Textbook outcome in liver surgery for intrahepatic cholangiocarcinoma: defining predictors of an optimal postoperative course using machine learning

Background

We sought to define textbook outcome in liver surgery (TOLS) for intrahepatic cholangiocarcinoma (ICC) by considering the implications of perioperative outcomes on overall survival (OS).

Methods

Using a multi-institutional database, TOLS for ICC was defined by employing novel machine learning (ML) models to identify perioperative factors most strongly predictive of OS ≥ 12 months. Subsequently, clinicopathologic factors associated with achieving TOLS were investigated.

Results

A total of 1556 patients with ICC were included. The ML classification models demonstrated that the absence of post-hepatectomy liver failure, intraoperative blood loss <750 mL, absence of major infectious complications, and R0 resection were the perioperative outcomes associated with prolonged OS, thereby defining TOLS for ICC. On multivariable analysis, older age, ASA class >2, lymph node metastasis, receipt of neoadjuvant therapy, advanced T status, poor histological grade and microvascular invasion were independently associated with lower odds of achieving TOLS (all p-values<0.05). Overall, 60.2 % (n = 936) of the patients achieved TOLS, demonstrating markedly improved OS and recurrence-free survival (RFS) than individuals who did not (both p < 0.05).

Conclusion

A standardized definition of TOLS for ICC was established that may be used to evaluate hospital performance at the patient level and help optimize surgical outcomes for patients with ICC.