Risk Factors Linked to Tumor Recurrence of Human Hepatocellular Carcinoma After Hepatic Resection SHYH-CHUAN JWO.' JEN-HWEI'CHI[
,' ' GAR-YANGCHAU,'
CHE-CHUAN LOONG'AND WING-YU LUI'
'Division of General Surgery Department of Surgery. Veterans General Hospital, and "Graduate Institute of Clinical Medicine. National Yang-Ming Medical College Taipei, Taiwan I 1 2 1 7, Republic of China
A total of 238 patients who received curative hepatic resectionsduring the last 10 yr were observed to search for the risk factors linked to early tumor recurrence of human hepatocellular carcinoma after hepatectomy. The results revealed that tumor size, tumor appearance and DNA ploidy were the factors in predicting tumor recurrence after resection for hepatocellular carcinoma. Patients with a tumor size less than or equal to 5 cm or a tumor appearance of the solitary type had better disease-free survival than did those with a tumor size greater than 5 cm or a tumor appearance of multiple/daughter nodule types (p < 0.06). Although patients with pattern III (aneuploid with 2 2 GO/Gl peaks) hepatoma had fewer statistically significant differences (p = 0.19) than did those with pattern I (diploid) or pattern II (aneuploid with single GO/G1 peak) tumors in predicting tumor recurrence,they did have poorer results in terms of the overall survival rate (p < 0.05). We conclude that patients with hepatocellular carcinoma having the aforementioned risk factors should be observed 1992: 16:1367-137 1.1 closely. (HEPATOLOGY
HCC is one of the most malignant diseases in Asian countries, and it also is the leading cause of cancer death in men and the third leading cause in women in Taiwan (1).Although much effort has been devoted to curative treatment of HCC and the results have given us some encouragement for the short term ( 2 - 4 ~ ,long-term results are still unsatisfactory. It has been generally accepted that postoperative tumor recurrence is the main cause of poor prognosis in patients with resectable HCC (5-7).Many factors. including tumor size, cut margin and portal vein invasion, for example, have been studied to predict the risk of tumor recurrence after resection for HCC (8-10). Because some controversy exists regarding the factors just described, different
Received May 4, 1992; accepted July 21, 1992 This work was partly supported by a grant from the (:linical Research Centpr. Veterans General Hospital. Institute of Biomedical .Science. Academia Sinica. Republic of China. Address reprint requests to. Shyh-('huan Jwu.Divlsion of General Surgery Department of Surgery, Veterans Grnrral Hospital. Rhih-Pal. Taipei. Taiwan 11217, Republic of China 31/1/41673
mechanisms leading to early and late recurrence have been proposed (8-10).Previous studies have shown that measurement of DNA content is useful in correlating biological behavior with the prognosis of neoplasms t 11- 13 ) . It has been suggested that change of DNA ploidy in human HCC has no correlation with survival of patients with HCC after hepatectomy (14,15).However, different conclusions have been drawn in some series (16). Recently, the change of DNA ploidy in human hepatomas was grouped into three patterns using a newly classified system. The patterns included the following: pattern I is a diploid tumor; pattern I1 is an aneuploid tumor with a single GO/G1 peak;and pattern I11 is an aneuploid tumor with more than one GO/Gl peak. We found that patients with pattern I11 hepatoma have had a tendency for early tumor recurrence after hepatic resection. Therefore we are interested in searching for changes of DNA content in conjunction with other documented factors to identify patients who will have a high risk of early tumor recurrence after hepatic resection. MATERLALS AND METHODS Between August 1981 and August 1991,3,649 patients with documented primary HCC were admitted to Veterans General Hospital, Taipei, Taiwan, and 308 of them received laparotomy for resection of HCC. Excluding those with palliative resection and operative mortality, 238 patients receiving curative hepatic resection were retrospectively reviewed in correlation with tumor recurrence and various risk factors, such as the presence of HBsAg, cirrhosis, tumor size, tumor appearance, tumor grading, tumor ploidy and cut margin, for example. After curative resection for HCC, which was defined as total removal of gross tumors (4, 9), patients were observed every 2 to 4 mo with a-fetoprotein; abdominal ultrasonography, abdominal computerized tomography or both; and celiac angography if necessary. Diagnosis of tumor recurrence was made when at least two of those four modalities were interpreted as positive findings. The mean follow-up time was 55.9 mo. The presence of HBsAg and cirrhosis were from welldocumented records. The resected specimens were recorded in tumor size, tumor appearance and gross cut margin. The cut margin was defined as the shortest distance between the tumor edge and the cut surface of the resected specimens. Tumor size was measured as the maximal diameter of the biggest tumor.
1367
1368
JWO ET AL.
HEPATOLOGY
extrahepatic recurrence 14 ( 6 . 9 % )
‘a and i n l r a tic recurrence
(8.8%)
FIG.1. Distribution of postoperative status in 238 patients receiving curative resection for HCC.
