Ascitic fluid analysis in hepatocellular carcinoma

zyxw zyxwvu 677 Ascitic Fluid Analysis in Hepatocellular Carcinoma zy Agosfino Colli, M.D., Massimo Cocciolo, M.D., Carlo Xiva, M.D., Loredana Marcassoli, Mariangela Pirola, Palma Di Gregorio, Biol.Sc.D., and Guglielmo Buccino, Bio1.Sc.D. zyxwvutsrqp Background. Ascites in patients with hepatocellular carcinoma (HCC) is a poorly characterized subgroup of malignancy-related ascites. Not only the underlying liver disease, but also the tumor growth and spread contributes to the ascites formation. The authors differentiated ascites in HCC from other types of ascites. Methods. The authors analyzed the ascitic fluid of 185 consecutive patients (89 liver cirrhosis, 33 HCC, 31 peritoneal carcinomatosis, 22 liver metastases, 10 spontaneous bacterial peritonitis). Resulfs. Each subgroup showed a typical pattern. Compared with the cirrhotic patients, those with HCC showed a higher frequency of positive cytologic findings (4 of 33 versus 0/89, P < 0.004), elevated fibronectin concentration (10/33 versus 8/89, P < 0.004), and elevated polymorphonuclear cell count (10/33 versus 5/89 P < 0.004). cated cirrhosis, on the renal f ~ n c t i o n However, .~ the growth and spread of HCC could, at least hypothetically, worsen fluid accumulation into the peritoneal cavity by increasing portal pressure with thrombosis or compression of intrahepatic portal branches or by obstructing lymph channels and infiltrating the peritoneum. Despite the large amount of information about ascitic fluid analysis and its diagnostic value, we have not found any study devoted to characterize the ascitic fluid in patients with HCC. Recently, some biochemical parameters (serumascites albumin gradient [SAAG], ascitic fibronectin, cholesterol, sialic acid) have been reported to yield a near-perfect discrimination between malignant and nonmalignant ascites, even better than cytologic examination.4-" However, in these studies, patients with HCC are not examined at all or are considered as a small and marginal subgroup. To differentiate ascites in HCC from other types of ascites, particularly from sterile uncomplicated ascites in cirrhosis of the liver, we analyzed, measuring these new parameters (SAAG, fibronectin, cholesterol, and sialic acid) a series of 185 consecutive ascites (33 from patients with HCC overimposed on liver cirrhosis, 89 sterile uncomplicated ascites in cirrhosis of the liver, 31 peritoneal carcinomatosis, 22 from patients with liver metastasis, and 10 spontaneous bacterial peritonitis). zyxwvutsrqp zyxwvu Conclusions. A significant number of patients with ascites and HCC patients showed signs of peritoneal infiltration with positive cytologic findings and increased concentration of fibronectin. Moreover, neutrocytic ascites without signs of superinfection is relatively common (30%). Cancer 1993; 72677-82. Key words: hepatocellular carcinoma, serum-ascites albumin gradient, spontaneous bacterial peritonitis, fibronectin, diagnostic paracentesis. Hepatocellular carcinoma (HCC) is a common complication of long-standing liver disease and, at least in Western Countries and in Japan, more than 80% of the cases are overimposed on a cirrhotic liver.',2 Ascites in patients with HCC is considered a sign of the severity of the underlying liver disease with portal hypertension and hypoalbuminemia. Its response to the treatment with diuretics depends, as in uncompliFrom Ospedale "C. Borella," Giussano (Milano), Italy. Address for reprints: Agostino Colli, M.D., Divisione di Medicina, Ospedale di Giussano, via Milano, 65, 20034 Giussano (Milano) Italy. Accepted