Вы находитесь на странице: 1из 6

Journal of Hepatology 34 (2001) 570575

www.elsevier.com/locate/jhep

Serum a-fetoprotein for diagnosis of hepatocellular carcinoma in


patients with chronic liver disease: inuence of HBsAg and
anti-HCV status
Franco Trevisani 1,*, Paola Emanuela D'Intino 1, Antonio Maria Morselli-Labate 2,
Giuseppe Mazzella 2, Esterita Accogli 2, Paolo Caraceni 1, Marco Domenicali 1,
Stefania De Notariis 1, Enrico Roda 2, Mauro Bernardi 1
1
Dipartimento di Medicina Interna, Cardioangiologia, Epatologia, University of Bologna, Via Massarenti, 9, 40138 Bologna, Italy
2
Dipartimento di Medicina Interna e Gastroenterologia, University of Bologna, Bologna, Italy

See Editorial, pages 603605

Background: It is not established whether virological status affects the efciency of a-fetoprotein (AFP) as a hepa-
tocellular carcinoma (HCC) marker among patients with chronic liver disease (CLD).
Methods: We enrolled in a case-control study 170 HCC and 170 CLD patients, matched for age, sex, CLD and
HBsAg/anti-HCV status. The AFP sensitivity, specicity, positive (PPV) and negative (NPV) predictive values were
calculated. PPV and NPV were evaluated for three additional HCC prevalences (5, 10, and 20%).
Results: The best discriminating AFP value was 16 ng/ml. A value of 20 ng/ml (above which investigations for HCC
are recommended) had equivalent sensitivity (60.0 vs. 62.4%) and specicity (90.6 vs. 89.4%). PPV of 20 ng/ml was
84.6% but decreased to 25.1% at 5% tumor prevalence. NPV was 69.4% and rose to 97.7% at 5% prevalence. In the
different groups of infected patients PPV ranged from 80.0 to 90.9%, falling to 17.434.5% at 5% prevalence. In non-
infected patients PPV was 100% at any HCC prevalence. NPV ranged from 59.0 to 73.0%, reaching 96.598.1% at 5%
prevalence.
Conclusions: In CLD patients, AFP monitoring misses many HCCs and inappropriately arouses suspicion of malig-
nancy in many patients. Its usefulness is barely affected by the infection responsible for CLD. An AFP elevation could
be more indicative of HCC in non-infected patients.
q 2001 European Association for the Study of the Liver. Published by Elsevier Science B.V. All rights reserved.
Keywords: a-Fetoprotein; Hepatocellular carcinoma; Chronic liver disease; Diagnosis; Hepatitis virus B; Hepatitis
virus C

1. Introduction lance programs based on serial abdominal ultrasonography


and serum a-fetoprotein (AFP) assay [3]. Although recent
Hepatocellular carcinoma (HCC) is one of the most evidence indicates that the fucosilated fraction of AFP may
common cancers in humans, mostly occurring in patients be a more useful marker than total AFP [47], the availability
with chronic liver disease (CLD) [1]. Its early detection of this assay is still conned to few laboratories. Moreover,
may be important, since effective treatments are now avail- des-gamma-carboxyprothrombin, another commercially
able for the management of non-advanced neoplasms [2]. In available marker of HCC, has a sensitivity highly dependent
high-risk patients, this goal is accomplished through surveil- on cancer size, being superior to AFP only in large HCCs [8].
Thus, AFP remains the oncomarker universally utilized for
Received 23 March 2000; received in revised form 12 September 2000; monitoring high-risk patients for HCC in clinical practice.
accepted 20 October 2000 The hepatitis B virus (HBV) surface antigen (HBsAg)
* Corresponding author. Tel.: 139-051-6364918; fax: 139-051-340877.
status may modulate the diagnostic accuracy of AFP. In
E-mail address: trevi@unibo.it (F. Trevisani).

