Impact of Portal Vein Embolization on Patient's Long-term Survival, Complications and Cancer Recurrence.

For patients with colorectal liver metastasis (CLM), the prospect of long-term survival relies on liver resection. Wistfully, more than 75 % of patients with CLM are initially unresectable, due to an insufficient future liver remnant (FLR) volume

In order to increase FLR volume, most patients will first receive chemotherapy to reduce the tumor load (downsizing).

When chemotherapy is insufficient to provide an adequate postoperative FLR, portal vein embolization (PVE) can be performed. About 50-70 % of patients undergoing PVE obtain a sufficient liver hypertrophy to allow liver resection.

While PVE is recognised for its efficacy to induce liver hypertrophy, some studies expressed substantial concerns regarding the potential adverse effect of this intervention on pre-resection tumor progression, increased risk of cancer recurrence following resection and reduced overall survival following resection

Those studies suggested that the need to perform PVE should be assessed thoroughly for each patient and that chemotherapy should be maintained during the whole hypertrophy process in order to contain the potential adverse effect of PVE on tumor progression.

Other studies found no significant association between PVE and negative oncological outcomes.

As mentioned in almost every study cited above, more data is needed to provide a clearer vision regarding the impact of PVE on tumor progression and cancer recurrence following liver resection.

The aim of this study is to compare the overall and disease-free survival of PVE-requiring patients to the ones who underwent upfront surgery (NoPVE).

As a secondary objective, the impact of several covariates (related to surgery, patient's condition and disease stage) on survival and cancer recurrence will be tested.

Our hypothesis are that 1) PVE might be associated with a lower overall survival and a higher risk of cancer recurrence in univariate analysis but 2) this association will not remain significant when other covariates are included in the proportional COX hazard models.