for publication March 23, 1993. Patients zyxwv We examined 185 consecutive patients with ascites admitted to our medical division between January 1986 and September 1991. We classified them into five groups: Group 1 Eighty-nine patients had sterile uncomplicated cirrhotic ascites. Diagnosis of cirrhosis was based on clinical 678 zyxwvutsrqp zyxwvutsrq zyxwvuts zyxw CANCER August 1, 1993, Volume 72, No. 3 grounds and histologically confirmed in 5 1 of them. In every patient, the serum alpha-fetoprotein was lower than 10 ng/ml and the ultrasound (US) abdomen scan showed no evidence of malignancy. Microbiologic cultures of the fluid gave negative results. Group 2 filtration; and one with liver cirrhosis and abdominal non-Hodgkin lymphoma. Methods In every patient, a diagnostic paracentesis was performed within 24 hours from the admission to our hospital, and usually before the beginning of any treatment. Some patients were already in chronic treatment with diuretics, but without oliguria and evidence of fluid accumulation. None of the patients had received antibiotic treatment before investigation. Paracentesis was performed under aseptic conditions using a 20-gauge needle and aspirating at least 350 ml of fluid: 20 ml were immediately injected, at patient’s bedside, into blood culture bottles with aerobic and anaerobic culture media (”Liquoid” Brain Heart Infusion and Thioglycollate, Roche, Basel, Switzerland) and observed for an adequate period, up to a month. The ascitic fluid was also cultured for Mycobacterium tuberculosis in special media. Leukocyte and PMN counts were performed by nonautomated means. Cytologic examination was performed within 1hour on stained smears of the sediment of centrifuged 250 ml of fluid. Chemical analyses (glucose, protein, albumin, amylase, lactate dehydrogenase, triglyceride, cholesterol) were performed by the techniques usually applied for blood samples. For alpha-fetoprotein and carcinoembryonic antigen determination, an immunoenzymatic assay was used (IMx AFP and 1Mx CEA, Abbott GmbH Wiesbaden, Germany). The fibronectin concentration was determined by nephelometric assay with a specific immune serum (Behringwerke AG, Marburg, W. Germany) in 5 ml of ascitic fluid freshly collected in a plastic tube containing ethylenediamine tetraacetic acid. The sialic acid concentration was determined with enzymatic-colorimetric assay (Boehringer Mannheim, W. Germany) in 5 ml of ascitic fluid. Serum total protein and albumin concentrations were measured for determination of SAAG on a venous blood sample taken immediately before paracentesis. zyxwvu Thirty-three patients had ascites of 70 patients with HCC in liver cirrhosis. The diagnosis was based on histologic study of adequate specimens obtained by USguided biopsy or laparoscopy or at necropsy. Multiple tumor masses or a large single mass with a diameter > 5 cm were found in five patients. In all of the patients, the finding of ascites was concomitant with the diagnosis of HCC. Group 3 Thirty-one patients had peritoneal carcinomatosis: 7 with and 24 without liver metastases. Diagnosis was based on positive ascitic cytologic findings at paracentesis and surgical or laparoscopic findings of peritoneal infiltration in patients with ovarian (13), colonic (6), uterine (2), renal (2), pancreatic (l),breast (l), lymphoma (l),or unknown (5) malignancy. Group 4 Twenty-two patients had liver metastases, no evidence of liver cirrhosis, and negative ascitic cytologic results. Diagnosis was made by histologic or cytologic study on specimens obtained by US-guided biopsy or fine-needle aspiration