0168-8278/01/$20.00 q 2001 European Association for the Study of the Liver. Published by Elsevier Science B.V. All rights reserved.
PII: S01 68-8278(00)0005 3-2
F. Trevisani et al. / Journal of Hepatology 34 (2001) 570575 571

endemic areas of HCC, AFP seems to be less useful in classication [24]. HCC gross pathology was evaluated according to
HBsAg positive than HBsAg negative patients [9], leading previous studies [20,25] as solitary, multinodular, diffuse (tumor mass
not clearly dened with an indistinct boundary) and massive. A more
to many false positive results in the former [10,11]. This detailed stratication was also used: single #3 cm, single .3 cm without
observation has not been tested in the Caucasian population, satellites (contiguous lesions much smaller in size), single .3 cm with
where several features of HCC may differ from those found satellites, and multinodular plus diffuse. In this classication massive
in endemic areas [12,13]. Other studies have shown that tumors were included among single HCCs of .3 cm with or without
HCCs occurring in hepatitis C virus (HCV)-infected satellites, as appropriate. The degree of cellular anaplasia was dened
according to Edmondson and Steiner [26]. The size of HCC was measured
patients are more often associated with AFP elevation by ultrasonography and in patients with multiple lesions the diameter of the
than those harboring in HBsAg carriers [14,15]. However, largest node was utilized for statistics. The presence of portal or caval vein
this nding has not been conrmed [1618]. Therefore, the thrombosis was dened according to the imaging technique work-up
impact of virological status on the diagnostic accuracy of including angiography in 64 patients. To detect distant metastases, patients
AFP in HCC detection still remains largely unsettled. underwent chest X-ray and abdominal ultrasonography. Bone scintigraphy
and CT scans of the chest and brain were made when clinically indicated.
This case-control study aimed to identify the best cut-off
value of serum AFP to discriminate CLD patients with and
without HCC, and to assess whether HBV and HCV infec- 2.1. Statistical analysis
tions can modulate the reliability of AFP.
The results were expressed as the mean ^ standard deviation (SD) or
median and range. The statistical evaluation was performed by means of
MannWhitney U-test, x 2-test (with the Yates correction for dichotomous
2. Patients and methods variables), McNemar test, Wilcoxon matched-pairs signed-ranks test and
hierarchical log-linear models [27].
Among patients with HCC superimposed on CLD seen from January The ROC curve and the corresponding area under the curve were calcu-
1993 to June 1996 in our institution, we retrospectively selected those in lated to provide the accuracy of serum AFP in distinguishing HCC and CLD
whom information on HBsAg, antibody to HCV (anti-HCV), AFP level and patients [28]. Non-parametric estimates of the area under the ROC curve
disease of the extratumoral liver was reported in the clinical record. Patients and the respective standard error were applied [29]. The best cut-off value
with liver disease due to genetic and autoimmune disorders, primary biliary was chosen as the value that maximized the likelihood ratio (LR) obtained
cirrhosis and sclerosing cholangitis were excluded. Two hundred and ten using the following formula: LR (probability of true positive1probabil-
cases fullled these criteria. Among them, we were able to match 170 cases ity of true negative)/(probability of false positive1probability of false nega-
(135 males and 35 females) with 170 controls with CLD seen during the tive) [30].
same period according to the following criteria: age (within 6 years), sex, Since positive (PPV) and negative (NPV) predictive values depend on
underlying CLD (cirrhosis/chronic hepatitis), HBsAg and HCV status. In the prevalence of the disease [31], these variables were evaluated not only
control patients, the presence of HCC was ruled out by ultrasonography and in our population but also considering three HCC prevalences (5, 10, and
also by excluding patients who developed HCC during the following 6 20%) closer to a clinical setting [22,23,3236].
months. In HCC patients, stepwise logistic regression analysis was performed to
The diagnosis of HCC was based on histologic or cytologic ndings in control simultaneously for age, sex, ALT, virological status, ChildPugh
128 patients, while it was conrmed by clinical and imaging data or class, cancer stage, vascular thrombosis and metastases in determining an
necropsy in the remainder. The diagnosis of cirrhosis was supported by elevated AFP level (.20 ng/ml). Categorical variables were considered as
biopsy in 132 patients, and was based on the presence of clinical and follows: HBsAg1/anti-HCV2, HBsAg2/anti-HCV1, HBsAg1/anti-
laboratory features of portal hypertension (the presence of esophageal HCV1, and HBsAg2/anti-HCV2; HCC stage: single #3 cm, single .3
varices and/or collateral circulation at endoscopy and ultrasonography) in cm without satellites, single .3 cm with satellites, and multinodular plus
the remainder. The diagnosis of chronic hepatitis or brosis was based on diffuse; vascular thrombosis: no/yes; metastases: no/yes. Edmondson's
liver histology. grade was not included in this model to prevent a great reduction in the
Liver function tests (serum bilirubin, alkaline phosphatase, g-glutamyl- sample size. In the subgroup of patients in whom Edmondson's grade was
transpeptidase (g-GT), alanine aminotranspherase (ALT) and albumin, and available, an additional analysis was performed including this variable
prothrombin activity (PT)) were determined using commercially available (dichotomized as grades III and IIIIV), together with those signicantly
kits. associated with an elevated AFP in the rst model.
HBV markers were tested by radioimmunoassay or enzyme-linked All statistical evaluations were performed running the SPSS/PC1 statis-
immunosorbent assay (Abbott Laboratories, Chicago, IL; Sorin Biomedica, tical package on a personal computer [37]. A two-tailed P value less than
Saluggia, Italy). Anti-HCV was detected by ELISA II (Ortho Diagnostic 0.05 was considered statistically signicant.
Systems, Raritan, NJ) (until November 1993) and III (Ortho Diagnostic
Systems, Neckargemund, Germany).
AFP was measured by conventional assays (radioimmunoassay, Eiken
Chemical Co., Tokyo, Japan; LA-AFP test, Poli, Milan, Italy; immunoen- 3. Results
zymatic assay, Abbott Laboratories, Rome, Italy). The analysis of the clin-
ical utility of AFP was conducted using the following cut-off values: The mean age was 60.2 ^ 8.9 years in HCC and
60.0 ^ 9.2 years in control patients.
the best discriminating value provided by the receiver-operating char-
acteristic (ROC) curve, HCC was solitary in 84 patients, multinodular in 63,
the value of 20 ng/ml, above which investigations for HCC are currently diffuse in 18 and massive in 5. HCC was single #3 cm in
recommended [19,20], 23 patients, single .3 cm without satellites in 48, and single
the values of 100, 200 and 400 ng/ml these two latter values are .3 cm with satellites in 18. Edmondson's grade was speci-
currently utilized as conrmatory tests for HCC diagnosis in the
ed in 91 cases (3 grade I, 37 grade II, 42 grade III, and 9
presence of focal solid lesions of the liver [8,2123].
grade IV). Information on portal or caval thrombosis was
The severity of liver cirrhosis was assessed according to ChildPugh available in 149 HCC patients. Thrombosis was present in
572 F. Trevisani et al. / Journal of Hepatology 34 (2001) 570575

