Evaluation of Metabolic Characteristics and Viability of Lipiodolized Hepatocellular Carcinomas Using ¹⁸F-FDG PET/CT

DISCUSSION

HCCs have been shown to accumulate ¹⁸F-FDG to varying degrees, with the size and degree of differentiation of the tumor and serum AFP concentration being important predictors of tumor visualization by ¹⁸F-FDG PET. We found that 65% of low-grade and 62% of high-grade viable lipiodolized HCCs, compared with 37% of low-grade and 75% of high-grade treatment-naïve HCCs, showed increased ¹⁸F-FDG uptake. In addition, lipiodolized HCCs showed no difference in ¹⁸F-FDG uptake with respect to serum AFP concentration or other clinical factors, except tumor size. Previous studies have reported that only 50%–55% of primary HCCs, compared with normal liver parenchyma, show increased ¹⁸F-FDG uptake (9,10). The relatively low sensitivity of ¹⁸F-FDG PET in the detection of HCC may be attributed to high levels of glucose-6-phosphatase, especially in low-grade and well-differentiated HCCs that show low expression of glucose transporter 1 (9,11,16,17). High levels of glucose-6-phosphatase cause dephosphorylation of ¹⁸F-FDG-6-phosphate and leakage of ¹⁸F-FDG back to the circulation. As the HCC stage becomes more advanced, lesions tend to become undifferentiated (18). Our findings were similar, in that the grade and size of treatment-naïve HCCs correlated with ¹⁸F-FDG uptake. Although Risse et al. reported that HCCs were not correlated with the degree of differentiation (19), this study included selected patients for ¹³¹I-lipiodol treatment and only 2 high-grade tumors. Thus, this interstudy difference from our study could affect the controversial result. Interestingly, however, we found that, in contrast to treatment-naïve HCCs, lipiodolized HCCs—particularly low-grade tumors—frequently showed increased ¹⁸F-FDG uptake. This may be caused by increased glycolysis in residual tumors after TACE. HCCs pretreated with TACE displayed significantly higher hexokinase II messenger RNA expression, and there was a significant correlation between the expression of hypoxia-inducible factor 1α (HIF-1α) protein and hexokinase II messenger RNA (20). Furthermore, both HIF-1α and hexokinase II protein were colocalized in cancer cells near necrotic regions. HIF-1α expression was observed only after TACE upregulation of hexokinase II, allowing cancer cells to obtain sufficient energy for growth under the condition of acute hypoxia caused by TACE. During the abrupt devascularization caused by TACE, HIF-1α expressed in HCCs tends to induce hexokinase II expression, mediating a phenotypic change enabling cells to adapt to a hypoxic environment.

The increased ¹⁸F-FDG uptake observed in lipiodolized, compared with naïve, tumors may be also attributable to a more aggressive growth of residual viable tumor when TACE is not completely effective. TACE has been shown to stimulate tumor angiogenesis, thus increasing the proliferative activity of tumor cells to some degree (21–25). After TACE, tumor cells are exposed to an extremely hypoxic or even anoxic environment. Because of ischemic damage, vascular endothelial growth factor and basic fibroblast growth factor, which act synergistically, are overexpressed after TACE (21,23). Levels of these angiogenic factors were found to be significant predictors of clinical outcomes in patients with HCC (26,27). The long-term survival of HCC patients after TACE remains unsatisfactory (28), and the 18-mo recurrence rate after TACE has been reported to be higher than that for untreated patients (5). Thus, the increase in ¹⁸F-FDG uptake after TACE may be caused by increased glycolysis and the aggressive behavior of any residual viable tumor.

Patients with HCCs detectable by ¹⁸F-FDG PET have been found to have significantly higher serum AFP concentrations than do patients with undetectable HCCs (12). Serum AFP concentration has been found to correlate significantly with standardized uptake value (SUV) or the tumor-to-normal ratio, indicating that AFP is involved in glucose metabolism and cell proliferation in HCC (29). We found, however, that ¹⁸F-FDG uptake was not correlated with serum AFP concentration, in patients with either lipiodolized HCCs or treatment-naïve HCCs. No correlation of AFP concentration and ¹⁸F-FDG uptake in our study might be attributable to the interstudy differences in patient population. It has been known that serum AFP values are frequently elevated in patients with chronic liver disease, even in the absence of HCC. In addition, many factors associated with raised AFP include the etiology and severity of liver disease, sex, and race (30). In contrast to the previous studies (12,29), our study population consisted of surgically resectable patients without distant metastasis, a high proportion of patients with high-grade tumors, and a relatively high proportion of patients with hepatitis B virus infections.