of the liver metastases: 7 gastric carcinoma, 4 colonic carcinoma, 2 pancreatic carcinoma, 2 lung carcinoma, 2 ovarian carcinoma, 2 lymphoma, 1 breast carcinoma, 1 renal carcinoma, and 1 adenocarcinoma of unknown origin. Group 5 Ten patients had spontaneous bacterial peritonitis with positive bacteriologic culture of the ascites in absence of evident visceral contamination. In all of the patients, the polymorphonuclear cell (PMN) count was > 250/ mm3. Seven patients had cirrhosis of the liver, one had HCC in cirrhosis, one had peritoneal carcinomatosis from gastric carcinoma, and one had liver metastases from colonic carcinoma and negative cytologic results. To simplify the study, we excluded four patients with ascites: one with tubercular peritonitis; one with neoplastic infiltration of the pericardium and evidence of constrictive pericarditis; one with amyloidosis, congestive heart failure, nephrotic syndrome, and liver in- Statistical Analysis For each parameter, a comparison among groups was performed with analysis of variance and the StudentNeuman-Keuls test. We preselected the cutoff values for each parameter indicating complicated ascites according to the data of previous studies: particularly SAAG < 1.1 mg/dl, ascitic fibronectin > 75 mg/l, cholesterol > 50 mg/dl, sialic acid > 30 mg/dl, leukocyte count > 500/mm3, PMN > 250/mm3. We calculated and analyzed the frequency rates exceeding the cutoff zyxwvutsrqp zyxwvutsrq zyxwvutsrqpo zyxwvutsr zyxwvutsrqpon zyxwvutsrq zyxwvutsrq zyxwvuts Ascites in HCC/Colli et al. 679 Table 1. Results of Ascitic Fluid Analysis in 175 Patients With Ascites and Comparison Between Sterile Uncomplicated Cirrhotic Ascites and Malignancy-Related Ascites Subgroups Parameter Serum-ascites albumin gradient (g/dl) Fibronectin (mg/l) Sialic acid (mg/dl) Cholesterol (mg/dl) Leukocyte count (cells/mm3) Polymorphonuclear cell count (cells/mm3) Group 1 Group 2 (n = 89) Mean 2 S D (n = 33) Mean f SD P versus 1 (n = 31) Mean ? SD P versus 1 1.9 f 0.5 32.1 f 30.1 16.2 f 88.1 30.1 k 33.9 2.1 f 0.7 29.0 62.8 21.0 f 18.1 24.7 f 26.4 NS NS NS NS 0.8 f 0.7 204.2 f 93.0 59.0 f 18.4 93.4 f 26.2 < 0.01 < 0.01 < 0.01 < 0.01 1.5 f 0.8 84.9 -+ 74.4 34.7 2 137.6 68.2 f 44.3 NS < 0.01 < 0.01 NS 422 f 571 809 f 860 NS 3011 f 2294 < 0.01 1055 2 1595 NS 103 f 102 765 f 768 NS 1622 f 1066 < 0.01 55 -+ 39 NS * Group 3 Group 4 (n = 22) Mean f SD P versus 1 Group 1: sterile uncomplicated cirrhotic ascites; Group 2: hepatocellular carcinoma in liver cirrhosis; Group 3: peritoneal carcinomatosis; Group 4: liver metastases; SD: standard deviation. for each group of ascites with chi-square test, with Yates modification, and the Bonferroni adjustment where needed; a P value < 0.05 was considered significant.” We excluded from the comparisons the 10 patients with spontaneous bacterial peritonitis (SBP). ResuIts Results are expressed as mean k SD unless otherwise indicated. None of the parameters considered showed a significant difference between HCC and sterile uncomplicated cirrhotic ascites, comparing the mean values; however, the comparison of the rates of frequency of values exceeding the cutoffs showed a significant difference for fibronectin, cytologic findings, leukocyte count, and PMN count ( P < 0.001). In patients with other subtypes of malignancy-related ascites (groups 3 and 4), not only fibronectin but also sialic acid and cholesterol concentrations were higher than in group 1. SAAG was lower in patients with peritoneal carcinomatosis. Table 1 shows the mean value and SD in groups 1 to 4. In Table 2 the rates of SAAG, fibronectin, sialic