19 cases, 12 of them showing a multinodular or diffuse


HCC. Metastases were detected in 6 patients.
HCC was associated with a non-cirrhotic CLD (chronic
hepatitis or brosis) in 13 patients (7.6%) and with cirrhosis
in 157 patients (92.4%).
Twenty-seven HCC patients were HBsAg1/anti-HCV2,
103 were HBsAg2/anti-HCV1, 17 were HBsAg1/anti-
HCV1, and 23 were HBsAg2/anti-HCV2.
Information on alcohol intake was available in 157 HCC
and 128 control patients. Forty-eight HCC patients (30.6%)
and 38 control patients (29.7%) were heavy drinkers (.80
g/day of alcohol for at least 5 years) (P 0:238). Among
seronegative cases (HBsAg2/anti-HCV2), liver disease
was due to alcohol abuse in 13 HCC and 12 control patients,
while it was considered cryptogenic in the remainders.
The ChildPugh class was available in 150 HCC and 155
control individuals. An advanced cirrhosis was more
frequent in the latter (class A: 46.0 vs. 32.3%; class B:
40.0 vs. 33.5%; class C: 14.0 vs. 34.2%; P , 0:001). Plasma
ALT (76.1 ^ 57.2 vs. 70.7 ^ 55.1 IU/l, P 0:262), alka-
line phosphatase (246.2 ^ 132.0 vs. 287.7 ^ 561.7 IU/l,
Fig. 1. Receiver-operating characteristic (ROC) curve of serum AFP to
P 0:926), and PT (71.2 ^ 15.7 vs. 63.0 ^ 18.3, discriminate between patients with HCC and those with CLD. The area
P 0:051) did not signicantly differ. Conversely, HCC under the ROC curve was 0.819 ^ 0.023. The gure reports the best
patients showed higher levels of g-GT (99.0 ^ 117.3 vs. discriminating value between patients with or without HCC found in
79.5 ^ 100.5 IU/l, P 0:032) and albumin (3.54 ^ 0.49 our population (16 ng/ml), the limit above which investigations for this
vs. 3.37 ^ 0.65 g/dl, P , 0:001), and a lower bilirubin cancer are currently recommended (20 ng/ml), and three abnormal
values (100, 200, and 400 ng/ml).
(2.19 ^ 2.84 vs. 3.39 ^ 5.87 mg/dl, P 0:028).