The gross tumor appearance was categorized into three types, which included the following: solitary tumor; tumor with daughter nodules in the same sector or segment (Couinaud’s classification)(17);and multiple tumors in different sectors or segments but confined in the same lobe. The specimens were also fixed with 10%formalin and stained with hematoxylin and eosin for examination of their pathological change. The differentiation of tumor cells was graded into four degrees as described by Edmondson and Steiner (18). DNA content, analyzed with fresh, frozen tissue or paraffinembedded samples, was measured with a flow cytometer (Epics Profile; Coulter Electronics, Hialeah, FL) as previously described (19, 20). The ploidy status was defined as diploid (2N) when the GO/G1 peak was superimposed after the addition of human peripheral blood lymphocytes to the samples with a range from 1.9N to 2.1N; aneuploid was defined if the discrete GO/G1 peak or peaks were outside the diploid range (1.9N to 2.1N). The DNA ploidy status of human hepatomas could be classified into three patterns. These patterns were the following: pattern I included hepatomas with diploid DNA distribution; pattern I1 included hepatomas with aneuploid DNA content and with only one prominent GO/Gl peak, and pattern I11 included hepatomas with aneuploid DNA content and with more than one GO/G1 peak. The data were analyzed using a statistical package program (BMDP Statistical Software, Inc., Los Angeles, CA). We used the BMDP 4F univariate and 2L multivariate analysis program to analyze risk factors for tumor recurrence and the 1L program to calculate disease-free survival by the Kaplan and Meier method (21). We compared the difference between categories in each prognostic factor using the generalized Wilcoxon’s test and Savage (Mantel-Cox) test. Any p values less than 0.05 were considered to be statistically sigdicant.
RESULTS
Among 3,649patients with HCC, 308 received laparotomies for resection of HCC. The operability rate was 8.4%, and the curative resectability rate was 86.7% (267 of 308 patients). The operative mortality rate was 8.4% (26 of 308). Among 238 patients receiving curative resection, 122 (51.3%) had tumor recurrence, and, of these 122 patients, 87 (71.3%) had intrahepatic recurrence. Fourteen patients (11.5%) had extrahepatic recurrence ( 1patient with a brain tumor, 1 patient with a lung tumor, 1 patient with a bone tumor and 11 patients with tumors affecting more than two organs); 21 patients t 17.2%) had both intrahepatic and extrahepatic recurrence (12 patients with lung tumors, 7 patients with bone tumors and 2 patients with chest wall tumors) (Fig. 1). Risk factors such as sex, presence of HBsAg, cirrhosis, tumor size, tumor appearance, tumor grading, tumor ploidy and cut margin were investigated in correlation with tumor recurrence. The results are shown in Table 1. No statistically significant difference was seen between different distances of cut margin and different differentiations of the Edmondson-Steiner tumor grading. Tumor size, tumor appearance and sex were statistically significant risk factors for tumor recurrence both by univariate and multivariate analysis (p < 0.05). Patients with HCC less than or equal to 5 cm had higher disease-free survival rates than did those with HCC greater than 5 cm (p < 0.005)(Fig. 2). Patients with a tumor appearance of the solitary type had higher disease-free survival rates than did those with tumors of
1369
FACTORS LINKED TO RECURRENCE OF HCC
Vol. 16, No. 6, 1992
TABLE1. Risk factors for tumor recurrence after resection of HCC
Rislrfactora
No. of patients
Tumor recurrence No
Yes
Lost follow-up
216 22
87 12
115 7
14 3
p p
= =
0.001" 0.035*
58 150
26 56
25 88
7 6
p
=
0.338
73 153
33 61
36 83
4
p
=
0.662
141 62 8
64 22 1
64 38 6
13 2
p p
=
0.050" 0.017'
111 115
54 38
50 68
7 9
p p
=
*--A
T55 ( n = l l l )
*-*B
T>5(n=115)
pvalue
Sex
M F HBsAg Negative Positive Cirrhosis Absent Present Tumor appearance solitary Daughter nodule Multiple nodules Tumor size 4 5 em > 5 em Tumor grading I
II III
Iv Tumor ploidyd I
n
n1
=
1
14 45 34 2
5 18 14 0
8 23 17 1
1 4 3 1
p
92 18 47
36 6 26
51 11 21
5 1 0
p
89 53 26
39 21 12
45 27 13
5 5 1
p
= =
0.026" 0.036'' 0.809
I
20
40
60
80
1ZO(rnonlhs)
100
FIG. 2. Disease-free survival after curative resection for HCC according to different tumor size. A = tumor size I 5 em; B = tumor size > 5 cm. A statistically sigmficant difference was seen between A and B (p < 0.005).
(X)
I00
o-oA:
90
'-*B.
80
A-b
(n=l41) (n= 62) C. (n= 8 )
70
60 =
0.192
50 40
30
Surgical margin < 1 em 1to3cm > 3 cm
04 0
9
20
=
0.931
"statistieally significant in univariate analysis. *Statistidy significant in multivariate analysis. 'Graded by Edmondson and Steiner's classification (18). dDNA ploidy, patterns I, I1 and 111, as defined in "Materials and Methods."
10
04 0
I
20
40
60
80
100
lZO(mon1hr)
FIG. 3. Disease-free survival after curative resection for HCC according to variation in tumor appearance. A = solitary tumor; B = tumor with daughter nodules; C = tumor with multiple nodules in different sectors or segments. Statistically s@cant differences were seen between A and B (p < 0.03) and A and C (p < 0.002), but no significant difference was seen between B and C (p = 0.11).
