ORIGINAL ARTICLE 
Annals of Nuclear Medicine Vol. 8 No. 2, 139-145, 1994 

Analysis of factors affecting uptake of Tc-99m Sn-N-pyridoxyl-5-methyltryptophan by hepatocellular carcinoma 

Yoshihisa HASEGAWA,1 Shunichi NAKANO,1 Tomotaka SOBUE,2 Makoto FUJITA,3 Shingo ISHIGURO,4 Yo SASAKI,5 Shingi IMAOKA,5 Sachiko TANAKA,6 Hiroshi KASUGAI,6 Atsuo INOUE 6 and Jyunnosuke KOJIMA 6 

Departments of 1 Nuclear Medicine, 2 Field Research, 3 Radiology, 4 Pathology, 5 Surgery, and 6 Internal Medicine, The Center for Adult Diseases, Osaka 

We performed Tc-99m PMT imaging in 176 patients with HCC and evaluated factors affecting 99mTc-PMT uptake by HCC with a logistic model. The probability of HCC showing increase in uptake of the radioisotope was 104.6 times higher in patients with the Ed I type than in those with the Ed III type and 12.1 times higher in patients with a tumor diameter of 5.0-7.9 cm than in those with a tumor diameter of 2.0-5.0 cm. Among the other variables, the serum AFP level and sex were suggested to have effects similar to those of the tumor size on Tc-99m PMT uptake by HCC. The grade of morphological differentiation of the tumor was therefore most markedly related to Tc-99m PMT uptake. 

Key words: Tc-99m PMT, hepatocellular carcinoma, grade of tumor differentiation 

INTRODUCTION 
RECENT ADVANCES in various imaging techniques and treatment methods have markedly improved the prognosis of hepatocellular carcinoma (HCC). 1-4 For the diagnosis of HCC, selective arteriography and liver biopsy under ultrasonic guidance are often performed. 5-7 However, these techniques are invasive, and there is need for noninvasive imaging methods that allow differentiation of HCC from other liver tumors. 
Hepatobiliary imaging is useful for the qualitative diagnosis of HCC. 8,9 Among agents used in this imaging, Tc-99m (Sn) N-pyridoxyl-5-methyltryptophan (Tc-99m PMT, Nihon Medi-Physics, Takarazuka, Japan) provides clear visualization of HCC in distinction from the surrounding liver areas. 10 However, the rate of clear visualization of HCC with Tc-99m PMT is only 51-56% 8,10 In order to make clear the role of Tc-99m PMT imaging in qualitative diagnosis of HCC, it is important to clarify factors affecting Tc-99m PMT uptake by hepatic tumor, and further to determine the degree of their contribution if there are multiple factors. 

Received July 5, 1993, revision accepted February 16, 1994. 
For reprint contact: Yoshihisa Hasegawa, M.D., Department of Nuclear Medicine, The Center for Adult Diseases, Osaka, 3-3, Nakamichi-1-chome Higashinari-ku, Osaka 537, JAPAN. 

Logistic regression analysis has recently been used as a method of multivariate analysis to evaluate the causes of disease in the epidemiological field. When several factors are involved in the development of a certain result, the degree of enhancement of the probability of the result by one factor can be estimated by this method. 11 
In this study, we analyzed possible factors associated with HCC-positive findings in Tc-99m PMT imaging with a logistic model. 