3.1. AFP levels HCC prevalences. When the cut-off value of 20 ng/ml was
considered, the decline in cancer frequency had an impress-
Serum AFP was signicantly higher in HCC than control ive impact on the PPV, which fell to an estimated value of
patients (median 41 ng/ml (range 2100 000 ng/ml) vs. 6 25.1% at the lowest prevalence. The best PPV at the lowest
ng/ml (range 1450 ng/ml), P , 0:001). In the two groups, HCC prevalence was observed by using 200 ng/ml, which
AFP was .20 ng/ml in 60.0 and 9.4% of cases (P , 0:001), provided a correct allocation of two-thirds of cases. Each
respectively. cut-off value reached a good NPV (.90%) when a HCC
Fig. 1 depicts the ROC curve of AFP in the whole popu- prevalence of 10% was considered.
lation. The best discriminating value was 16 ng/ml, but both
sensitivity and specicity of 20 ng/ml were almost equiva- 3.2. Virological status and AFP
lent (Table 1). Therefore, 20 ng/ml was used as the best cut-
In each group, an elevated AFP (.20 ng/ml) was more
off for the analyses performed in patient subgroups. The
frequent in HCC than in control patients (HBsAg1/anti-
sensitivity of the other cut-off values was exceedingly low.
HCV2: 59.3 vs. 14.8%, P 0:004; HBsAg2/anti-
In HCC patients, ChildPugh class did not affect the
HCV1: 67.0 vs. 10.7%, P , 0:001; HBsAg1/anti-
probability of showing an elevated AFP (class A: 59.4%;
HCV1: 58.8 vs. 5.9%, P 0:004; HBsAg2/anti-HCV2:
class B: 61.7%; class C: 57.1%; P 0:995). The same was
30.4 vs. 0%, P 0:016).
true for the gross pathology of cancer (solitary: 52.4%;
multinodular: 69.8%; diffuse: 61.1%; massive: 60.0%; Table 1
P 0:205; single #3 cm: 52.2%; single .3 cm without Sensibility and specicity of ve serum levels of AFP for the diagnosis
satellites: 54.2%; single .3 cm with satellites: 50.0%; of HCC in the population under study a
multinodular plus diffuse: 67.9%; P 0:333) and Edmond-
AFP cut-off (ng/ml) Sensitivity (%) Specicity (%)
son's grade (grade III: 57.5%; grade IIIIV: 64.7%;
P 0:628). Finally, no correlation was found between 16 62.4 89.4
HCC diameter, available in 136 patients, and AFP elevation 20 60.0 90.6
(P 0:427). 100 31.2 98.8
200 22.4 99.4
Table 2 reports the PPV and NPV of the AFP cut-offs 400 17.1 99.4
obtained in our population, where the tumor prevalence was
a
50%, and those that were calculated for three additional AFP, a-fetoprotein; HCC, hepatocellular carcinoma.
F. Trevisani et al. / Journal of Hepatology 34 (2001) 570575 573