acid, cholesterol, leukocyte count, PMN count exceeding the cutoff values and the results of cytologic examination in each group are shown. Table 3 shows the diagnostic value of fibronectin, leukocyte count, PMN count, and cytologic findings in the differential diagnosis between HCC and sterile uncomplicated cirrhotic ascites. Table 4 shows the results of ascitic fluid analysis in the 10 patients with SBP (group 5). Serum alpha-fetoprotein was elevated (> 200 ng/ ml) in 21 of 33 patients with HCC and always normal, by definition, in patients with cirrhosis and in those with other malignancies. Ascitic concentration was >15 ng/dl only in patients with elevated serum concentration; in all of the patients, the concentration in the ascitic fluid was lower than serum concentration. SAAG was lower than 1.1 g/dl in the exudative range, in 21 of 31 patients with peritoneal carcinomatosis, in 3 of 33 with HCC, in 7 of 22 with liver metastases, and in only 1 of 89 patients with cirrhosis. Concomitant liver metastases (seven) or liver cirrhosis (one) were demonstrated by US abdomen scan or biopsy or both in the patients with peritoneal carcinomatosis and SAAG > 1.1. Ascitic sialic acid was elevated in 30 of 31 patients with peritoneal carcinomatosis and in 10 of 22 patients with liver metastases (P < 0.004 versus group 1). Ascitic cholesterol was elevated in 16 of 89 cirrhotic and in all 31 peritoneal carcinomatosis. Fibronectin was elevated in 8 of 89 patients with cirrhosis, 10 of 33 HCC, 2 of 22 metastases, and in 30 of 3 1 peritoneal carcinomatosis. Leukocyte count and PMN count were elevated in all patients with the infected ascites (SBP)in which culture was always positive. However, an increase in PMN count with a negative culture was found in 24 of 31 patients with carcinomatosis and in 10 of 33 HCC, with a significant difference with group 1 (cirrhosis of the liver) and group 4 (liver metastases). Finally, cytologic study was positive in 4 of 33 patients with HCC, with no false-positive results in group 1. Positive cytologic findings were, in every case, confirmed by an adequate liver biopsy. Figure 1 shows the results of PMN count, cytologic examination, and fibronectin concentration in the patients with HCC (group 2). zyxwvuts 680 zyxwvutsrqp zyxwvutsrq zyxwvuts zyxwvuts zyxwvutsr zy CANCER August 1, 1993, Volume 7 2 , No. 3 Table 2. Frequency Rate of Patients With Parameters That Exceed the Cutoff Values and Comparison Between Sterile Uncomplicated Cirrhotic Ascites and Malignancy-Related Ascites Subgroups Parameter Serum-ascites albumin gradient (< 1.1g/dl) Fibronectin (> 75 mg/l) Sialic acid (> 30 mg/dl) Cholesterol (> 50 mg/dl) Leukocyte count (> 500/mm3) Polymorphonuclear cell count (> 250/mm3) Positive cytologic results Group 1 (YO) Group 2 (%) (n = 33) P versus I Group 3 (%) (n = 31) P versus 1 Group 4 (YO) (n = 22) 3 (9) 10 (30) 6 (18) 3 (9) 19 (58) NS < 0.04 NS NS < 0.04 21 (68) 30 (97) 30 (97) 31 (100) 31 (100) < 0.004 < 0.004 < 0.004 < 0.004 < 0.004 7 (32) 8 (37) 15 (68) 15 (68) 11 (50) 10 (30) 4 (12.1) < 0.004 < 0.04 24 (77) 31 (100) < 0.004 < 0.004 (n = 89) 1(1) 8 (9) 7 (8) 16 (18) 26 (29) 5 (6) 0 0 0 P versus 1 NS < 0.04 < 0.02 < 0.02 < 0.02 NS NS Group 1: sterile uncomplicated cirrhotic ascites; Group 2: hepatocellular carcinoma in liver cirrhosis; Group 3: peritoneal carcinomatosis; Group 4: liver metastases. Discussion Malignancy-related ascites is classified into five main subgroups on pathophysiologic basis: (1) peritoneal carcinomatosis without liver metastases; (2) peritoneal carcinomatosis with liver metastases; (3) massive liver metastases without