In this study, we go beyond comparing the 3 method types addressed in our previous paper and aim to identify risk factors for PV complications through both univariate and multivariate analyses. Anatomical assessment of dual portal vein configuration Donor PV configurations were assessed using preoperative computed tomography and magnetic resonance imaging. For each case, image slices that most clearly depicted both the right anterior (RAPV) and right posterior (RPPV) branches were selected. The spatial relationship of these 2 branches was quantified by measuring their separation distance and the branching angle on both axial and coronal planes. The branching angles on both the axial and coronal planes were measured. The specific imaging protocol and measurement technique have been described in detail in our previous report [13]. Donors with ambiguous anatomy or a common channel configuration were excluded from this assessment. To ensure consistency in geometric measurements, all anatomical assessments were performed by hepatobiliary transplant specialists using a predefined standardized protocol. To evaluate reproducibility, both interobserver and intraobserver reliability tests were conducted in a randomly selected subset of 30 patients, demonstrating excellent agreement across axial and coronal angles and distances. Clinical and anatomical variables Graft-related variables PV geometry was characterized by measuring the diameters of the RAPV and RPPV, the interbranch distances in axial and coronal planes, and the RAPV–RPPV branching angles in both planes. Measurements were obtained from preoperative imaging and confirmed intraoperatively. Surgical variables Operative factors included total recipient operation time, bench preparation time, and PV anastomosis time. The method of PV reconstruction (direct venoplasty, Y-graft interposition, or patch grafting) was documented [13]. Additional procedures such as shunt ligation, splenectomy, and intraoperative thrombectomy were recorded when performed. Postoperative variables Outcomes assessed included length of hospital stay, overall and major Clavien-Dindo (CD) complications, and PV-related complications. PV complications were defined as stenosis or thrombosis, while major PV complications referred to CD grade ≥III events requiring interventional treatment, including stent placement, balloon dilatation, or thrombectomy. Definition of outcomes The primary outcome of interest was the occurrence of major PV complications, defined as CD grade III or higher events directly attributable to PV anastomotic problems such as stenosis, thrombosis, or flow impairment necessitating intervention. Secondary outcomes included the time to complication and overall graft survival. Statistical analysis Continuous data are summarized as mean values with standard deviations. Depending on the distribution pattern, comparisons between groups were made using either the Student t-test or the Mann-Whitney U-test. Categorical variables were analyzed with the chi-square test or Fisher exact test, as appropriate. Variables showing a P-value below 0.10 in the univariate analysis were subsequently included in a multivariate logistic regression model to identify factors independently associated with the outcome. A P-value of less than 0.05 was considered statistically significant. All statistical analyses were carried out using IBM SPSS Statistics ver. 25.0 (IBM Corp.). RESULTS Cohort characteristics and outcomes Among the 111 recipients of LDLT with dual PV grafts, the mean RAPV and RPPV diameters were 6.3 ± 1.3 mm and 6.5 ± 1.4 mm, respectively. The interbranch distance measured 8.8 ± 2.5 mm in the axial plane and 9.7 ± 3.2 mm in the coronal plane. The mean branching angles were 61.7° ± 18.6° (axial) and 59.2° ± 19.9° (coronal), demonstrating substantial variability in PV geometry. The mean recipient operation time was 414.3 ± 99.1 minutes, with bench and PV anastomosis times averaging 108.6 ± 39.7 minutes and 10.1 ± 4.7 minutes, respectively. Adjunctive procedures included shunt ligation in 2 patients (1.8%), splenectomy in 2 (1.8%), and intraoperative thrombectomy in 23 (20.7%). The mean hospital stay was 7.8 ± 2.9 days. Overall, 90 patients (81.1%) developed at least 1 CD complication, including 80 (72.1%) with major events (grade ≥III). PV complications occurred in 41 patients (36.9%), of whom 16 (14.4%) experienced major PV complications requiring intervention (Supplementary Table 1). Geometric predictors of portal vein complications RAPV–RPPV geometry did not differ significantly between patients with and without major complications requiring intervention (all P > 0.60). In contrast, patients who developed PV complications had a longer coronal interbranch distance (10.6 ± 3.0 mm vs. 9.2 ± 3.1 mm, P = 0.031) and wider branching angles in both axial (67.8° ± 19.2° vs. 58.1° ± 17.4°, P = 0.009) and coronal planes (65.9° ± 19.2° vs. 55.2° ± 19.3°, P = 0.005), whereas axial distance was not different (P = 0.391). These findings suggest that wider angles and greater coronal separation of RAPV and RPPV are significant predictors of overall PV complications, although not of major events (Table 1). Regression analysis of risk factors for overall and major portal vein