Table 2 Table 4
PPV and NPV for the diagnosis of HCC of four serum levels of AFP PPV and NPV of serum AFP (20 ng/ml) for the diagnosis of HCC
calculated for our population (where the cancer prevalence was 50%) according to the virological status of patients a
and for three additional tumor prevalences a
Patients HCC prevalence (%) PPV (%) NPV (%)
AFP cut-off (ng/ml) HCC prevalence (%) PPV (%) NPV (%)
HBsAg1/anti-HCV2 50 80.0 67.6
20 50 84.6 69.4 20 50.0 89.3
20 61.4 90.1 10 30.8 95.0
10 41.5 95.3 5 17.4 97.5
5 25.1 97.7 HBsAg2/anti-HCV1 50 86.3 73.0
100 50 96.4 58.9 20 61.1 91.5
20 86.9 85.2 10 41.1 96.1
10 74.6 92.8 5 24.8 98.1
5 58.2 96.5 HBsAg1/anti-HCV1 50 90.9 69.6
200 50 97.4 56.1 20 71.4 90.1
20 90.5 83.7 10 52.6 95.4
10 80.9 92.0 5 34.5 97.7
5 66.7 96.1 HBsAg2/anti-HCV2 50 100.0 59.0
400 50 96.7 54.5 20 100.0 85.2
20 87.9 82.7 10 100.0 92.8
10 76.3 91.5 5 100.0 96.5
5 60.4 95.8 a
They were calculated for our population (cancer prevalence: 50%) and
a
PPV, positive predictive value; NPV, negative predictive value; HCC, for three additional tumor prevalences. PPV, positive predictive value;
hepatocellular carcinoma; AFP, a-fetoprotein. NPV, negative predictive value; AFP, a-fetoprotein; HCC, hepatocellular
carcinoma; HBsAg, hepatitis B surface antigen; anti-HCV, antibody to
hepatitis C virus.
Table 3 reports the sensitivity and specicity of AFP of
.20 ng/ml according to the virological status. The sensitiv- analysis, the model retained 139 HCC patients. In this
ity was higher in HBsAg2/anti-HCV1 patients model, only the ALT value was an independent predictor
(P 0:025) and lower in seronegative individuals of an elevated AFP level (P 0:001). In the additional
(P 0:011) as compared to the overall cohort. Conversely, model including ALT and Edmondson's grade (84 patients),
the specicity did not signicantly differ between the ALT remained the only independent predictor of abnormal
groups. AFP levels (P 0:007).
Table 4 shows the PPV and NPV of AFP of .20 ng/ml
evaluated at different tumor prevalences. In all three groups
of infected patients, PPV markedly decreased with the 4. Discussion
decline of tumor prevalence. On the contrary, in the group
of seronegative patients PPV did not change according to Serum AFP is the most widely used oncomarker for
cancer prevalence due to the lack of false positive cases. A suspecting HCC. In our population, AFP showed an accu-
fairly elevated NPV was seen in every group at each HCC racy of 82% and according to previous studies [38] 16 ng/ml
prevalence. was the best cut-off to identify HCC superimposed to CLD.
Nonetheless, we adopted 20 ng/ml as the best cut-off since
3.3. Stepwise logistic regression analysis this value had an equivalent sensitivity and specicity, and
it is currently considered the limit above which investiga-
When age, sex, ALT, virological status, ChildPugh tions for HCC are needed [19,20].
class, cancer stage, vascular thrombosis and metastases This cut-off had a rather good specicity but a low sensi-
were simultaneously checked in the logistic regression tivity. However, attention should be paid to predictive
values which help the physician in clinical practice more
Table 3
Sensitivity and specicity of serum AFP of .20 ng/ml for the diagnosis than the specicity and sensitivity. In our population, the
of HCC according to the virological status of patients a NPV of AFP (indicating the probability that an individual
with a normal serum AFP is free from HCC) was unsatis-
Patients (n) Sensitivity (%) Specicity (%)
factory. It should be pointed out that predictive values are
HBsAg1/anti-HCV2 (27) 59.3 85.2 critically inuenced by the prevalence of the disease [31]
HBsAg2/anti-HCV1 (103) 67.0* 89.3 and in our case-control study the HCC prevalence was much
HBsAg1/anti-HCV1 (17) 58.8 94.1 greater (50%) than that expected in a clinical setting. We
HBsAg2/anti-HCV2 (23) 30.4** 100.0 therefore calculated the predictive values for cancer
a
AFP, a-fetoprotein; HCC, hepatocellular carcinoma; HBsAg, hepatitis frequencies matching those found in clinical practice
B surface antigen; anti-HCV, antibody to hepatitis C virus. *P 0:025 and [22,23,3236]. With a HCC prevalence of 5% the NPV
**P 0:011 vs. the overall sensitivity (hierarchical log-linear models). became very high (about 98%), suggesting that most
574 F. Trevisani et al. / Journal of Hepatology 34 (2001) 570575

