This study was performed retrospectively. Between 1980 and 2007, 646 HCC patients underwent radical resection in our institute. The patients included 525 men and 121 women, with a meanage ± SD of 61.9±9.40 years. Of the 646 patients, 237 only had hepatitis B, 257 only had hepatitis C, 85 had both hepatitis B and C and the remaining 67 patients had no viral infection. According to the Child-Pugh classification, 565 patients were scored as A and the remaining 81 patients as B. All the patients underwent pathologically complete (R0) resection The 3-year and 5-year survival rates for all the patients were 75.6% and 62.4%, respectively. All the patients were regularly followed up for 52.4±44.4 months (range, 1.2-248.2 months) in our department. Of the 646 patients, 483 developed postoperative HCC recurrence [intrahepatic recurrence in 372 (57.6%) and extrahepatic recurrence, with or without intrahepatic lesions, in 111 patients (17.2%)], while 163 patients (25.2%) had no recurrence.

The study protocol was approved by the Human Ethics Review Committee of Osaka University and asigned informed consent form was obtained from each subject.

Lung metastasis was diagnosed using a combination of imaging techniques, such as computed tomography and magnetic resonance imaging; histopathological findings, such as fine-needle biopsy; and the levels of tumor markers, such as serum α-fetoprotein and serum protein induced by vitamin K absence or antagonist-II. Pulmonary resection was indicated only when the metastatic lesions were limited to the lung and intrahepatic lesions were absent or controllable. In all other cases, systemic chemotherapy was performed based on the following criteria, as previously reported (14): Patients were required to be aged 20–75 years, with an Eastern Cooperative Oncology Group performance status of ≤2, a life expectancy of ≥ 12 weeks, measurable or assessable disease, adequate bone marrow function (hemoglobin concentration ≥8.0 g/dl, leukocyte count 2,500-12,000 /µl and platelet count ≥80,000 /µl) and adequate hepatic and renal reserve (total bilirubin ≤1.5 mg/dl, aspartate aminotransferase and alanine aminotransferase ≤100 IU/l, blood urea nitrogen ≤30 mg/dl and serum creatinine ≤1.5 mg/dl).

Among the 111 patients with extrahepatic recurrence, with or without intrahepatic recurrence, lung metastasis was observed as the first distant metastasis in 62 patients; these patients were enrolled in the present study. The patient characteristics are reported in Table I. The patients included 46 men and 16 women, with a mean age ± SD of 60.3±10.5 years. A total of 13 patients only had hepatitis B, 26 only had hepatitis C, 12 had both hepatitis B and C and 11 had no viral infection. The average resected tumor size at primary surgery was 66.8±46.3 mm and histopathological grading revealed poorly, moderately and well-differentiated HCC in 48, 12 and 2 patients, respectively. Vascular invasion was observed in 25 patients (portal vein invasion in 17, hepatic vein invasion in 2 and both portal and hepatic vein invasion in 6 patients). Primary recurrence was manifested as an intrahepatic lesion in 42, lung lesion in 16 and both intrahepatic and lung lesions in 4 patients. The average disease-free interval (DFI) for all enrolled patients was 11.1±14.8 months, while the distant recurrence-free interval (DRFI), defined as the period from primary surgery to the appearance of distant metastasis, was 16.2±22.4 months. S-1 treatment was administered to 11 patients (S-1 treatment group), while the remaining 51 patients (non-treatment group) received other systemic chemotherapy with a traditional regimen, such as the combination of adriamycin, 5-FU and cisplatin (CDDP). The criteria for S-1 treatment were almost identical to those for other chemotherapeutic regimens and the 51 patients who did not receive S-1 treatment were analyzed as a historical control. Two patients underwent pulmonary metastastectomy and did not receive S-1 treatment.