MATERIALS AND METHODS 
The subjects consisted of 176 patients (143 males and 33 females) with HCC. Their age ranged from 36 to 76 years (mean, 58.7 years). The diagnosis of HCC was made by histological examination in 64 patients, and selective arteriography as well as ultrasonography, CT, and the serum AFP level in the other 112. In the 64 patients, the grade of differentiation of tumor cells was classified according to Edmondson and Steiner's criteria 12 by histological examination after evaluation of Tc-99m PMT uptake. For Tc-99m PMT imaging, 185-555 MBq (5-15 mCi) of Tc-99m PMT was intravenously injected, and imaging of the liver was performed after 5 minutes and 1 , 2, 3, and 5 hours. Tc-99m PMT uptake by the tumor was measured on images 3-5 hours after injection. The tumor sites were determined by Tc-99m colloid imaging, CT scan, and selective arteriography. When multiple tumors were present in the liver, the uptake by the largest tumor was measured. The Tc-99m PMT uptake by the tumor was compared with that by the non-tumor area of the liver and was classified into the following two groups. ( 1 ) increased uptake (more radioactivity in the tumor than in the surrounding liver tissues). (2) non-increased uptake (radioactivity in the tumor almost equal to or less than that in the surrounding liver tissues). The association between Tc-99m PMT uptake by HCC and various characteristics of the patient (sex, age, and serum bilirubin) and the tumor (the grade of histological differentiation, tumor size, and serum AFP) was evaluated. Patients with HCC less than 2.0 cm in diameter were excluded from the study, because they showed a very low incidence of positive results in a Tc-99m PMT study. The significance of the relationship between Tc-99m PMT uptake by HCC and each variable was analyzed by chi-square test. In addition, to exclude the confounding effects among various variables, analysis was done with a logistic model. 11 At first, in all 176 patients, the association between radioactivity uptake by HCC and variables excluding the grade of histological differentiation was analyzed. A similar analysis was then done in 64 patients histologically examined. In the analysis with a logistic model, the covariate variables (sex, age, serum bilirubin, serum AFP, tumor size, and Edmondson's classification) were classified into 2-4 categories depending on the nature of each. For the covariate variables that have n categories, n-1 dummy variables, the values of which were set to be 0 or 1 for each patient, were made and included in the logistic model simultaneously. Odds ratios were calculated by taking the natural exponent of estimated beta coefficients for each dummy variable. In this study, the odds ratio indicates the ratio of the effect of a category on the increase in the incidence of positive results in Tc-99m PMT images to that of a designated baseline category for each factor (positive rate ratio) (Table 3). Analysis was done with PROC LOGISTIC of SAS. 13 

RESULTS 
Tc-99m PMT uptake by HCC and various variables The relationship between Tc-99m PMT uptake by HCC and various variables excluding the grade of histological differentiation was evaluated in the 176 patients with HCC. In Tc-99m PMT images, the incidence of positive results was higher in the females than in the males, in patients aged >= 60 years than in those < 60 years, in patients with an AFP level < 400 ng/ml than in those with a level >= 400 ng/ml, and in patients with a tumor 5.0-7.9 cm in diameter than in those with a tumor 2.0-4.9 cm or >= 8.0 cm in diameter. Among these variables, sex (p = 0.044), and the serum AFP level (p = 0.026) were significantly related to HCC-positive findings (Table 1). 
In the 64 patients histologically examined, there were relationships similar to those observed in the 176 patients between HCC-positive findings and various variables excluding age and serum bilirubin. However, no significant differences were found in the chi-square test. Only the grade of histological differentiation was significantly related to HCC-positive findings (p = 0.01, Table 3). 

Analysis with a logistic model 
( 1 ) Tc-99m PMT uptake by HCC and various variables except the grade of histological differentiation in 176 patients with HCC: Analysis with the logistic model showed a definite relationship between HCC-positive findings and sex, the serum AFP level and the tumor size. In Tc-99m PMT images, the probability of positive findings (positive rate ratio) was 0.41 in the males when the females were used as a reference (1.0), and 0.30 in patients showing an AFP level > 1,000 ng/ml when patients with an AFP < 20 ng/ml were used as a reference. Similarly, the positive rate ratio was 3.0 in patients with a tumor 5.0-7.9 cm in diameter when those with a tumor 2.0-4.9 cm were used as a reference, showing a significant difference (Table 2). 
(2) Tc-99m PMT uptake by HCC and various variables in 64 patients histologically examined: Analysis with the logistic model showed a definite relationship between HCC-positive findings and the grade of histological differentiation as well as the tumor size. The probability of positive findings in Tc-99m PMT images (positive rate ratio) in patients with Edmondson I (Ed I) type HCC was 104.6 when those with Ed III HCC were used as a reference, being significantly high. On the other hand, the positive rate ratio in patients with Ed II HCC was 1.8, being only slightly high. The positive rate ratio in patients with a tumor 5.0-7.9 cm in diameter was 12.1 when those with a tumor 2.0-4.9 cm in diameter were used as a reference, showing statistical significance (Table 4). 

CASES 
Patient 1: a 59-year-old male 
He visited our hospital due to general malaise persisting for 3 months. Since ultrasonography showed a tumor (about 6 cm in diameter) in the posterior segment of the right liver lobe, he was admitted. On admission, the serum bilirubin level was 1.2 mg/dl, and the serum AFP level was less than 20 ng/ml. A Tc-99m Sn colloid image (Fig. 1A) showed a defect in the posterior segment of the right lobe. Imaging 5 hours after a Tc-99m PMT injection (Fig. 1B) revealed marked uptake at the site of the defect. Selective arteriography showed a hypervascular tumor in S7. Right lobectomy was performed, and the diameter of the resected tumor was 6.0 cm. Histopathologically, the tumor primarily consisted of Ed I type HCC. 

