Recently, 2-[18F]-fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET)/computed tomography (CT) has been used for diagnosing malignant tumors, and many reports have described its role for predicting the malignant potential of these tumors. However, the clinical efficacy of PET/CT for diagnosing hepatocellular carcinoma (HCC) remains controversial, and only a few reports have described the predictive value of its findings for pathological malignant potential and prognosis (1–3).

Hepatic resection is performed as a standard therapy for HCC in Japan (4,5). However, recurrence of HCC after resection is known to occur at a high rate, and early recurrence is considered to be a significant prognostic factor for death. Although macroinvasion of HCC to the portal vein is also a factor for poor prognosis (6), most patients with HCC without macro-tumor thrombosis suffer from recurrence after resection. Previous reports have investigated prognostic markers for early recurrence and survival, including the doubling time of pre-operative serum α-fetoprotein (AFP) and protein induced by vitamin K absence or antagonist II (PIVKA-II) (7), complication with diabetes mellitus (8), hepatic steatosis (9) and tumor node metastasis (TNM) stage (10), although these are not adequately sensitive. In the present study, we investigated the predictive value of PET/CT for the pathological malignant potential of HCC as a new indicator for early recurrence after hepatic resection.

From April 2006 to October 2009, 53 patients with naïve HCC, examined by PET/CT and treated by hepatic resection, were enrolled. None had poorly controlled diabetes mellitus. All were examined using PET/CT (Discovery ST Elite 16; GE Healthcare Japan Co. Ltd., Tokyo, Japan) within the month prior to resection. PET/CT was performed 60 min after a bolus injection of F-18 FDG (3 MBGq/kg). Accumulations of FDG [standardized uptake value (SUVmax)] in HCC and non-HCC areas of the liver as well as the ratio of SUVmax (R-SUV), which indicated the tumor to non-tumor ratio, were determined. In cases with multiple HCC, the SUVmax was calculated for the main nodule. From these findings, we evaluated prognostic factors for early recurrence, which was defined as recurrence within 2 years of resection. Moreover, R-SUV values were compared to the pathological findings, including microvascular invasion (vp), micro-intrahepatic metastasis (im) and gross type of HCC (11,12). The patients were divided into two groups, low R-SUV (n=19) and high R-SUV (n=34), and their clinical parameters were compared.

One patient was classified as TNM stage I, 35 as stage II, 14 as stage III and 3 as stage IV, based on the results of imaging examinations (abdominal ultrasonography and dynamic CT). There were no cases with extrahepatic metastasis. R-SUV values ranged from 1.0 to 6.9. In pathological analyses, all were diagnosed as typical HCC. PIVKA-II (≥200 mAU/ml), fucosylated AFP (AFP-L3) (≥15%), tumor size (≥5 cm) and high R-SUV (≥1.5) were found to be risk factors for early recurrence in a univariate analysis (P<0.05, respectively) (Table I). In a multivariate analysis, high R-SUV (≥1.5) was the only risk factor (P<0.05) (Table II). The recurrence-free rate in the low R-SUV group was higher than that in the high R-SUV group (1- and 2-year recurrence-free rates: 100 and 67%, 67 and 17%, respectively; P<0.01) (Fig. 1). While the frequencies of high levels of PIVKA-II (≥200 mAU/ml) and AFP-L3 (≥15%) were greater in the high R-SUV group (52.9 and 38.2% vs. 21.1 and 10.5%, respectively; P<0.01), there were no significant differences in regard to the frequencies of high levels of AFP (≥100 ng/ml), tumor diameter ≥5 cm, Child-Pugh class, number of tumors and TNM stage between the groups (Table III).

Patients with HCC nodules rated as Edmondson III (13) had a higher R-SUV value (3.0±1.8) than those rated as I and II (1.4±0.3 and 1.9±0.9, respectively; P<0.01). Patients with nodules showing vp(+) and im(+), and with non-boundary type of nodules (single nodular type with extranodular growth, confluent multinodular or invasive type) had higher R-SUV values than those with vp(−), im(−) or boundary type (vaguely nodular or single nodular type) (3.6±2.4 vs. 2.0±0.9, 3.5±2.3 vs. 1.9±0.8 and 2.9±1.8 vs. 1.6±0.5, respectively; P<0.01). Throughout the observation period, extrahepatic metastasis was observed in 2 cases of stage II; these cases had high R-SUV (2.3 and 2.4, respectively). Fig. 2 shows representative results of a patient with low R-SUV (1.4) whose pathological findings were single nodular type, Edmondson I, and who was negative for both vp and im. By contrast, Fig. 3 presents the results of a representative patient with high R-SUV (1.9) whose pathological findings were confluent multinodular type, Edmondson III and positive for vp, though the tumor size was small (2.5 cm in diameter).

In Japan, the shortage of donors for liver transplantation is a major obstacle to the treatment of HCC, thus hepatic resection is often performed as curative therapy (4,5). FDG-PET/CT is a functional imaging modality that is used to measure the glucose metabolism of malignant tumors, although its clinical efficacy has not been established. The ability of PET to detect HCC in the liver was found to be less effective than that of contrast enhanced CT (14). On the other hand, Yoon et al (15) and Sugiyama et al (16) reported that PET is useful for the screening of extrahepatic metastasis from HCC. Recently, the usefulness of FDG-PET for predicting HCC recurrence following liver transplantation was proposed (17,18). However, few reports have described FDG-PET as useful for predicting prognosis after resection (1,2). Kawamura et al found that even in patients diagnosed in the early phase of HCC, a high R-SUV value among other prognostic scores may indicate poor prognosis or the need for radical treatment (3). The present results are similar to past reports, which showed that a high R-SUV value is capable of predicting early recurrence after resection, and that the relationship between a high R-SUV and pathological malignant potential is associated with positive findings for im or vp (19), higher Edmondson grade and worse gross type (11,12) in resected specimens. Since the average R-SUV of Edmondson I was <1.5, we set 1.5 as the cut off; this cut off value predicted the early recurrence of HCC.

Full body scanning with PET/CT is useful for the screening of extrahepatic metastasis and staging in patients with large HCC (15). An FDG-PET/CT examination is non-invasive and useful for predicting the malignant pathological potential of HCC before resection without the need for a biopsy. However, patients with high R-SUV values must be followed carefully with imaging modalities after resection.

In conclusion, we found that HCC patients with a high R-SUV value (≥1.5) had an elevated risk of early recurrence after resection, while R-SUV was also shown to be related with pathological findings. Thus, R-SUV is proposed as a useful predictive marker for the early recurrence of HCC before surgical resection.