complications In the univariate assessment, several anatomical characteristics showed a positive association with the likelihood of PV complications. A wider axial branching angle (odds ratio [OR], 1.031; 95% confidence interval [CI], 1.006–1.055; P = 0.015), a broader coronal angle (OR, 1.028; 95% CI, 1.006–1.051; P = 0.012), and a longer coronal distance (OR, 1.163; 95% CI, 1.018–1.327; P = 0.026) were all linked to higher complication rates. Among operative variables, direct venoplasty tended to have a protective effect (OR, 0.399; 95% CI, 0.170–0.935; P = 0.035), while intraoperative thrombectomy was correlated with a greater risk of PV events (OR, 3.211; 95% CI, 1.225–8.414; P = 0.018). When these factors were re-evaluated in the multivariate logistic model, the axial angle remained an independent determinant of PV complications (OR, 1.041; 95% CI, 1.003–1.079; P = 0.033). Thrombectomy also continued to show a significant adverse impact (OR, 3.913; 95% CI, 1.200–12.756; P = 0.024). After statistical adjustment, the apparent benefit of direct venoplasty was no longer observed, and none of the other demographic, anatomical, or operative variables retained independent significance (Table 2). When the analysis was restricted to major PV complications (CD grade ≥III), univariate results identified larger RPPV diameter (OR, 1.521; 95% CI, 1.029–2.246; P = 0.036) and wider axial angle (OR, 1.032; 95% CI, 1.000–1.064; P = 0.049) as significant risk factors. Multivariate analysis confirmed both parameters as independent predictors: increased RPPV diameter (OR, 1.792; 95% CI, 1.024–3.135; P = 0.041) and wider axial angle (OR, 1.078; 95% CI, 1.015–1.146; P = 0.015). No other demographic, anatomical, or surgical factors, including type of anastomosis, were significantly associated with major complications (Table 3). Taken together, these findings indicate that while multiple anatomical and operative factors contribute to the overall risk of PV complications, the occurrence of major complications is specifically determined by the anatomical geometry of the dual portal system—particularly RPPV diameter and axial angle—rather than by surgical technique. This highlights the importance of preoperative anatomical assessment in risk stratification and donor selection. Threshold analysis indicated that an RPPV diameter ≥7.0 mm and an axial RAPV–RPPV angle ≥65–70° were associated with a substantially higher incidence of major PV complications. Management of major portal vein complications In patients who developed CD grade ≥III PV complications, the patterns and management strategies varied according to the reconstruction method. Direct venoplasty cases were mainly complicated by branch stenosis of the RAPV or RPPV, which occurred both early and during follow-up, and were effectively treated with stent insertion. Y-graft reconstructions showed a broader spectrum of events, including main PV thrombosis or stenosis in addition to branch lesions, most of which developed in the early postoperative period. These were managed with endovascular therapy such as MPV or branch stenting and, in selected cases, balloon dilatation was performed. Patch graft cases were uncommon, but RPPV stenosis requiring stent placement was observed. Overall, despite variation in complication type by anastomotic technique, the majority of events were successfully controlled with minimally invasive endovascular interventions, with stenting representing the cornerstone of management (Table 4). Outcomes according to anastomosis type and major complications As shown in Fig. 1, the cumulative incidence of CD grade ≥III PV complications did not differ significantly among the 3 reconstruction techniques. Over long-term follow-up (Fig. 1A), as well as within the first 2 postoperative months (Fig. 1B, highlighted in the red dashed circle), the rates of major PV complications were comparable between direct venoplasty, Y-graft interposition, and patch grafting (log-rank P = 0.209). Similarly, overall survival did not vary according to the anastomosis method, with closely aligned survival curves across groups throughout the observation period (Fig. 1C, P = 0.388). In contrast, the presence of CD grade ≥III complications, regardless of surgical technique, was associated with significantly reduced survival. Patients who developed major complications had an estimated 100-month survival of approximately 80%, compared with nearly 100% among those without complications (Fig. 1D, P = 0.014). These findings indicate that while the choice of anastomosis type does not influence either complication rates or long-term survival, the occurrence of major PV complications exerts a substantial negative impact on outcomes. DISCUSSION This study provides one of the largest single-center analyses of right lobe LDLT with dual PV grafts, focusing on predictors of PV complications. More than one-third of patients developed PV complications, highlighting the inherent technical challenges of dual PV reconstruction. Importantly, the occurrence of major complications was determined primarily by anatomical geometry rather than surgical technique, with RPPV diameter and wider axial angle emerging as independent predictors. These findings suggest that vascular geometry, rather