Patient 2: a 55-year-old male He visited our hospital due to decreased appetite persisting for about I month. Since ultrasonography showed a tumor (about 10 cm) in the right hepatic lobe, he was admitted. The serum bilirubin level was 0.9 mg/dl, and the serum AFP level was less than 20 ng/ml. Liver imaging with Tc-99m Sn phytate (Fig. 2A) demonstrated a large filling defect in the dorsal inferior portion of the right hepatic lobe. Tc-99m PMT imaging (Fig. 2B) showed an accumulation of radioactivity in this defect similar to that in the surrounding liver areas. Selective arteriography revealed a large hypervascular tumor extending from S6 to S7 (Fig. 2C). Right lobectomy was performed, and the size of the resected tumor was 13 cm. Histologically, the tumor primarily consisted of Ed III type HCC.

DISCUSSION 
Hepatobiliary imaging is considered to be useful for the qualitative diagnosis of HCC. 8,9 Tc-99m PMT is superior to other agents for hepatobiliary imaging in clear visualization of HCC as a positive image. However, only about half of HCCs are positively visualized even by Tc-99m PMT imaging. The other HCCs are difficult to differentiate from other hepatic tumors. In this study, we analyzed patient and tumor factors affecting the results of Tc-99m PMT imaging of HCC. The possible association between Tc-99m PMT uptake by HCC and various variables excluding the grade of histological differentiation was evaluated with a logistic model in the 176 patients. The incidence of visualization of HCC as a positive image (positive rate ratio) was 3.0 times higher in patients with an HCC size of 5.0-7.9 cm than in those with an HCC size of 2.0-4.9 cm. The positive rate ratio was 0.30 times lower in patients with an AFP >= 1,000 ng/ml than in those with an AFP < 20 ng/ml and 0.41 times lower in the males than in the females. In analysis with a logistic model, when A in two categories (A and B) is used as a reference for a certain variable, the reciprocal of the positive rate ratio of the other (B) is considered to approximate the positive rate ratio of the former (A) when the latter (B) is used as a reference. Therefore, the probability of positive findings increased by 3.3 (1/0.30) times in patients with an AFP level < 20 ng/ml when patients with an AFP level >= 1,000 ng/ml were used as a reference and by 2.4 times (1/0.41) in the females when males were used as a reference. Analysis of HCC in the 176 patients showed that 3 variables (sex, serum AFP level and tumor size) were related to Tc-99m PMT uptake by the hepatic tumor to similar degrees. 
On the other hand, in the 64 patients with HCC in whom the grade of histological differentiation could be determined, the probability of HCC-positive findings was 105 times higher in patients with Ed I type HCC than in those with Ed III type HCC, the difference being significant. The positive rate ratio was 12.1 times higher in patients with an HCC 5.0-7.9 cm in diameter than in those with an HCC 2.0-4.9 cm in diameter, the difference being significant. The ratio was also 2.7 times higher in patients with an HCC of more than 8.0 cm in diameter than in those with an HCC 2.0-4.9 cm in diameter, although the difference was not significant. 
The uptake of a hepatobiliary imaging agent by the HCC is thought to be due to the hepatocytic function remaining in the hepatic tumor. A functionally differentiated HCC is also presumed to be morphologically differentiated. The degree of histologic differentiation of HCC has been reported to be significantly correlated with the uptake of hepatobiliary imaging agents by the hepatic tumor. 14 However, it is important to quantitatively estimate the pure effect of histological features on the tumor uptake of Tc-99m PMT after removing the effects of other factors. In the present study with a logistic model, the grade of morphological differentiation was shown to be the predominant determining factor in the taking up of Tc-99m PMT by HCC among the various factors studied. The tumor size was also an independent factor influencing the Tc-99m PMT uptake by HCC after adjustment for the effects of other variables, even though it had a smaller effect on the tumor uptake of radioactivity than the histological findings. The lower incidence of HCC showing positive results on Tc-99m PMT imaging in patients with an HCC 2.0-4.9 cm in diameter than in those with an HCC 5.0-7.9 cm in diameter might partially be due to the limitation in detection of small HCC by this method. On the other hand, the lower odds ratio for positive visualization of HCC in patients with a tumor larger than 8.0 cm in diameter as compared with that in those 5.0-7.9 cm in diameter might be caused by the presence of hypovascularity and the necrotic tissues in the advanced tumors. 15 