Because recurrence of HCC is frequent and the main cause of poor prognosis after extirpation of HCC, many factors have been studied to predict the risk of tumor recurrence after hepatic resection. Previous results concerning the predominant ratio and higher recurrent rate of HCC in men were explained by the presence of the androgen-dependent hormone receptor in hepatoma cells (22). Because this predominance in male patients is characteristic in our hospital and causes a sampling bias in patient selection, it is hard to draw any conclusion about the relationship between sex and the recurrence of HCC. Yoshida et al. (9) considered that the cut margin was not linked to tumor recurrence in small HCC ( 5 4 cm), whether or not an adequate cut margin ( 2 1 cm) was there, because the opportunity for portal vein or capsule invasion was very low. On the other hand, a DISCUSSION correlation existed between an inadequate cut margin The early diagnosis of HCC by developed diagnostic ( < 1 cm) and tumor recurrence in large HCC ( > 4 cm) modalities and improved surgical techniques has in- because the opportunity for portal vein or invasion creased the operability and resectability of human HCC. increased (9). In our results, patients with HCC less
daughter nodules (p < 0.03) or tumors of multiple types (p c 0.002) (Fig. 3). Although a less statistically significant difference existed between patients with HCC of different DNA ploidy in terms of the disease-free survival rate (Fig. 41, patients with pattern I11 HCC had a greater tendency for early tumor recurrence after hepatic resection than did those with pattern I or pattern 11 HCC. The median disease-free survival of patients with HCC of DNA patterns I, I1 and I11 was 19 mo, 17 mo and 8 mo, respectively. In addition, patients with pattern I11 HCC had poorer results in terms of overall survival rates (p < 0.02) (Fig. 5). The results suggested that HCC with pattern I11 DNA ploidy status had more malignant tumor behavior than did HCC with pattern I or pattern I1 DNA ploidy.
1370
JWO ET AL.
HEPATOLOGY
:“t 70
60 .50
~-
40
-.
30 -20 --
:
10.-
+
04
20
40
60
80
I00
120 (months)
FIG. 5. Overall survival rate after curative resection for HCC according to different DNA ploidy (patterns I, I1 and 111, as described in “Materials and Methods”). I = diploid HCC; I1 = aneuploid HCC containing only one prominent GO/G1 peak; IIZ = aneuploid HCC containing two or more GO/G1 peaks. A statistically significant difference was seen between patterns I and I11 (p < 0.003), but no significant difference was seen between patterns I and I1 (p = 0.26) or patterns I1 and I11 (p = 0.27).
than or equal to 5 cm had a significantly higher disease-free survival rate than did those with HCC greater than 5 cm; however, no significant differences in various distances of the cut margin ( < 1 cm, 1 to 3 cm and > 3 cm) were noted. It was generally accepted that patients with solitary HCC had a better prognosis than did those with HCC having daughter or multiple nodules in terms of overall survival (2,23). In this study, patients with solitary HCC also had better disease-free survival than did those with daughter or multiple nodules. Regarding the increased opportunity for tumors to spread, the possibility of multicentric metachronous or unnoticed synchronous lesions beyond the resected area has previously been considered to be the cause of tumor recurrence in HCC with the tumor appearance of daughter/multiple types (24).Because tumor grading by the Edmondson Steiner classification was not considered to be a factor for predicting tumor recurrence after resection of HCC in many studies (25,26), tumor behavior was analyzed by the measurement of DNA content and cell-cycle analysis. DNA ploidy of HCC, previously categorized into diploidy and nondiploidy, was reported to have no correlation with overall survival in patients with HCC after hepatic resection (14,151. However, Fujimoto et al. (27) and Ishizu (28)reported that patients with aneuploid HCC had a poorer overall survival rate than patients with diploid HCC. In our study, although the differences among newly classified DNA ploidy patterns I, I1 and I11 were less statistically significant (p = 0.191, patients with pattern I11 HCC tended to have a lower disease-free survival rate and a shorter median disease-free survival (8mo) than did those with pattern I or I1 HCC. It has been proposed that early tumor recurrence (within 2 yr) and late recurrence (after 2 yr) after resection of HCC are probably caused by different mechanisms. Some researchers have proposed that distant spreading of tumor thrombi by way of the portal
vein, direct invasion into the surrounding boundary by the original lesion, a multicentric metachronous lesion or unnoticed synchronous tumors were the main causes of early recurrence (9, 10, 24,291,whereas late recurrence in the liver remnant was a new lesion primary in origin (8, 30). Because the median disease-free survival in patieqts with DNA ploidy pattern I11 HCC ( 8 mo) is much shorter than that of pattern I1 or pattern I (17 mo and 19 mo, respectively), we speculate that tumor recurrence in patients with pattern 1/11and pattern I11 might involve a different mechanism. Thus a new lesion primary in origin might obscure our result and make it less statistically significant. However, further study on the molecular level and analysis is being conducted. In conclusion, those patients with resectable HCC who have high-risk factors, a large tumor ( > 5 cm), a tumor with daughter or multiple nodules or HCC with pattern I11 ploidy need to be observed closely after hepatic resection to detect early tumor recurrence. We also believe that measuring DNA ploidy may provide a new treatment strategy for HCC after curative hepatic resection. Acknowledgments: We thank L.H. Wu and L.J. Tai for their excellent technical assistance. REFERENCES 1. National Health Administration, Taiwan Provincial Health Department and Taipei City Health Department, Republic of China. Health statistics. Tapei: National Health Administration 1979;Z: 34-137. 2. Lee CS, Sung JL, Hwang LY, Sheu JC, Chen DS, Lin TY, Beasley P. Surgical treatment of 109 patients with symptomatic and asymptomatic hepatocellular carcinoma. Surgery 1986;99: 481-490. 3. Wu MC, Zhang XH, Chen H. Hepatic resection for primary liver cancer: report of 400 cases. Chin Med J 1986;99:175-180. 4. Kanematsu T, Matsumata T, Takenaka K, Yoshida Y, Higashi H, Sugimachi K. Clinical management of recurrent hepatocellular carcinoma after primary resection. Br J Surg 1988;75:203-206.
