The prospectively maintained database of the Cancer Institute Hospital, Tokyo, Japan was queried to identify patients who underwent initial hepatectomy for CRLMs at this hospital from January 2004 through December 2012. We excluded patients who underwent R2 resection. During the study period, 462 patients underwent initial hepatectomy for CRLMs with curative intent. Of these, 265 patients (57.4 %) were ≤ 64 years of age with a median of 56 years; 151 patients (32.7%) were 65–74 years with a mean age of 68 years, and the other 46 patients (9.9%) were ≥ 75 years with a median age of 80 years. The study population was divided into three age groups: ≤ 64 years, 65–74 years, and ≥ 75 years of age. Baseline characteristics, perioperative course, and long-term outcomes were compared retrospectively between the three groups.
The Institutional Review Board of the Cancer Institute Hospital approved this study (Protocol 2018-1033).
Indications for Hepatectomy for CRLMs
In our institute, indications for resection of CRLMs during the January 2004 through December 2012 period consisted of: (1) no comorbid conditions that precludes hepatic resection, (2) all liver tumors that were amenable to resection would have a clear margin, leaving at least 30% of noncancerous remnant liver without a potentially ischemic or congested area, and (3) no unresectable extrahepatic tumors. The indications for repeat hepatectomy for liver recurrence were the same as those for initial hepatectomy. No age restriction was set for initial or repeat hepatectomy as long as patients met the above criteria.
Routine use of preoperative chemotherapy was not adopted until 2010. After 2010, preoperative chemotherapy was routinely performed for patients with ≥ four CRLMs or those with CRLMs > 50 mm or those with resectable extrahepatic metastases by imaging studies .
Surgical Procedure and Postoperative Outcomes
Parenchymal-sparing hepatectomy was the standard procedure regardless of the number or size of CRLMs. Major hepatectomy, which was defined as resection of ≥ 3 segments, was performed only when CRLMs were close to major Glisson’s pedicles. Following laparotomy and liver mobilization, fundamental intraoperative ultrasonography was performed to confirm the tumors detected by preoperative imaging and to search for new occult lesions. Resecting of all the tumors were intended, including newly detected nodules and disappearing CRLMs by preoperative imaging. Liver transections were performed by the crushing technique using the LigaSure vessel sealing system (Valleylab, Boulder, CO, USA), as reported previously . Surgical margins were measured from the resected specimens. A positive surgical margin was defined as microscopic evidence of tumor at the resection margin.
The severity of postoperative complications was assessed according to the Clavien–Dindo classification; grade IIIa or worse was defined as a major complication. Any complications that developed within 90 days after the operation were included .
Patient follow-up consisted of measuring serum tumor markers (carcinoembryonic antigen and carbohydrate antigen) at every visit as well as enhanced computed tomography every 3–6 months. Although adjuvant chemotherapy was not routinely administered, it was given to: (1) patients who were included in clinical studies, (2) patients who underwent simultaneous resection of advanced primary disease, and (3) patients who had advanced CRLMs judged by a multidisciplinary team.
Associations between variables with categorical data were sought using either Fisher’s exact test or Pearson’s Chi-squared test. The Mann–Whitney’s U-test was applied to continuous variables between the three groups. Survival curves were generated by the Kaplan–Meier method, and comparisons between the groups were performed using a log-rank test. Overall survival (OS), disease-specific survival (DSS), and recurrence-free survival (RFS) were defined as the interval from the date of primary hepatectomy to the date of all death, death attributed to colorectal cancer, and recurrence, respectively. Statistical significance was assessed using a two-tailed test across p < 0.05. All the statistical analyses were performed using JMP software, version 10 (SAS Institute Inc., Cary, NC, USA).