than the method of reconstruction, dictates hemodynamic stability at the anastomotic site. A larger RPPV diameter may predispose to flow turbulence or size mismatch, while a wider axial angle may create unfavorable flow dynamics that promote stenosis or thrombosis. In contrast, coronal parameters did not show independent associations with major events, reinforcing that axial orientation has a more direct impact on portal flow. These observations underscore the importance of meticulous preoperative assessment of PV geometry, ideally through 3-dimensional imaging and quantitative modeling, to guide donor selection and surgical planning [1415]. In particular, an RPPV diameter ≥7.0 mm and an axial R APV–RPPV branching angle ≥65°–70° emerged as practical anatomical thresholds associated with a markedly increased risk of major PV complications. These parameters may serve as useful preoperative criteria for donor selection and surgical planning, although prospective validation is required before formal clinical adoption. In contrast, the method of PV reconstruction—direct venoplasty, Y-graft interposition, or patch grafting—was not independently associated with either overall or major complications. Although direct venoplasty appeared protective in univariate analysis, this association disappeared after multivariate adjustment, indicating that no single technique is inherently superior. These findings are consistent with prior studies, which have shown comparable outcomes across reconstruction methods when performed by experienced surgeons. Interestingly, a shorter axial distance emerged as an independent risk factor for adverse outcomes. This may be explained by our previous observation that direct venoplasty cases typically had the shortest interbranch distance; when the separation is only moderately short, however, forcing a direct venoplasty rather than adopting an alternative method may impose undue tension or misalignment at the anastomotic site, thereby predisposing to complications. Thus, technique selection should be individualized according to donor anatomy and intraoperative conditions, while the priority should remain on managing anatomical risk factors rather than favoring a particular method. The prognostic implications of major PV complications were particularly striking. Patients who developed CD grade ≥III PV complications had significantly lower long-term survival compared with those without such events, whereas survival outcomes did not differ by anastomosis type. Most complications were successfully managed with endovascular interventions, especially stent placement, which represented the cornerstone of treatment. Nonetheless, survival remained compromised once a major complication occurred, even after successful intervention, underscoring the critical need for prevention rather than reliance on rescue therapies. These findings extend beyond prior comparative studies of anastomotic techniques by incorporating detailed quantitative measurements of PV geometry. By identifying RPPV diameter and axial angle as independent predictors of major complications, this study provides a framework for risk stratification and the development of predictive models. Such models could refine donor selection, optimize surgical strategies, and guide future work employing computational flow dynamics to further elucidate hemodynamic consequences of anatomical variations. Future studies incorporating computational fluid dynamics modeling or standardized Doppler-based flow assessments will be essential to validate how anatomical geometry influences hemodynamic behavior at the anastomotic site. This study has inherent methodological constraints. Because it was conducted retrospectively at a single institution, the findings may not be fully applicable to other centers or broader patient populations. The small number of patch graft cases constrained the statistical power of subgroup analyses and conclusions derived from this subgroup should therefore be interpreted with caution. Moreover, discrepancies between preoperative imaging-based measurements and intraoperative assessments may introduce measurement variability and potential observer bias. Finally, the retrospective design inherently limits causal inference. Future multicenter prospective studies with standardized imaging protocols, operative techniques, and long-term follow-up will be essential to validate and generalize our findings across different surgical teams and patient populations. In conclusion, the risk of major PV complications in dual PV LDLT is determined by anatomical geometry, particularly RPPV diameter and axial angle, rather than surgical technique. These parameters independently predict adverse outcomes and should therefore be incorporated into preoperative evaluation and donor selection. Preventive strategies based on anatomical risk stratification, rather than reliance on specific reconstruction methods, hold the key to improving long-term graft survival in this technically complex setting. Importantly, the integration of anatomical risk assessment into donor selection algorithms and structured preoperative planning workflows may provide a practical means to minimize complication risk and enhance clinical outcomes.