patients with normal AFP are correctly allocated to the P , 0:001). Since ALT levels are surrogate markers of
group without cancer. Unfortunately, the clinical usefulness CLD activity, which can stimulate the `non-tumoral' secre-
of this prediction is diminished by the low sensitivity of tion of AFP [10,11,19,20], it is conceivable that a feeble
AFP. In fact, although false negative cases were a minimal extratumoral production contributed to abate the sensitivity
percentage of the patients with normal AFP, they accounted of AFP in seronegative patients. This may also explain the
for 40% of all HCC cases (Table 1). This conrms the poor absence of false positive cases, which rose to 100% both
efciency as a `screening test' demonstrated by AFP in specicity and PPV. Such gures would indicate that an
prospective studies [8,19,32,36,39]. AFP rise in non-viral CLD patients may be considered
Another important pitfall of this AFP cut-off was the highly indicative of HCC occurrence. However, owing to
unacceptably low PPV (that is, the probability that a patient the low number of seronegative individuals, our results need
with an abnormal AFP has HCC), which fell to 25% at the to be conrmed in a larger series.
lowest cancer prevalence. This gure is close to that The best sensitivity of AFP found in anti-HCV1/
reported in a prospective study on cirrhotic patients by HBsAg2 supports the association between HCV infection
using the cut-off value of 15 ng/ml [8]. Therefore, AFP and AFP elevation previously reported in HCC patients
monitoring alone in patients with CLD would lead to suspi- [14,15]. The greater propensity to produce AFP by these
cion of malignancy in a high proportion of individuals with- patients does not have an explanation. It cannot be attributed
out cancer, inappropriately eliciting stressful and expensive to a higher activity of the underlying CLD since ALT levels
investigations in three out of four cases with elevated AFP. did not differ among the three groups of infected patients
In this respect, 200 ng/ml was the best cut-off value, yield- (KruskalWallis test: P 0:293, data not shown).
ing a correct allocation into the neoplastic group of two- Lastly, the kind of viral infection did not greatly affect the
thirds of patients with an elevated AFP. However, the use predictive values at any HCC prevalence, and the PPV
of this cut-off value as a screening test cannot be recom- became unacceptably low in all infected patients when a
mended because almost 80% of HCC would be missed. Our clinical setting was simulated. We may conclude that
ndings strengthen the suggestion that the use of AFP as a among patients with viral CLD, the type of infection does
screening test for HCC in CLD patients should be aban- not inuence the clinical usefulness of AFP.
doned [19]. In conclusion, our study showed that in patients with CLD
Serum AFP is also utilized as a `conrmatory test' to (1) the AFP value of 20 ng/ml, above which investigations for
discriminate HCC from other solid lesions of the liver. HCC are currently recommended, has a discriminating
The levels of 200 [8,21] and 400 ng/ml [22,23] are power almost equivalent to the best cut-off (16 ng/ml), (2)
commonly utilized for this purpose. In our population, not the diagnostic efciency of AFP is poor, dampening its use as
only these values but also 100 ng/ml showed a very high a screening test for HCC detection in fact, AFP monitoring
specicity, indicating that even 100 ng/ml could be conr- alone allows many tumors to escape identication and
matory for HCC. However, this assumption needs to be arouses the suspicion of malignancy in too many patients
veried through studies in which the control group is without cancer, unnecessarily increasing medical costs and
formed by CLD patients with hepatic nodules other than patient anxiety, (3) the specicity of 100 ng/ml AFP is high,
HCC. making it advisable to assess the role of this cut-off value as a
Previous studies have suggested that the etiology of liver conrmatory test for HCC, (4) the predictive values, which
disease inuences the probability of nding an abnormal are the most helpful parameters in clinical practice, are barely
AFP in HCC patients, but they did not analyze the impact affected by the kind of viral infection responsible for the
of these ndings in HCC diagnosis [9,14,15,40]. Moreover, underlying liver disease, and (5) an elevated AFP could be
data denying the inuence of etiology in AFP utility have more indicative of HCC in non-infected than infected
also been produced [1618]. The distribution of the different patients.
etiologic groups found in our study was similar to that
reported by a recent large multicentric Italian study [15].
We found that the sensitivity of an elevated AFP (.20 References
ng/ml) reached the lowest value in patients without viral
infection, owing to the very low proportion (30%) of secret- [1] World Health Organization. 1994 World Health Statistics Annual.
ing HCCs. Similar ndings have been obtained in serone- Geneva: WHO, 1995.
[2] Bruix J. Treatment of hepatocellular carcinoma. Hepatology
gative non-Caucasian patients [14,16] and Italian alcoholic 1997;25:5962.
patients [40], who frequently have normal AFP at the time [3] Collier J, Sherman M. Screening for hepatocellular carcinoma. Hepa-
of HCC detection. A precise explanation for this phenom- tology 1998;27:273278.
enon is not available. It is worth noting, however, that the [4] Kuromatsu R, Tanaka M, Tanikawa K. Serum alpha-fetoprotein and
only independent predictor of an AFP elevation in our study lens culinaris agglutinin-reactive fraction of alpha-fetoprotein in
patients with hepatocellular carcinoma. Liver 1993;13:177182.
was the ALT level, and seronegative individuals showed [5] Sato Y, Nakata K, Kato Y, Shima M, Ishii N, Koji T, et al. Early
signicantly lower ALT values than the remaining groups recognition of hepatocellular carcinoma based on altered proles of
taken together (43.6 ^ 30.1 vs. 81.4 ^ 58.8 IU/l, alpha-fetoprotein. N Engl J Med 1993;328:18021806.
F. Trevisani et al. / Journal of Hepatology 34 (2001) 570575 575