peritoneal infiltration; (4)chylous ascites (rare); and (5) HCC. Every subgroup is expected to have a characteristic cytologic and biochemical pattern.13 To make it easy, if peritoneal infiltration cytologic results are positive and if liver metastases occur, SAAG is > 1.1g/dl. Our data fail to demonstrate a typical pattern in the subgroup of patients with HCC (Tables I and 2). The SAAG-that is, the oncotic gradient between serum and ascites correlated with portal hypertension (the hydrostatic counterbalancing pre~sure)'~-as expected, almost always produced elevated (30/33,91 YO)results. Cytologic findings were positive in four cases (12%), according to the autopsy findings of peritoneal infiltration in another series.I5 Moreover, ascitic fibronectin concentration was elevated in 10 of 33 patients (3 of which had positive cytologic findings). In a recent report, ascitic fibronectin was elevated in 3 of 16 patients with HCC without peritoneal metastases.' Fibronectin is an intracellular matrix glycoprotein released from malignant cells dissociating into the peritoneal cavity.l 6 It was found in elevated concentrations in peritoneal carcinomatosis (30/31, 97% in our series). However, two other markers of malignant implants in the peritoneum-cholesterol" and sialic acid"-showed a lower sensitivity (3/33, or 9%; and 6/33, or IS%, respectively) in patients with HCC. Alpha-fetoprotein concentration in ascites was elevated (> 15 ng/rn1)l3 in 12 of 33 (36%) patients with HCC but only in 50% of patients with elevated serum level (> 200 ng/ml: 21/33, 63%). Thus its measurement in the ascites seems not to yield additive information. Finally, we found a statistically significant increase in frequency of an elevated leukocyte count and PMN count in ascites of patients with HCC compared with sterile uncomplicated cirrhotic ascites. Increased leukocyte count is a nonspecific finding often related to diuretic therapy and concentration of ascitic fluid.17 On the contrary, increased PMN count (> 250/mm3) in ascitic fluid is regarded as the most accurate index of superinfection and dictate antibiotic therapy." Apart from 10 cases of SBP (7 in cirrhosis, 1 in HCC, 1 in massive metastasis, and 1in peritoneal carcinomatosis), Table 3. Sensitivity, Specificity, and Diagnostic Accuracy of Fibronectin, Leukocyte Count, and Polymorphonuclear Count in Differential Diagnosis Between Sterile Uncomplicated Cirrhotic Ascites and Ascites in Hepatocellular Carcinoma Group 1 Parameter Fibronectin (> 75 mg/l) Leukocyte count (> 500/mm3) Polymorphonuclear cell count (> 250/mm3) Positive cytologic findings (n = 89) Group 2 (n = 33) Sensitivity Specificity (YO) Diagnostic accuracy (YO) 8 26 10 19 30.3% 57.6% 91 70.8 74.6 75.3 5 0 10 4 30.3% 12.1% 94.4 100 77 85.3 Group 1: sterile uncomplicated cirrhotic ascites; Group 2: hepatocellular carcinoma in h e r cirrhosis. zyxwvutsrqpo zyxwvutsr zyxwv Ascites in HCC/Colli et al. 681 Table 4. Ascitic Fluid Analysis in 10 Patients With Spontaneous Bacterial Peritonitis Type of ascites Fibronectin Leukocyte count (ml/@ (cells/m3) Cirrhosis Cirrhosis Cirrhosis Cirrhosis Cirrhosis Cirrhosis Cirrhosis Hepatocellular carcinoma Peritoneal carcinomatosis Liver metastases 60 26 39 10 10 56 60 1100 1900 2700 6000 900 2800 26600 1000 1600 2160 5400 855 2576 26000 5 11500 11270 129 38 6000 300 5400 290 a 0 0 0 Escherichia coli E. coli E. coli Diplococcus pneumoniae E. coli Yersinia enferocolitica E. coli zyxwvu E. coli E. coli Proteus niirabilis zyxwvutsrq in which culture was positive, we found neutrocytic ascites only in 5 of 89 (6%) patients with cirrhosis and in 