Unlike the analysis in all the patients, analysis involving histological findings in the 64 patients showed no definite effects of serum AFP or sex on Tc-99m PMT uptake by HCC. This may be due to the smaller number of patients and addition of the grade of histological differentiation to the variables. In the 64 patients histologically examined, variables excluding Edmondson's classification were analyzed with a logistic model. The positive rate ratio between categories with different AFP levels and between the sexes was similar to that obtained in the analysis of the 176 patients, though it was not significant. In the 64 patients, sex and the AFP level may also have affected Tc-99m PMT uptake by HCC. In the animal experiments, the serum AFP level was found to be related to the grade of histological differentiation. 16,17 This relationship may explain why the serum AFP level affected the Tc-99m PMT uptake by HCC in the study of the 176 patients. However, in the analysis of the 64 patients, the marked effects of the grade of histological differentiation on 99mTc-PMT uptake by HCC might have resulted in underestimation of the effects of AFP. In contrast, the grade of morphological differentiation of HCC has not been reported to differ with the sex, to our knowledge. The reason why sex was seen to affect Tc-99m PMT uptake by HCC in the study of the 176 patients, therefore, is not yet clear. In order to determine whether AFP and sex are associated with HCC-positive findings in 99mTc-PMT imaging, additional cases of HCC at known grades of histological differentiation are needed. 
Extrahepatic metastases of HCC taking up hepatobiliary imaging agents have been reported to frequently produce bile pigments in tumor cells. 18,19 However, it has not yet been clarified whether the uptake of a hepatobiliary imaging agent by the primary tumor of HCC is related to the bile formation in the tumor tissues or not. Of the 64 patients with HCC examined histologically, only 7 showed morphological evidence of bile formation in the tumor tissues of the liver. All 7 patients showed increased uptake of Tc-99m PMT by the hepatic tumors, whereas 32 (56.1%) of the remaining 57 patients who showed no bile pigments in the tumor tissues, did not take up a great deal of Tc-99m PMT (p < 0.02). The ability of hepatic tumors to produce bile was therefore also proved to be closely correlated with Tc-99m PMT uptake by the hepatic tumors. In the estimation of the effects of various factors including the effect of bile production on tumor uptake of Tc-99m PMT with a logistic model, the positive rate ratios for each level of variables did not converge into meaningful values as a whole. The difficulty in the analysis might be caused by the low frequency of HCC production of bile pigments despite the close relationship between the ability to form bile in the tumor cells and the tumor uptake of radioactivity. To accurately evaluate the effect of bile formation by HCC on Tc-99m PMT uptake by the hepatic tumor, it is necessary to analyze more cases of HCC examined histologically for the presence of bile pigments. 
In our study, the grade of Tc-99m PMT uptake by the hepatic tumor was classified into two groups; "increased uptake" and "non-increased uptake". The latter group included two conditions of radioactivity uptake by the hepatic tumor; 1 ) equal uptake which showed radioactivity in the hepatic tumor equal to that in the surrounding liver tissues, and 2) decreased uptake showing less radio-activity in the hepatic tumor than in the surrounding liver. However, in patients with small hepatic tumors which did not show any filling defect on the liver images with Tc-99m colloid or Tc-99m PMT, it is difficult to differentiate the two conditions of Tc-99m PMT uptake by the hepatic tumor. Therefore, these two grades of Tc-99m PMT uptake by the hepatic tumor were combined into the group "non-increased uptake". 
In the present study, the grade of morphological differentiation was most markedly related to Tc-99m PMT uptake by HCC. The tumor size was also related, but to a lesser degree. The serum AFP level and sex had effects similar to those of the tumor size. Thus, imaging of the hepatobiliary system not only is useful for qualitatively diagnosing HCC but also provides information associated with the grade of morphological differentiation. 
A recent report suggested the usefulness of MRI for differentiating early stage Ed I type HCC from Ed II or III type HCC. 20 However, in about a half of Ed I liver tumors, the tumor and the surrounding areas were noted to be isointense on T1- and T2- weighted images. 20 On the other hand, there is a report that evaluated the grade of the differentiation of glucose metabolism of HCC by means of positron camera and 18FDG. 21 This method is still very expensive. In future, however, the two methods may be used to determine the grade of HCC differentiation. 
Even Ed III HCC incorporates Tc-99m PMT, though its incidence is lower than Ed I and II HCC. Therefore, the Tc-99m PMT uptake by HCC is not always in complete agreement with the grade of morphological differentiation. The uptake of a hepatobiliary agent by HCC originally represents the degree of functional differentiation of tumor cells. Tc-99m PMT imaging is a simple and inexpensive method. We hope that this method, which provides a specific index of the grade of HCC differentiation, will be widely used. 

