Experiences with revision for BA patients with recurrent jaundice after initial OKPE or ILKPE have rarely been reported. In two studies, CJ rates of 60%  and 83.3%  were obtained after revision of OKPE for patients with initially successful OKPE. Naruhiko et al. showed good results of RLKPE for patients with recurrent jaundice after initial OKPE. The authors found that 10/12 patients had normal bilirubin concentrations after RLKPE . In our study, 20/25 (80%) patients with sudden cessation of bile drainage achieved normal bilirubin concentrations after RLKPE. In addition, the 3-year and 5-year SNL rates of patients who underwent RLKPE were comparable to those of patients after unrevised ILKPE and were significantly better than those of patients after failed ILKPE. These satisfactory results were attributed to a rigorous selection of candidates with poor jaundice reduction after the initial Kasai operation. If the indications for revision were relaxed, the postoperative results would be quite different. In a survey from the Japanese Biliary Atresia Registry, the revision rate was 21% and the CJ rate was 35% among 2630 BA patients after the failed initial OKPE during 1989–2011 .
The indications for RLKPE in our center were bile drainage that stopped abruptly after ILKPE and no improvement after 2 weeks of antibiotic treatment. However, a consensus has never been reached for the optimal timing of revision worldwide. Some reports have shown that revision could be effective for patients in whom jaundice suddenly recurred after a favorable initial reduction, irrespective of the time since the initial surgery [14, 15]. However, Shirota suggested that the revision should be performed as soon as irreversible jaundice is recognized . In our study, the time interval from the cessation of bile drainage to revision was 21.9 ± 8.0 days, which should be shortened to avoid further liver damage. RLKPE is not recommended for patients with bile drainage failure for more than 60 days or patients with severe ascites or aggravated liver function. For patients over 1 year old with sudden bile drainage failure, antibiotics should be administered first, and liver transplantation is recommended if conservative treatment is ineffective.
Technically, RLKPE is relatively easy to perform, partly due to the omission of the Roux limb anastomosis, which is a time-consuming step in ILKPE. Under the magnified and clear view of the laparoscope, the Roux limb is easily identified by only removing the greater omentum adhered to the porta hepatis. After partly dismantling the tail of the Roux limb, the fibrous occlusive hilar plate could be spotted. At this point, the main portal vein and its branches should be clearly and accurately marked, and the resection of the fibrous plate should be confined to the area between the portal branches. The resection level of the fibrous plate depends on the bile drainage status, which is consistent with that in ILKPE. The last step of portoenterostomy in RLKPE is the same as that in ILKPE [15–18].
However, some authors were reluctant to repeat the Kasai operation because of the possibility of a high incidence of perioperative complications (e.g., uncontrolled intraoperative bleeding) and longer ORT . Perineal adhesion was more common in patients who underwent RLKPE. Specifically, in the vicinity of the porta hepatis, dense fibrous granulation tissue sometimes may result in unanticipated bleeding and damage to the Roux limb . In some cases, oozing bleeding from the remains of the hilar fibrous plate is difficult to stop under laparoscopy. In addition, accidental portal hemorrhage from iatrogenic injuries may place patients at great risk. In our study, however, fewer peritoneal adhesions around the porta hepatis were observed during RLKPE, at least partially due to the laparoscopic technique used for ILKPE. Under laparoscopy, the image could be zoomed in, and the portal vein could be easily recognized, which may help to protect it from accidental injuries when dissecting the hilar fibrous plate . Active bleeding at the fibrous stump can be stopped by compression with a gauze pad for a period. In the event of uncontrolled hemorrhage, conversion to OKPE should be adopted without taking risks to ensure the safety of the patients. The above measures adopted in our hospital may make RLKPE safer. In our study, the blood loss volume, conversion rate, ORT and incidence of APOC of RLKPE were not significantly different from those of ILKPE. Perhaps the technique of RLKPE may repeatedly be used for revision in patients with recurrent jaundice postoperatively.
Two limitations of our study deserve comment. First, the evaluation of RLKPE was based on retrospective data. Second, there were only a few cases of RLKPE compared with the number of unrevised ILKPE and failed ILKPE cases. Nonetheless, we recommend RLKPE as the preferred surgical treatment for patients with recurrent jaundice after the initial Kasai operation. If the time interval between the cessation of bile drainage and revision is shortened, the postoperative outcomes could be improved.