In our analysis, we regarded lipiodolized HCCs with ¹⁸F-FDG uptake either greater than or similar to that in the surrounding normal liver tissue as viable. These tumors were considered viable because PET/CT was able to identify the margins of lipiodolized HCCs, and necrotic tissue, compared with normal liver, usually shows significantly decreased ¹⁸F-FDG uptake (11). Of the 30 viable lipiodolized HCCs, 29 (97%) showed ¹⁸F-FDG uptake higher than or similar to that in the surrounding normal liver tissue and could therefore be correctly diagnosed as viable on PET/CT. However, 26 of these 30 viable lipiodolized HCCs showed enhancement on multiphasic contrast-enhanced CT. In comparing the results of ¹⁸F-FDG PET/CT with multiphasic contrast-enhanced CT, we found that 6 tumors yielded discordant results. PET/CT correctly diagnosed 3 viable HCCs in patients with increased AFP levels, whereas these 3 patients were false-negatives on multiphasic contrast-enhanced CT. The other 3 nonviable HCCs in patients with normal AFP levels were true-negatives on multiphasic contrast-enhanced CT. Torizuka et al. reported that increased or similar ¹⁸F-FDG uptake by a tumor with respect to surrounding normal tissue suggested the presence of residual viable tumor tissue, whereas decreased or absent ¹⁸F-FDG uptake indicated greater than 90% necrosis, showing that therapy had been effective (14). In addition, these PET findings appeared to reflect tumor viability more accurately than did intratumor lipiodol retention on CT. Chen et al. reported that ¹⁸F-FDG PET detected 20 of 30 (67%) recurrent HCCs in patients showing increasing AFP concentrations after surgical resection or interventional therapy, when results of conventional examinations were normal (31). These findings suggest that ¹⁸F-FDG PET/CT may be useful for detecting viable lipiodolized HCCs in patients with increased serum AFP but in whom no demonstrable viability is seen on multiphasic contrast-enhanced CT images.

After TACE, HCCs undergo ischemic necrosis, and liver tissue surrounding the HCC may become inflamed, especially during the early postembolic period (32). ¹⁸F-FDG accumulates in inflammatory cells such as neutrophils, lymphocytes, and macrophages (33). Of the 8 nonviable HCCs studied in the present report, 3 showed increased ¹⁸F-FDG uptake (false-positives). Because of the retrospective design of our study, how much inflammatory cells were infiltrated in false-positive cases could not be confirmed. However, because all of our false-positive cases were observed in the early postembolic period and 75% of viable HCCs in the early postembolic period showed increased ¹⁸F-FDG uptake regardless of pathologic grade, it was determined that inflammatory cells may contribute to the overall ¹⁸F-FDG uptake in nonviable lipiodolized HCCs, but also viable tumors, within 3 mo after TACE.

The presence of highly attenuating materials, such as metal and lipiodol (34), during ¹⁸F-FDG PET/CT may result in artifacts. If lipiodol affects ¹⁸F-FDG uptake, the error must be present more or less in lipiodolized HCCs. Uncorrected images may be helpful if the lesions are located in the periphery of the liver but not for central lesions. Moreover, respiratory motion artifacts at the interface between the lungs and diaphragm may be present on CT–attenuation-corrected PET images. These artifacts may be caused by undercorrection of attenuation of the upper liver, because lung volumes are larger on CT scans than on emission PET images (35,36). Therefore, we evaluated PET/CT images, including PET images that were not corrected for attenuation, to detect any artificially increased activity. We observed 2 lipiodolized HCCs with falsely elevated activity on attenuation-corrected, compared with noncorrected, PET/CT images, suggesting the importance of evaluating noncorrected images in the detection of viable lipiodolized HCCs.

Our study had several limitations. First, the retrospective design may introduce biases in the diagnostic accuracy of ¹⁸F-FDG PET/CT and multiphasic contrast-enhanced CT. Moreover, we included only patients who underwent surgery after ¹⁸F-FDG PET/CT for the determination of pathologic correlations. Thus, the relatively small number of nonviable HCCs may have affected the diagnostic specificity data. Second, we did not quantitatively measure ¹⁸F-FDG uptake. Although SUV is the most commonly used quantitative measurement of ¹⁸F-FDG uptake, SUV can be influenced by factors other than glucose metabolism and there is no absolute threshold for determining tumor viability. Because viable HCCs can be detected by contrast between the tumor and the surrounding liver, we evaluated HCC uptake of ¹⁸F-FDG by semiquantitative visual comparison with adjacent normal tissue. Increased or similar tumor ¹⁸F-FDG uptake, compared with that in surrounding normal liver, as measured by the SUV ratio, has been reported to indicate tumor viability after interventional therapy (14). Third, we defined HCCs with less than 99% necrosis on histology as viable. Because a small lipiodolized HCC with more than 90% necrosis may show decreased ¹⁸F-FDG uptake despite the presence of residual viable tumor, as a result of a partial-volume effect, care should be taken in evaluating such small tumors. However, in our study, there was only 1 lipiodolized HCC with more than 90% necrosis, which showed increased ¹⁸F-FDG uptake.