[6] Takahashi H, Saibara T, Iwamura S, Tomita A, Maeda T, Onishi S, et Borzio F, et al. Liver cell dysplasia is a major risk factor for hepato-
al. Serum alpha-l-fucosidase activity and tumor size in hepatocellular cellular carcinoma in cirrhosis: a prospective study. Gastroenterology
carcinoma. Hepatology 1994;19:14141417. 1995;108:812817.
[7] Giardina MG, Matarazzo M, Morante R, Lucariello A, Varriale A, [24] Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R.
Guardasole V, et al. Serum a-l-fucosidase activity and early detec- Transection of the oesophagus for bleeding oesophageal varices. Br J
tion of hepatocellular carcinoma. A prospective study of patients with Surg 1973;60:646649.
cirrhosis. Cancer 1998;83:24682474. [25] Trevisani F, Caraceni P, Bernardi M, D'Intino PE, Arienti V, Amorati
[8] Pateron D, Ganne N, Trinchet JC, Aurousseau MH, Mal F, Meicler C, P, et al. Gross pathologic types of hepatocellular carcinoma in Italian
et al. Prospective study of screening for hepatocellular carcinoma in patients. Relationship with demographic, environmental and clinical
Caucasian patients with cirrhosis. J Hepatol 1994;20:6571. factors. Cancer 1993;72:15571563.
[9] Lee HS, Chung YH, Kim CY. Specicities of serum a-fetoprotein in [26] Edmondson HA, Steiner PE. Primary carcinoma of the liver. A study
HBsAg1 and HBsAg2 patients in the diagnosis of hepatocellular of 100 cases among 48,900 necropsies. Cancer 1954;7:462503.
carcinoma. Hepatology 1991;14:6872. [27] Siegel S. Non-parametric statistics for the behavioral sciences. New
[10] Liaw YF, Tai DI, Chen TJ, Chu CM, Huang MJ. Alpha-fetoprotein York: McGraw-Hill, 1956.
changes in the course of chronic hepatitis: relation to bridging hepatic [28] Zweig MH, Campbell G. Receiver-operating characteristic (ROC)
necrosis and hepatocellular carcinoma. Liver 1986;6:133137. plots: a fundamental evaluation tool in clinical medicine. Clin
[11] Chen DS, Sung JL. Relationship of hepatitis B surface antigen to Chem 1993;39:561577.
serum alpha-fetoprotein in nonmalignant diseases of the liver. Cancer [29] Hanley JA, McNeil BJ. The meaning and use of the area under a
1979;44:984992. receiver operating characteristic (ROC) curve. Radiology
[12] Okuda K, Peters RL, Simson IW. Gross anatomic features of hepato- 1982;143:2936.
cellular carcinoma from three disparate geographic areas. Proposal of [30] Pezzilli R, Morselli-Labate AM, Miniero R, Barakat B, Fiocchi M,
new classication. Cancer 1984;54:21652173. Cappelletti O. Simultaneous serum assays of lipase and interleukin-6
[13] Tiribelli C, Melato M, Croce LS, Giarelli L, Okuda K, Ohnishi K. for early diagnosis and prognosis of acute pancreatitis. Clin Chem
Prevalence of hepatocellular carcinoma and relation to cirrhosis:
1999;45:17621767.
comparison of two different cities of the world Trieste, Italy, and [31] Komaroff AL, Berwick DM. Decision theory and medical practice.
Chiba, Japan. Hepatology 1989;10:9981002.
In: Isselbacher KJ, Adams RD, Braunwald E, Pedersdorf RG, Wilson
[14] Tsai JF, Chang WY, Jeng JE, Ho MS, Lin ZY, Tsai JH. Frequency of
JD, editors. Principles of internal medicine. Update IV, New York:
raised a-fetoprotein level among Chinese patients with hepatocellular
McGraw-Hill, 1983. pp. 243254.
carcinoma related to hepatitis B and C. Br J Cancer 1994;69:1157
[32] Cottone M, Turri M, Caltagirone M, Maringhini A, Sciarrino E,
1159.
