Our previous study demonstrated the clinical impact of HA formation on predicting CR-POPF in the patients who received O-PD [17]. Building on this, here we aimed to verify this impact in a consecutive series of patients who received either O-PD or R-PD. The findings confirm that HA formation predicts CR-POPF, and further reveal that the incidence of E-HA formation was significantly lower in the R-PD group than in the O-PD group, suggesting an advantage of R-PD over O-PD in reducing HA formation and consequent CR-POPF.

In our previous study, the analysis of the CT value of HA let us speculate that the HA area is not fluid collection or anastomosis separation, but rather pancreatic parenchyma with reduced blood flow. Taking the new results into account together with this speculation, we suggest that an unknown factor may have lessened the reduction of blood flow at the PJ site in patients who underwent R-PD in comparison with those who received O-PD. Unfortunately, we could not identify this factor. However, we initially speculated that the reduced blood flow might result from excessive tension when creating the PJ and/or over-mobilization of the remnant pancreas for the PJ anastomosis. Since the PJ procedure is the same in both O-PD and R-PD, including the mobilization of the remnant pancreas, the factor may lie in the differences in the strength of ligation on the jejunal serosa covering the pancreatic stump during the PJ procedure. In robotic surgery, the lack of tactile sensation might lead to looser ligations due to concerns about applying excessive force, which could cause the thread to break. This may lead to less blood flow reduction at the PJ site and, consequently, a lower incidence of HA formation in R-PD [21, 22]. The photographs in Fig. 3 show the potentially looser ligation on the jejunal serosa covering the pancreatic stump during the PJ procedure. To test this hypothesis, we plan to analyze blood flow at the PJ site in the near future.

Photographs showing ligation on the jejunal serosa covering the pancreatic stump during the PJ procedure in R-PD. A The photograph shows the potentially looser ligation using 3–0 nonabsorbable monofilaments with the modified Blumgart method [18], which leads to less blood flow reduction at the PJ site in R-PD compared with O-PD. B The completed PJ anastomosis. O-PD, open pancreaticoduodenectomy; PJ, pancreatojejunostomy; R-PD, robotic pancreaticoduodenectomy

When considering the clinical applications of these results, one is that the progression to CR-POPF can be predicted by the HA findings in the patients exhibiting BL POPF, regardless of whether they underwent O-PD or R-PD. In this case series, the presence of an E-HA predicted progression to CR-POPF with 89.4% sensitivity and 85.1% specificity. When stratified by the surgical approach, these values were, respectively, 100% and 76.0% in O-PD, and 75.0% and 95.5% in R-PD, indicating the potential clinical utility of HA findings in both groups. Another important point concerns the incidence of E-HA formation based on the surgical approach. As summarized in Fig. 2, the incidence of E-HA in the O-PD group was consistent with our previous report [17]. The incidence was significantly lower in the R-PD group than in the O-PD group before PSM, and this trend was confirmed after PSM. Although the advantage of R-PD over O-PD remains inconclusive, our retrospective study provides valuable evidence suggesting that R-PD may have an edge over O-PD [6,7,8,9,10, 12]. Finally, as mentioned earlier, we have speculated on the underlying mechanism of HA formation, especially considering the reduced percentage of E-HA in the patients with R-PD. Understanding this mechanism, which leads to CR-POPF, could offer insights for preventing CR-POPF not only in R-DP but also in O-PD.

This study has several limitations. First, although the impact of HA was also confirmed in patients who underwent R-PD, the study is retrospective and includes a small number of patients, necessitating caution in interpreting the results. In particular, the small sample size applies to both pre- and post-PSM analyses, making it a significant limitation. Actually, we had considered collecting additional cases prior to reporting the results of the present study, but we prioritized disseminating the current findings and chose to publish with the currently available sample size. In the future, we plan to increase the number of cases and conduct further validation in conjunction with the aforementioned blood flow evaluation. Furthermore, due to the retrospective study design, R-PD and O-PD were not randomly assigned to the included patients, indicating that they remained incomparable even after performing PSM for the comparison. This represents a limitation inherent to the study design. Second, the included patients in this study were limited to those who met the diagnostic criterion of BL and underwent CE-CT assessment on PODs 3–14. In addition, regarding the timing of CT, although it was within the range of 3 to 14 days, it was not strictly set to a specific day. Therefore, while there was no significant difference in the duration from the surgery to CT between the E-HA group and the S-HA group, the results of this study may include potential bias related to the timing of CT. This limitation means that the findings in this study may apply only to similar patients. If applied otherwise, the universality of the results of this study might be compromised.

In summary, this study confirmed the clinical impact of HA formation in predicting CR-POPF in patients who received PD, including both O-PD and R-PD. Furthermore, the results suggest that R-PD, compared with O-PD, significantly reduces the incidence of E-HA formation, indicating a potential advantage of R-PD over O-PD.