Vol 16, No 6, 1992
F.4C'TOKS LINKED TO RE('L7RRENCE OF HCC
5. Nagao T, Inone S, Goto S, Kawano N. Hepatic resection for hepatocellular carcinoma: clinical features and long-term prognosis. Ann Surg 1987;205:33-40. 6. Nagao T, Inone S, Yoshimi F. Sodeyama M. Omori Y,Mizuta T. Kawano N, et al. Postoperat.ive recurrence of hepatocellular carcinoma. Ann Surg 1990;211:28-33. 7. Nagasue N, Yukaya H, Chang YC. Yamanoi A, Kohno H. Hayashi T, Nakamura T. Assessment of pattern and treatment of intrahepatic recurrence after resection of hepatocellular carcinoma. Surg Gynecol Obstet 1990;171:217-222. 8. Lin TY, Lee CS, Chen KM. Chen CC. Role of surgery In treatment of primary carcinoma of the liver- a 31-war experience. Br J Surg 1987;74:839-842. 9 . Yoshida Y, Kanematsu T. Matsumata ' l ~ ,Takenaka K.Sugimachi K. Surgical margin and recurrence after resection of hepatocellular carcinoma in patients with cirrhosis. Ann Surg 1989;209: 297-301. 10. Shirabe K, Kanematsu T, Matsumata T, Adachi E. Akazawa K. Sugimachi K. Factors linked to early recurrence of small hepatocellular carcinoma after hepatectorny: univariate and multivariate 199 1 ;14:802-8O5. analysis. HEPATOLOGY 11. Watson JV.Oncogenes, cancer and analyt.ical q t o l o p . Cytometry 1986;7:400-410. 12. Young GAR, Hedley DW, Rugg CA. Iland HJ. The prognostic significance of proliferative activity in poor histology nonHodglun's lymphoma: a flow c-ytometry study using archival material. Eur J Cancer Clin Oncol 198723:1497- 1504. 13. Ewers SB, Langstrom E, Baldetorp B, IGllander D. Flowcytometric DNA analysis in primary breast carcinomas and clinicopathologid correlations. C.ytometry 1984:5:408-419. 14. Ezaki T, Kanematsu T, Okamura T. Sonoda T, Sugimachi K. DNA analysis of hepatocellular carcinoma and clinicopatholog~implications. Cancer 1988;61:106-109. 15. Saeter G , Lee CZ, Schwarze PE. Ous S. Chen DS, Sung JI,. Seglen PO. Changes in ploidy distributions in human liver carcinogenesis. Cancer 1988;80:1480-1485. 16. Clark GM, Dressler LG, Owens MA. Pounds G , Oldaker '1'. McGuire WL. Prediction of relapse or survival in patients with node-negative breast cancer by DNA flow c.ytometry N Engl J Med 1989;320:627-633. 17. Couinaud C. Les enveloppes vasculobiliaries du foie ou capsule de Glisson: leur interet dans la chirurgie vesicularie. les resections hepatiques et l'abord du bile du foie Lvon Chir 1954;49:589
HA, Steiner PE. Primary carcinoma of the liver: a stud?; of 100 cases among 48900 necropsies. Cancer 1954;7:462503. Dressler LG, Seamer L, Owens MA, Clark GM, McCuire WL. Evaluation of a modeling system for S-phase estimation in breast cancer by flow cytometry. Cancer Res 1987;47:5294-5302. Coon JS, Landay AL, Weinstein RS. Flow cytometric analysis of paraffin embedded tumors: implications for diagnostic pathology. Hum Path01 1986;117:435-437. Kaplan EL, Meier P. Nonparametric estimation from incomplete observation. J Am Stat Assoc 1958;53:457-487. Nagasue N. Galizia G, Yukaya H, Kohno H, Chang YC, Hayashi T, Nakamura T. Better survival in women than in men after radical resection of hepatocellular carcinoma. Hepatogastroenterology 1989;36:379-383. Nagasue N , Yukaya H, Ogawa Y, Sasaki Y, Chang YC, Niimi K. Clinical experience with 118 hepatic resections for hepatocellular carcinoma. Surgery 1986;99:694-701. Nagasue N, Yukawa H, Ogawa Y, Sasak~Y, Chang YC, Niimi K. Second hepatic resection for recurrent hepatocelluar carcinoma. Br J Surg 1986;73:434-438. Gozzetti G, Mazziotti A, Cavallari A, Bellusci R, Bolondi L, Grigioni W, Bragaglia R, et al. Clinical experience with hepatic resection for hepatocellular carcinoma in patients with cirrhosis. Surg Gynecol Obstet 1988;166:503-510. Hsu HC, Sheu JC, Lin YH, Chen DS, Lee CS, Hwang LY, Beasley 1'. Prognostic histologic features of resected hepatocellular carcinoma in Taiwan: a comparison with resected large HCC. Cancer 1985;56:672-680. Fujimoto J. Okamoto E, Yamanaka N, Toyosaka A, Mitsunobu M. Flow cytometric DNA analysis of hepatocellular carcinoma. Cancer 1991;67:939-944, Ishizu H. Flow cytometric analysis of the nuclear DNA content of hepatocellular carcinoma. Jpn J Surg 1989;19:662-673. Sasalu Y, Imaoka S, Fujita M, Miyoshi Y, Ohigashi H, Ishikawa 0, Furukawa H, et al. Regional therapy in the management of intrahepatic recurrence after surgery for hepatoma. Ann Surg 1987:206:40-47. .4ckerman NB, Nallathambi MN, Patel KR, Panchacharam P, Williams SE. Second hepatoma developing 13 years after resection of first tumor. Arch Surg 1986;121:726-728.