10 of 33 (30%)with HCC ( P < 0.004). None of these patients showed symptoms indicative of peritoneal infection at paracentesis and during follow-up. We performed insemination and microbiologic cultures acthus uncording to the recommended detected infection seems improbable. In another small series, ascitic leukocyte and PMN counts were significantly higher in patients with HCC (4/6 and 3/6, respectively) than in cirrhotic patients (O/2O).I3 Moreover, the reported prevalence of sterile neutrocytic ascites in cirrhosis is about 3% to 8%;18higher prevalence (28% with a cutoff value of 150 PMN/mm3) was reported in a study which gathered in a single group cirrhosis and HCC2’ In peritoneal carcinomatosis, too, we noticed an elevated leukocyte count in 31 of 31 pa- 1 Polymorphonuclear cell count (cells/mm3) Culture tients, and an elevated PMN count in 24 of 31 (P < 0.004, with respect to sterile uncomplicated cirrhotic ascites), thus we suggest that the increase in the leukocyte count, more specifically, in the PMN count in ascites may represent a phlogistic reaction to neoplastic implant in the peritoneum or on the liver surface. In our series and in another recent report,’ superinfection of ascitic fluid seems not to increase fibronectin ascitic concentration. In our series of 10 cases of SBP, the only patient with elevated fibronectin (> 75 mg/l) had peritoneal carcinomatosis (Table 3). On the contrary, other reports emphasized the lack of specificity of increased ascitic fibronectin concentration demonstrating high concentration in SBP2*and in tubercular p e r i t ~ n i t i s . ~ ~ We found only 1patient (from 190 patients with ascites admitted to our medical division in 5 years) with tubercular ascites and she had elevated ascitic fibronectin (230 mg/l). In summary, ascitic fluid in a significant fraction of patients with HCC differentiates from sterile uncomplicated cirrhotic ascites (Table 4 and Fig. 1) showing signs of peritoneal infiltration (positive cytologic findings), aseptic increase of PMN count, and elevated concentration of fibronectin, a glycoprotein released from malignant cells. Ascitic fibronectin also was elevated in 8 of 22 (36%) patients with massive liver metastases and negative cytologic findings, confirming a high sensitivity in detection of malignancy-related as cite^.'-^," We cannot exclude minimal peritoneal infiltration without neoplastic cells shedding and, thus, these ascites should have been correctly classified in the peritoneal carcinomatosis group by more invasive diagnostic procedures (laparoscopy or laparotomy) or at autopsy, in spite of negative cytologic results. In clinical practice, in case of ascites of unknown cause, paracentesis and ascitic fluid analysis is informative. A neutrocytic ascites (with sterile adequate cul- zyxwv zyxw zyxwv z 0 0 0 : m : so0 Rlp5 Figure 1. Polymorphonuclear cell (PMN) count and fibronectin concentration in 33 patients with hepatocellular carcinoma (HCC, group 2) with positive (0)and negative ( 0 )cytologic examination. 682 zyxwvutsrqpo zyxwvuts zyxwvut CANCER August 1, 1993, Volume 72, No. 3 tures) or elevated concentration of fibronectin or both suggests the diagnosis of malignancy-related ascites, particularly HCC or massive liver metastases if the SAAG is elevated; a positive cytologic result is diagnostic for peritoneal infiltration, with liver metastases or cirrhosis if SAAG elevated. References 10. Jungst D, Gerbes AL, Martin R, Paumgartner G. Value of ascitic lipids in the differentiation between cirrhotic and malignant ascites. Hepatology 1986; 6:239-43. 