REFERENCES 
1. Liaw YF, Tai DI, Chu CM, Lin DY, Sheen IS. Chen TJ, et al. Early detection of hepatocellular carcinoma in patients with chronic type B hepatitis: a prospective study. Gastroenterology 90: 263-267, 1986. 
2. Yamada R, Sato M, Kawabata M, et al. Hepatic artery embolization in 120 patients with unresectable hepatoma. Radiology 148: 397-401, 1983. 
3. Kasugai H, Kojima J, Tatsuta M, Okuda S, Sasaki Y, Imaoka S, et al. Treatment of hepatocellular carcinoma by transcatheter arterial embolization combined with intraarterial infusion of a mixture of cisplatin and ethodized oil. Gastroenterology 97: 965-971, 1989. 
4. The liver cancer study group of Japan. Primary liver cancer in Japan. Cancer 60: 1400-1411, 1987. 
5. Boijsen E, Abrams HL. Roentgenologic diagnosis of primary cancer of the liver. Acta radiologica (diag.) 3: 257-277, 1965. 
6. Reuter SR, Redman HC, Siders DB. The spectrum angiographic findings in hepatoma. Radiology 94: 89-94, 1970. 
7. Tanaka S. Kitamura T, Kasugai H, Okano Y, Tatsuta M, Okuda S. Early diagnosis of hepatocellular carcinoma: usefulness of ultrasonically guided fine-needle aspiration biopsy. J Clin Ultrasound 14: 11-16, 1986. 
8. Hasegawa Y, Nakano S, Ibuka K. Hashizume T, Noguchi A, Sasaki Y, et al. Specific diagnosis of hepatocellular carcinoma by delayed hepatobiliary imaging. Cancer 57: 230-236, 1986. 
9. Lee VW, O'Brien MJ, Devereux DF. Morris PM, Shapiro JH. Hepatocellular carcinoma: Uptake of Tc-99m IDA in primary tumor and metastasis. AJR 143: 57-61, 1984. 
10. Hasegawa Y, Nakano S, Hashizume T, Noguchi A, Ibuka K, Kasugai H, et al. Comparison of delayed imaging with Tc-99m PMT and Tc-99m DEIDA for visualization of hepatoma. Clin Nucl Med 14: 526-531, 1989. 
11. Breslow NE, Day NE, Davis W. Unconditional logistic regression for large strata. In Statistical methods in cancer research. Vol. 1. The analysis of case-control studies. IARC Scientific publications, No. 32, pp. 191-246, 1980. 
12. Edmondson HA, Steiner PE. Primary carcinoma of the liver. A study of 100 cases among 48,900 necropsies, Cancer 7: 462-503, 1954. 
13. SAS Institute Inc. The logistic procedure. In SAS technical report. Cary, North Carolina: SAS Institute Inc., p-200, pp. 175-230, 1990. 
14. Calvet X, Pons F. Bruix J, et al. Technetium-99m-DISIDA hepatobiliary agent in diagnosis hepatocellular carcinoma: relationship between detectability and tumor differentiation. J Nucl Med 29: 1916-1920, 1988. 
15. Nakashima T. Vascular changes and hemodynamics in hepatocellular carcinoma. In Hepatocellular Carcinoma. Okuda K, and Peters RL (ed.). New York: John Wiley and Sons, pp. 169-203, 1976. 
16. Becker EF, Klein KM, Wolmann SR, et al. Characterization of hepatocellular carcinomas and initial transplantation generations. Cancer Research 33: 3330-3338, 1973. 
17. Sell S, Morris HP. Relationship of rat alpha-1-fetoprotein to growth rate and chromosome composition of Morris hepatoma. Cancer Research 34: 1413-1417, 1984. 
18. Yasunaga H, Hirota T, Tsuchigame T, et al. Clinical significance of Tc-99m HIDA scintigraphy for hepatomas and detection of their distant metastasis. 19: 523-528, 1982. 
19. Lee VW, Shapiro JH. Specific diagnosis of hepatoma using Tc-99m HIDA and other radionuclides. Eur J Nucl Med 8: 191-195, 1983. 
20. Muramatsu Y, Nawano S, Takayasu S, Moriyama N, Yamada T, Yamasaki S, et al. Early hepatocellular carcinoma: MR imaging. Radiology 181: 209-213, 1991. 
21. Enomoto K, Fukunaga T, Okazumi S, Asano T, Kikuchi T, Yamamoto H, et al. Can fluorodeoxyglucose positron emission tomography evaluate the functional differentiation of hepatocellular carcinoma. KAKU IGAKU (Jpn J Nucl Med) 28: 1353-1356, 1991.