Virdone R, et al. Early detection of hepatocellular carcinoma asso-
[15] Stroffolini T, Andreone P, Andriulli A, Ascione A, CraxI A, Chiar-
ciated with cirrhosis by ultrasound and alfafetoprotein: a prospective
amonte M, et al. Characteristics of hepatocellular carcinoma in Italy. J
study. Hepatogastroenterology 1988;35:101103.
Hepatol 1998;29:944952.
[33] Imberti D, Fornari F, Sbolli G, Buscarini E, Squassante L, Buscarini
[16] Kew MC, Houghton M, Choo QL, Kuo G. Hepatitis C virus antibo-
L. Hepatocellular carcinoma in liver cirrhosis. A prospective study.
dies in southern African blacks with hepatocellular carcinoma. Lancet
1990;335:873874. Scand J Gastroenterol 1993;28:540544.
[17] Saitoh S, Ikeda K, Koida I, Tsubota A, Arase Y, Chayama K, et al. [34] Propst T, Propst A, Dietze O, Judmaier G, Braunsteiner H, Vogel W.
Serum des-gamma-carboxyprothrombin concentration determined by Prevalence of hepatocellular carcinoma in alpha-1-antitrypsin de-
avidin-biotin complex method in small hepatocellular carcinomas. ciency. J Hepatol 1994;21:10061011.
Cancer 1994;74:29182923. [35] Curley SA, Izzo F, Gallipoli A, de Bellis M, Cremona F, Parisi V.
[18] Hwang SJ, Tong MJ, Lai PPC, Ko ES, Co RL, Chien D, et al. Evalua- Identication and screening of 416 patients with chronic hepatitis at
tion of hepatitis B and C viral markers: clinical signicance in Asian high risk to develop hepatocellular carcinoma. Ann Surg
and Caucasian patients with hepatocellular carcinoma in the United 1995;222:375383.
States of America. J Gastroenterol Hepatol 1996;11:949954. [36] Zoli M, Malagotti D, Bianchi G, Gueli C, Marchesini G, Pisi E.
[19] Sherman M, Peltekian KM, Lee C. Screening for hepatocellular carci- Efcacy of surveillance program for early detection of hepatocellular
noma in chronic carriers of hepatitis B virus: incidence and preva- carcinoma. Cancer 1996;78:977985.
lence of hepatocellular carcinoma in a north American urban [37] Norusis MJ. SPSS Inc. SPSS/PC1. Base system and advanced statis-
population. Hepatology 1995;22:432438. tics, version 5.0. Chicago, IL: SPSS, 1992.
[20] Trevisani F, D'Intino PE, Caraceni P, Pizzo M, Stefanini GF, [38] Bayati N, Silverman AL, Gordon ST. Serum alpha-fetoprotein levels
Mazziotti A, et al. Etiologic factors and clinical presentation of hepa- and liver histology in patients with chronic hepatitis C. Am J Gastro-
tocellular carcinoma. Difference between cirrhotic and noncirrhotic enterol 1998;93:24522456.
Italian patients. Cancer 1995;75:22202232. [39] Shinagawa T, Ohto M, Kimura K, Tsunetomi S, Morita M, Saisho H,
[21] Associazione Italiana per lo Studio del Fegato. Epatocarcinoma. et al. Diagnosis and clinical features of small hepatocellular carci-
Linee guida per la diagnosi e la terapia. Bologna: Tipograa Moderna, noma with emphasis on the utility of real-time ultrasonography. A
1998. study in 51 patients. Gastroenterology 1984;86:495502.
[22] Colombo M, De Franchis R, Del Ninno E, Sangiovanni A, De Fazio [40] Fasani P, Sangiovanni A, De Fazio C, Borzio M, Bruno S, Rochi G, et
C, Tommasini M, et al. Hepatocellular carcinoma in Italian patients al. High prevalence of multinodular hepatocellular carcinoma in
with cirrhosis. N Engl J Med 1991;325:675680. patients with cirrhosis attributable to multiple risk factors. Hepatol-
[23] Borzio M, Bruno S, Roncalli M, Colloredo Mels G, Ramella G, ogy 1999;29:17041707.

Вам также может понравиться