18. Edmondson
19 20 21
22
23 24
85.
26.
27 28.
29
30
1371
,' ' GAR-YANGCHAU,'
CHE-CHUAN LOONG'AND WING-YU LUI'
'Division of General Surgery Department of Surgery. Veterans General Hospital, and "Graduate Institute of Clinical Medicine. National Yang-Ming Medical College Taipei, Taiwan I 1 2 1 7, Republic of China
A total of 238 patients who received curative hepatic resectionsduring the last 10 yr were observed to search for the risk factors linked to early tumor recurrence of human hepatocellular carcinoma after hepatectomy. The results revealed that tumor size, tumor appearance and DNA ploidy were the factors in predicting tumor recurrence after resection for hepatocellular carcinoma. Patients with a tumor size less than or equal to 5 cm or a tumor appearance of the solitary type had better disease-free survival than did those with a tumor size greater than 5 cm or a tumor appearance of multiple/daughter nodule types (p < 0.06). Although patients with pattern III (aneuploid with 2 2 GO/Gl peaks) hepatoma had fewer statistically significant differences (p = 0.19) than did those with pattern I (diploid) or pattern II (aneuploid with single GO/G1 peak) tumors in predicting tumor recurrence,they did have poorer results in terms of the overall survival rate (p < 0.05). We conclude that patients with hepatocellular carcinoma having the aforementioned risk factors should be observed 1992: 16:1367-137 1.1 closely. (HEPATOLOGY
HCC is one of the most malignant diseases in Asian countries, and it also is the leading cause of cancer death in men and the third leading cause in women in Taiwan (1).Although much effort has been devoted to curative treatment of HCC and the results have given us some encouragement for the short term ( 2 - 4 ~ ,long-term results are still unsatisfactory. It has been generally accepted that postoperative tumor recurrence is the main cause of poor prognosis in patients with resectable HCC (5-7).Many factors. including tumor size, cut margin and portal vein invasion, for example, have been studied to predict the risk of tumor recurrence after resection for HCC (8-10). Because some controversy exists regarding the factors just described, different
Received May 4, 1992; accepted July 21, 1992 This work was partly supported by a grant from the (:linical Research Centpr. Veterans General Hospital. Institute of Biomedical .Science. Academia Sinica. Republic of China. Address reprint requests to. Shyh-('huan Jwu.Divlsion of General Surgery Department of Surgery, Veterans Grnrral Hospital. Rhih-Pal. Taipei. Taiwan 11217, Republic of China 31/1/41673
mechanisms leading to early and late recurrence have been proposed (8-10).Previous studies have shown that measurement of DNA content is useful in correlating biological behavior with the prognosis of neoplasms t 11- 13 ) . It has been suggested that change of DNA ploidy in human HCC has no correlation with survival of patients with HCC after hepatectomy (14,15).However, different conclusions have been drawn in some series (16). Recently, the change of DNA ploidy in human hepatomas was grouped into three patterns using a newly classified system. The patterns included the following: pattern I is a diploid tumor; pattern I1 is an aneuploid tumor with a single GO/G1 peak;and pattern I11 is an aneuploid tumor with more than one GO/Gl peak. We found that patients with pattern I11 hepatoma have had a tendency for early tumor recurrence after hepatic resection. Therefore we are interested in searching for changes of DNA content in conjunction with other documented factors to identify patients who will have a high risk of early tumor recurrence after hepatic resection. MATERLALS AND METHODS Between August 1981 and August 1991,3,649 patients with documented primary HCC were admitted to Veterans General Hospital, Taipei, Taiwan, and 308 of them received laparotomy for resection of HCC. Excluding those with palliative resection and operative mortality, 238 patients receiving curative hepatic resection were retrospectively reviewed in correlation with tumor recurrence and various risk factors, such as the presence of HBsAg, cirrhosis, tumor size, tumor appearance, tumor grading, tumor ploidy and cut margin, for example. After curative resection for HCC, which was defined as total removal of gross tumors (4, 9), patients were observed every 2 to 4 mo with a-fetoprotein; abdominal ultrasonography, abdominal computerized tomography or both; and celiac angography if necessary. Diagnosis of tumor recurrence was made when at least two of those four modalities were interpreted as positive findings. The mean follow-up time was 55.9 mo. The presence of HBsAg and cirrhosis were from welldocumented records. The resected specimens were recorded in tumor size, tumor appearance and gross cut margin. The cut margin was defined as the shortest distance between the tumor edge and the cut surface of the resected specimens. Tumor size was measured as the maximal diameter of the biggest tumor.
1367
1368
JWO ET AL.
HEPATOLOGY
extrahepatic recurrence 14 ( 6 . 9 % )
‘a and i n l r a tic recurrence
(8.8%)
FIG.1. Distribution of postoperative status in 238 patients receiving curative resection for HCC.