11, Colli A, Buccino G, Cocciolo M, Parravicini R, Mariani F, Scaltrini GC. Diagnostic accuracy of sialic acid in the diagnosis of malignant ascites. Cancer 1989; 63:912-6. 12. Zar 11-1.The binomial distribution. In: Zar JH.Biostatistical analysis. Englewood Cliffs, NJ: Prentice-Hall 1974:495-8. 13. Runyon BA, Hoefs JC, Morgan TR. Ascitic fluid analysis in malignancy-related ascites. Hepatology 1988; 8:1104-9. 14. Hoefs JC. Serum protein concentration and portal pressure determine the ascitic fluid protein concentration in patients with chronic liver disease. 1 Lab Cliii Med 1983; 102:260-73. 15. Yuki K, Hirohashi 5, Sakamoto M, Kanai T, Shimosato Y. Growth and spread of hepatocellular carcinoma. Cancer 1990; 66~2174-9. 16. Klaubert W, Hafter R, Gollwitzer R, Willmanns W, Graeff H. Klinische wertigkeit von fibrin(ogen)-reaktionsprodukten und fibronektin in der differentialdiagnose von aszites. Z Gastroenterol 1985; 23:458-62. 17. Hoefs JC. Increase in ascites white blood cell and protein concentration during diuresis in patients with chronic liver disease. HepatoLogy 1981; 1:249-54. 18. Hoefs JC. Diagnostic paracentesis: a potent clinical tool. Gastroenterology 1990; 98:230-6. 19. Runyon BA, Canawati HN, Akriviadis EA. Optimization of ascitic fluid culture technique. Gastroenterology 1988; 95:1351-5. 20. Runyon BA, Umland ET, Merlin T. Inoculation of blood culture bottles with ascitic fluid: improved detection of spontaneous bacterial peritonitis. Arch I n t e r n Med 1986; 147:73-6. 21. Pinzello G, Verdone R, Lojacono F, Ciambra M, Dardanoni G, Fiorentino G, et al. Is the ascitic fluid a reliable index in making the presumptive diagnosis of spontaneous bacterial peritonitis? Hepatology 1986; 6:244-7. 22. Runyon BA. Elevated ascitic fluid fibronectin concentration: a non-specific finding. I Hepatoi 1986; 3:219-22. 23. Villar M, Garcia-Bragado F, Vilardell M, Biosca M, Rodrigo MJ, Schwartz 5.Fibronectin concentration does not differentiate between malignant and nonmalignant ascites. Gastroenterology 1988; 44:556-7. zyxwv zyxwv zyxwvutsrqpon 1. Okuda K. Early recognition of hepatocellular carcinoma. Hepatology 1986; 61729-38. 2. Colombo M, De Franchis R, Del Ninno E, Sangiovanni A, De Fazio C, Tommasini M, et al. Hepatocellular carcinoma in Italian patients with cirrhosis. N Engl ] Med 1991; 325:675-80. 3. Mas A, Arroyo V, Rodes J, Bosch J. Ascites and renal failure in primary liver cell carcinoma. Br Med J 1975; 3:629-33. 4. Par6 P, Talbot J, Hoefs JC. Serum-ascites albumin concentration gradient: a physiologic approach to the differential diagnosis of ascites. Gastroenterology 1983; 85:240-44. 5 Rector WJ, Reynolds IB. Superiority of the serum-ascites albumin difference over the ascites total protein concentration in separation of "transudative" and "exudative" ascites. Am J Med 1984; 77:83-5. 6 Albillos A, Cuervas-Mons V, Mill5 I, Cant6 T, Montes J, Barrios C, et al. Ascitic fluid polymorphonuclear cell count and serum to ascites albumin gradient in the diagnosis of bacterial peritonitis. Gastroenterology 1990; 98:134-40. 7 Scholmerich J, Volk BA, Kottgen E, Ehlers S, Gerok W. Fibronectin concentration in ascites differentiates between malignant and nonmalignant ascites. Gastroenterology 1984; 87:1160-4. 8 Colli A, Buccino G, Cocciolo M, Parravicini R, Mariani F, Scaltrini GC. Diagnostic accuracy of fibronectin in the differential diagnosis of ascites. Cancer 1986; 58:2489-93. 9 Prieto M, Gbez-Lecho MJ, Hoyos M, Castell JV, Carrasco D, Berenguer J. Diagnosis of malignant ascites: comparison of ascitic fibronectin, cholesterol and serum-ascites albumin difference. Dig Dis Sci 1988; 33:833-8.