Combination therapy with S-1 and IFN was performed as previously reported (14). S-1 (Taiho Pharmaceutical, Tokyo, Japan) was administered orally twice daily after a meal at a total dose of 80 mg/m² body surface. Three initial doses of S-1 were established according to body surface area as follows: <1.25 m², 80 mg/day; 1.25-1.5 m², 100 mg/day; and ≥1.5 m², 120 mg/day. IFN- α (OIF; Otsuka Pharmaceutical, Tokyo, Japan) was injected subcutaneously at a dose of 5×10⁶ units (5 MU)/m² body surface on days 1, 3 and 5 of each week. One course consisted of consecutive administration for 28 days, followed by ≥ 14 days of rest. The combination treatment schedule is summarized in Fig. 1. The non-steroidal anti-inflammatory drug diclofenac sodium was administered prior to IFN- α injection to alleviate fever, which is a common adverse effect of IFN- α.

Data are expressed as mean ± SD. Differences in continuous variables were evaluated by the Student's t-test. The Fisher's exact probability test was used to compare discrete variables. The DFI, DRFI and survival after distant recurrence (SADR) rates were estimated with the Kaplan-Meier method and compared using the log-rank test. The Cox's proportional hazard regression model with stepwise comparisons was used to analyze independent prognostic factors. P<0.05 was considered to indicate a statistically significant difference.

We evaluated the toxicity of S-1 treatment according to the National Cancer Institute-Common Toxicity Criteria for Adverse Events, version 3.0 (http://ctep.cancer.goc/protocolDevelopment/electronic_applications/docs/ctcaev3.pdf) by determining blood counts and biochemical profiles at least once every 2 weeks and monitoring patients for the occurrence of non-hematological toxicities, such as general fatigue, fever, nausea/vomiting, diarrhea, skin pigmentation, hand-foot syndrome and, particularly, depression. Hematological toxicity was observed in 5 patients (3 patients had thrombocytopenia, 2 had leukocytopenia and 1 had anemia). Toxicity >grade 3 was observed in only 1 patient (9.1%), who developed leukocytopenia but did not require withdrawal or dose reduction of either drug. Non-hematological toxicities were observed in 6 patients (4 patients had fever, 3 had increased serum bilirubin levels, 3 had general fatigue and 1 had dermatitis). All non-hematological toxicities were grade 1, except 1 patient (9.1%) with a grade 2 increase of serum bilirubin levels. There was no need for discontinuation of the combination therapy and no treatment-related mortality was reported during this study.

The characteristics of the S-1 treatment and non-treatment groups are summarized in Table II. There were no significant between-group differences in gender, age, viral infection status, or Child-Pugh classification. The average size of primary HCC in the S-1 treatment group was 84.4±35.7 mm and histopathological grading revealed that all HCCs were poorly differentiated and 6 were vascular invasion-positive. In the non-treatment group, the average primary HCC size was 63.0±47.7 mm and histopathological grading revealed that the HCCs were poorly, moderately and well-differentiated in 37, 12 and 2 patients, respectively, with 19 cases being vascular invasion-positive. In the S-1 treatment group, the DFI was 8.8±6.8 months and the DRFI was 14.9±11.2 months, while the DFI and DRFI of the non-treatment group were 11.6±16.1 and 16.5±24.3 months, respectively. These intervals did not significantly differ between groups. The Kaplan-Meier analysis also revealed no significant between-group differences in DRI or DRFI (Fig. 2). Pulmonary resection was performed in 2 patients who did not receive S-1 treatment.

SADR was higher among patients who received S-1 treatment compared to that in patients without S-1 treatment. All the patients in the S-1 treatment group survived for >1 year, with a 3-year survival rate of 81.8%, while the 1-year and 3-year survival rates in the non S-1 treatment group were 58.8 and 43.1%, respectively; this between-group difference was significant (P=0.0141) (Fig. 3). Of the 11 patients receiving S-1 treatment, 1 case exhibited a good response to S-1 combination therapy and the multiple lung metastases almost disappeared. To date, the patient has survived for >5 years while receiving continuous S-1 therapy, although some relapse has appeared (Fig. 4).

We investigated the prognostic significance of various clinicopathological factors in HCC patients with lung metastasis and only treatment with S-1 was found to be of prognostic significance for SADR in the univariate and multivariate analysis (P=0.018 and 0.0091, respectively). Child-Pugh score and venous invasion by the primary liver lesion exhibited marginal significance in both analyses (Table III).