The gross tumor appearance was categorized into three types, which included the following: solitary tumor; tumor with daughter nodules in the same sector or segment (Couinaud’s classification)(17);and multiple tumors in different sectors or segments but confined in the same lobe. The specimens were also fixed with 10%formalin and stained with hematoxylin and eosin for examination of their pathological change. The differentiation of tumor cells was graded into four degrees as described by Edmondson and Steiner (18). DNA content, analyzed with fresh, frozen tissue or paraffinembedded samples, was measured with a flow cytometer (Epics Profile; Coulter Electronics, Hialeah, FL) as previously described (19, 20). The ploidy status was defined as diploid (2N) when the GO/G1 peak was superimposed after the addition of human peripheral blood lymphocytes to the samples with a range from 1.9N to 2.1N; aneuploid was defined if the discrete GO/G1 peak or peaks were outside the diploid range (1.9N to 2.1N). The DNA ploidy status of human hepatomas could be classified into three patterns. These patterns were the following: pattern I included hepatomas with diploid DNA distribution; pattern I1 included hepatomas with aneuploid DNA content and with only one prominent GO/Gl peak, and pattern I11 included hepatomas with aneuploid DNA content and with more than one GO/G1 peak. The data were analyzed using a statistical package program (BMDP Statistical Software, Inc., Los Angeles, CA). We used the BMDP 4F univariate and 2L multivariate analysis program to analyze risk factors for tumor recurrence and the 1L program to calculate disease-free survival by the Kaplan and Meier method (21). We compared the difference between categories in each prognostic factor using the generalized Wilcoxon’s test and Savage (Mantel-Cox) test. Any p values less than 0.05 were considered to be statistically sigdicant.
RESULTS
Among 3,649patients with HCC, 308 received laparotomies for resection of HCC. The operability rate was 8.4%, and the curative resectability rate was 86.7% (267 of 308 patients). The operative mortality rate was 8.4% (26 of 308). Among 238 patients receiving curative resection, 122 (51.3%) had tumor recurrence, and, of these 122 patients, 87 (71.3%) had intrahepatic recurrence. Fourteen patients (11.5%) had extrahepatic recurrence ( 1patient with a brain tumor, 1 patient with a lung tumor, 1 patient with a bone tumor and 11 patients with tumors affecting more than two organs); 21 patients t 17.2%) had both intrahepatic and extrahepatic recurrence (12 patients with lung tumors, 7 patients with bone tumors and 2 patients with chest wall tumors) (Fig. 1). Risk factors such as sex, presence of HBsAg, cirrhosis, tumor size, tumor appearance, tumor grading, tumor ploidy and cut margin were investigated in correlation with tumor recurrence. The results are shown in Table 1. No statistically significant difference was seen between different distances of cut margin and different differentiations of the Edmondson-Steiner tumor grading. Tumor size, tumor appearance and sex were statistically significant risk factors for tumor recurrence both by univariate and multivariate analysis (p < 0.05). Patients with HCC less than or equal to 5 cm had higher disease-free survival rates than did those with HCC greater than 5 cm (p < 0.005)(Fig. 2). Patients with a tumor appearance of the solitary type had higher disease-free survival rates than did those with tumors of
1369
FACTORS LINKED TO RECURRENCE OF HCC
Vol. 16, No. 6, 1992
TABLE1. Risk factors for tumor recurrence after resection of HCC
Rislrfactora
No. of patients
Tumor recurrence No
Yes
Lost follow-up
216 22
87 12
115 7
14 3
p p
= =
0.001" 0.035*
58 150
26 56
25 88
7 6
p
=
0.338
73 153
33 61
36 83
4
p
=
0.662
141 62 8
64 22 1
64 38 6
13 2
p p
=
0.050" 0.017'
111 115
54 38
50 68
7 9
p p
=
*--A
T55 ( n = l l l )
*-*B
T>5(n=115)
pvalue
Sex
M F HBsAg Negative Positive Cirrhosis Absent Present Tumor appearance solitary Daughter nodule Multiple nodules Tumor size 4 5 em > 5 em Tumor grading I
II III
Iv Tumor ploidyd I
n
n1
=
1
14 45 34 2
5 18 14 0
8 23 17 1
1 4 3 1
p
92 18 47
36 6 26
51 11 21
5 1 0
p
89 53 26
39 21 12
45 27 13
5 5 1
p
= =
0.026" 0.036'' 0.809
I
20
40
60
80
1ZO(rnonlhs)
100
FIG. 2. Disease-free survival after curative resection for HCC according to different tumor size. A = tumor size I 5 em; B = tumor size > 5 cm. A statistically sigmficant difference was seen between A and B (p < 0.005).
(X)
I00
o-oA:
90
'-*B.
80
A-b
(n=l41) (n= 62) C. (n= 8 )
70
60 =
0.192
50 40
30
Surgical margin < 1 em 1to3cm > 3 cm
04 0
9
20
=
0.931
"statistieally significant in univariate analysis. *Statistidy significant in multivariate analysis. 'Graded by Edmondson and Steiner's classification (18). dDNA ploidy, patterns I, I1 and 111, as defined in "Materials and Methods."
10
04 0
I
20
40
60
80
100
lZO(mon1hr)
FIG. 3. Disease-free survival after curative resection for HCC according to variation in tumor appearance. A = solitary tumor; B = tumor with daughter nodules; C = tumor with multiple nodules in different sectors or segments. Statistically s@cant differences were seen between A and B (p < 0.03) and A and C (p < 0.002), but no significant difference was seen between B and C (p = 0.11).
Because recurrence of HCC is frequent and the main cause of poor prognosis after extirpation of HCC, many factors have been studied to predict the risk of tumor recurrence after hepatic resection. Previous results concerning the predominant ratio and higher recurrent rate of HCC in men were explained by the presence of the androgen-dependent hormone receptor in hepatoma cells (22). Because this predominance in male patients is characteristic in our hospital and causes a sampling bias in patient selection, it is hard to draw any conclusion about the relationship between sex and the recurrence of HCC. Yoshida et al. (9) considered that the cut margin was not linked to tumor recurrence in small HCC ( 5 4 cm), whether or not an adequate cut margin ( 2 1 cm) was there, because the opportunity for portal vein or capsule invasion was very low. On the other hand, a DISCUSSION correlation existed between an inadequate cut margin The early diagnosis of HCC by developed diagnostic ( < 1 cm) and tumor recurrence in large HCC ( > 4 cm) modalities and improved surgical techniques has in- because the opportunity for portal vein or invasion creased the operability and resectability of human HCC. increased (9). In our results, patients with HCC less
daughter nodules (p < 0.03) or tumors of multiple types (p c 0.002) (Fig. 3). Although a less statistically significant difference existed between patients with HCC of different DNA ploidy in terms of the disease-free survival rate (Fig. 41, patients with pattern I11 HCC had a greater tendency for early tumor recurrence after hepatic resection than did those with pattern I or pattern 11 HCC. The median disease-free survival of patients with HCC of DNA patterns I, I1 and I11 was 19 mo, 17 mo and 8 mo, respectively. In addition, patients with pattern I11 HCC had poorer results in terms of overall survival rates (p < 0.02) (Fig. 5). The results suggested that HCC with pattern I11 DNA ploidy status had more malignant tumor behavior than did HCC with pattern I or pattern I1 DNA ploidy.
1370
JWO ET AL.
HEPATOLOGY
:“t 70
60 .50
~-
40
-.
30 -20 --
:
10.-
+
04
20
40
60
80
I00
120 (months)
FIG. 5. Overall survival rate after curative resection for HCC according to different DNA ploidy (patterns I, I1 and 111, as described in “Materials and Methods”). I = diploid HCC; I1 = aneuploid HCC containing only one prominent GO/G1 peak; IIZ = aneuploid HCC containing two or more GO/G1 peaks. A statistically significant difference was seen between patterns I and I11 (p < 0.003), but no significant difference was seen between patterns I and I1 (p = 0.26) or patterns I1 and I11 (p = 0.27).
than or equal to 5 cm had a significantly higher disease-free survival rate than did those with HCC greater than 5 cm; however, no significant differences in various distances of the cut margin ( < 1 cm, 1 to 3 cm and > 3 cm) were noted. It was generally accepted that patients with solitary HCC had a better prognosis than did those with HCC having daughter or multiple nodules in terms of overall survival (2,23). In this study, patients with solitary HCC also had better disease-free survival than did those with daughter or multiple nodules. Regarding the increased opportunity for tumors to spread, the possibility of multicentric metachronous or unnoticed synchronous lesions beyond the resected area has previously been considered to be the cause of tumor recurrence in HCC with the tumor appearance of daughter/multiple types (24).Because tumor grading by the Edmondson Steiner classification was not considered to be a factor for predicting tumor recurrence after resection of HCC in many studies (25,26), tumor behavior was analyzed by the measurement of DNA content and cell-cycle analysis. DNA ploidy of HCC, previously categorized into diploidy and nondiploidy, was reported to have no correlation with overall survival in patients with HCC after hepatic resection (14,151. However, Fujimoto et al. (27) and Ishizu (28)reported that patients with aneuploid HCC had a poorer overall survival rate than patients with diploid HCC. In our study, although the differences among newly classified DNA ploidy patterns I, I1 and I11 were less statistically significant (p = 0.191, patients with pattern I11 HCC tended to have a lower disease-free survival rate and a shorter median disease-free survival (8mo) than did those with pattern I or I1 HCC. It has been proposed that early tumor recurrence (within 2 yr) and late recurrence (after 2 yr) after resection of HCC are probably caused by different mechanisms. Some researchers have proposed that distant spreading of tumor thrombi by way of the portal
vein, direct invasion into the surrounding boundary by the original lesion, a multicentric metachronous lesion or unnoticed synchronous tumors were the main causes of early recurrence (9, 10, 24,291,whereas late recurrence in the liver remnant was a new lesion primary in origin (8, 30). Because the median disease-free survival in patieqts with DNA ploidy pattern I11 HCC ( 8 mo) is much shorter than that of pattern I1 or pattern I (17 mo and 19 mo, respectively), we speculate that tumor recurrence in patients with pattern 1/11and pattern I11 might involve a different mechanism. Thus a new lesion primary in origin might obscure our result and make it less statistically significant. However, further study on the molecular level and analysis is being conducted. In conclusion, those patients with resectable HCC who have high-risk factors, a large tumor ( > 5 cm), a tumor with daughter or multiple nodules or HCC with pattern I11 ploidy need to be observed closely after hepatic resection to detect early tumor recurrence. We also believe that measuring DNA ploidy may provide a new treatment strategy for HCC after curative hepatic resection. Acknowledgments: We thank L.H. Wu and L.J. Tai for their excellent technical assistance. REFERENCES 1. National Health Administration, Taiwan Provincial Health Department and Taipei City Health Department, Republic of China. Health statistics. Tapei: National Health Administration 1979;Z: 34-137. 2. Lee CS, Sung JL, Hwang LY, Sheu JC, Chen DS, Lin TY, Beasley P. Surgical treatment of 109 patients with symptomatic and asymptomatic hepatocellular carcinoma. Surgery 1986;99: 481-490. 3. Wu MC, Zhang XH, Chen H. Hepatic resection for primary liver cancer: report of 400 cases. Chin Med J 1986;99:175-180. 4. Kanematsu T, Matsumata T, Takenaka K, Yoshida Y, Higashi H, Sugimachi K. Clinical management of recurrent hepatocellular carcinoma after primary resection. Br J Surg 1988;75:203-206.
Vol 16, No 6, 1992
F.4C'TOKS LINKED TO RE('L7RRENCE OF HCC
5. Nagao T, Inone S, Goto S, Kawano N. Hepatic resection for hepatocellular carcinoma: clinical features and long-term prognosis. Ann Surg 1987;205:33-40. 6. Nagao T, Inone S, Yoshimi F. Sodeyama M. Omori Y,Mizuta T. Kawano N, et al. Postoperat.ive recurrence of hepatocellular carcinoma. Ann Surg 1990;211:28-33. 7. Nagasue N, Yukaya H, Chang YC. Yamanoi A, Kohno H. Hayashi T, Nakamura T. Assessment of pattern and treatment of intrahepatic recurrence after resection of hepatocellular carcinoma. Surg Gynecol Obstet 1990;171:217-222. 8. Lin TY, Lee CS, Chen KM. Chen CC. Role of surgery In treatment of primary carcinoma of the liver- a 31-war experience. Br J Surg 1987;74:839-842. 9 . Yoshida Y, Kanematsu T. Matsumata ' l ~ ,Takenaka K.Sugimachi K. Surgical margin and recurrence after resection of hepatocellular carcinoma in patients with cirrhosis. Ann Surg 1989;209: 297-301. 10. Shirabe K, Kanematsu T, Matsumata T, Adachi E. Akazawa K. Sugimachi K. Factors linked to early recurrence of small hepatocellular carcinoma after hepatectorny: univariate and multivariate 199 1 ;14:802-8O5. analysis. HEPATOLOGY 11. Watson JV.Oncogenes, cancer and analyt.ical q t o l o p . Cytometry 1986;7:400-410. 12. Young GAR, Hedley DW, Rugg CA. Iland HJ. The prognostic significance of proliferative activity in poor histology nonHodglun's lymphoma: a flow c-ytometry study using archival material. Eur J Cancer Clin Oncol 198723:1497- 1504. 13. Ewers SB, Langstrom E, Baldetorp B, IGllander D. Flowcytometric DNA analysis in primary breast carcinomas and clinicopathologid correlations. C.ytometry 1984:5:408-419. 14. Ezaki T, Kanematsu T, Okamura T. Sonoda T, Sugimachi K. DNA analysis of hepatocellular carcinoma and clinicopatholog~implications. Cancer 1988;61:106-109. 15. Saeter G , Lee CZ, Schwarze PE. Ous S. Chen DS, Sung JI,. Seglen PO. Changes in ploidy distributions in human liver carcinogenesis. Cancer 1988;80:1480-1485. 16. Clark GM, Dressler LG, Owens MA. Pounds G , Oldaker '1'. McGuire WL. Prediction of relapse or survival in patients with node-negative breast cancer by DNA flow c.ytometry N Engl J Med 1989;320:627-633. 17. Couinaud C. Les enveloppes vasculobiliaries du foie ou capsule de Glisson: leur interet dans la chirurgie vesicularie. les resections hepatiques et l'abord du bile du foie Lvon Chir 1954;49:589
HA, Steiner PE. Primary carcinoma of the liver: a stud?; of 100 cases among 48900 necropsies. Cancer 1954;7:462503. Dressler LG, Seamer L, Owens MA, Clark GM, McCuire WL. Evaluation of a modeling system for S-phase estimation in breast cancer by flow cytometry. Cancer Res 1987;47:5294-5302. Coon JS, Landay AL, Weinstein RS. Flow cytometric analysis of paraffin embedded tumors: implications for diagnostic pathology. Hum Path01 1986;117:435-437. Kaplan EL, Meier P. Nonparametric estimation from incomplete observation. J Am Stat Assoc 1958;53:457-487. Nagasue N. Galizia G, Yukaya H, Kohno H, Chang YC, Hayashi T, Nakamura T. Better survival in women than in men after radical resection of hepatocellular carcinoma. Hepatogastroenterology 1989;36:379-383. Nagasue N , Yukaya H, Ogawa Y, Sasaki Y, Chang YC, Niimi K. Clinical experience with 118 hepatic resections for hepatocellular carcinoma. Surgery 1986;99:694-701. Nagasue N, Yukawa H, Ogawa Y, Sasak~Y, Chang YC, Niimi K. Second hepatic resection for recurrent hepatocelluar carcinoma. Br J Surg 1986;73:434-438. Gozzetti G, Mazziotti A, Cavallari A, Bellusci R, Bolondi L, Grigioni W, Bragaglia R, et al. Clinical experience with hepatic resection for hepatocellular carcinoma in patients with cirrhosis. Surg Gynecol Obstet 1988;166:503-510. Hsu HC, Sheu JC, Lin YH, Chen DS, Lee CS, Hwang LY, Beasley 1'. Prognostic histologic features of resected hepatocellular carcinoma in Taiwan: a comparison with resected large HCC. Cancer 1985;56:672-680. Fujimoto J. Okamoto E, Yamanaka N, Toyosaka A, Mitsunobu M. Flow cytometric DNA analysis of hepatocellular carcinoma. Cancer 1991;67:939-944, Ishizu H. Flow cytometric analysis of the nuclear DNA content of hepatocellular carcinoma. Jpn J Surg 1989;19:662-673. Sasalu Y, Imaoka S, Fujita M, Miyoshi Y, Ohigashi H, Ishikawa 0, Furukawa H, et al. Regional therapy in the management of intrahepatic recurrence after surgery for hepatoma. Ann Surg 1987:206:40-47. .4ckerman NB, Nallathambi MN, Patel KR, Panchacharam P, Williams SE. Second hepatoma developing 13 years after resection of first tumor. Arch Surg 1986;121:726-728.
18. Edmondson
19 20 21
22
23 24
85.
26.
27 28.
29
30
1371
