REVIEW

Diagnostic value of alpha-L-fucosidase for hepatocellularcarcinoma: a meta-analysis

Yuling Gan & Qiuzhen Liang & Xinghua Song

Received: 6 October 2013 /Accepted: 16 December 2013# International Society of Oncology and BioMarkers (ISOBM) 2014

Abstract Alpha-fetoprotein (AFP) is the primary marker fordetecting hepatocellular carcinoma (HCC) and has been usedwidely in the clinic, but AFP is a biomarker characterized bypoor sensitivity and specificity. Alpha-L-fucosidase (AFU) hasbeen proposed as a tumor marker for diagnosis of HCC inmany studies. However, conclusions of its diagnostic valueare inconsistent. The current review aimed to evaluate thediagnostic value of AFU for HCC. After systematic reviewof 12 related studies, sensitivity, specificity, and diagnosticodds ratio (DOR) were pooled using random-effect models.Summary receiver operating characteristic (sROC) curve anal-ysis was used to summarize the overall test performance. Thepooled sensitivity for AFU was 0.72 (95 % confidence inter-val (CI) 0.69–0.76), while the pooled specificity was 0.78(95 % CI 0.74–0.81). DOR was 10.26 (95 % CI 5.99–17.59), and the area under the curve (AUC) was 0.8125.AFU had great value for the diagnosis of HCC as a serummarker.

Keywords Alpha-L-fucosidase . Hepatocellular carcinoma .

Diagnosis .Meta-analysis

Introduction

Hepatocellular carcinoma (HCC) is the sixth most commoncancer and the third cause of cancer-related mortality world-wide [1]. The incidence of HCC is on the increase, and it is a

major threat to our modern life [2]. Epidemiological studieshave indicated that HCC are related to chronic hepatitis B virus(HBV) and chronic hepatitis C virus (HCV) infection [3]. Alarge portion of patients diagnosed with HCC are at an ad-vanced stage andwith underlying liver dysfunction [4]. Patientsare often correlated with a poor prognosis. This is why earlydetection of HCC is extremely important in improving thesurvival of patients. Tumor markers are potential screeningtools for early diagnosis of tumors [5]. Alpha-fetoprotein(AFP) is the primary marker for detecting HCC and is usedwidely in the clinic. However, AFP is amarker characterized bypoor sensitivity and specificity, particularly during the earlystages of HCC [6]. Many patients with HCC show no elevatedlevel of AFP, while some patients with benign hepatic diseasescould have an elevated level of AFP [7]. Discovery of potentialbiomarkers is important in early diagnosis of HCC in the at-riskpopulation and improves the survival of patients.

Alpha-L-fucosidase (AFU) is a sort of lysosomal enzymepresent in all mammalian cells and hydrolyzes sugars contain-ing L-fucose [8]. AFU has been proposed as a tumor markersince many studies reported that its activity increases obvi-ously in the serum of HCC patients compared with that in thepatients with benign hepatic diseases [8, 9]. Unfortunately, itsdiagnostic value has not been widely accepted because con-clusions are inconsistent and even conflicting. The currentreview aims to evaluate the diagnostic value of AFU com-pared with that of AFP for the diagnosis of HCC by synthe-sizing and analyzing the results from research papers.

Methods

Search strategy

Studies published in English were mainly searched in PubMedand Google Scholar. The following keywords were used:

Electronic supplementary material The online version of this article(doi:10.1007/s13277-013-1563-8) contains supplementary material,which is available to authorized users.

Y. Gan :Q. Liang :X. Song (*)Department of Surgery of Bone Tumour, The First AffiliatedHospital of Xinjiang Medical University, Xinjiang 830000,People’s Republic of Chinae-mail: songxinghua19@163.com

Tumor Biol.DOI 10.1007/s13277-013-1563-8

“hepatocellular carcinoma,” “HCC,” “alpha-L-fucosidase,” and“AFU”. Both free text and MeSH search for keywords wereapplied. The search strategy is shown in Online Resource 1.The search was performed without restriction on the year of thepublication, study design, or published status. Additional stud-ies were obtained by a manual search of the references ofrelevant reviews.

Inclusion and exclusion criteria

Criteria that the included studies had to meet

1. All patients in experimental group were diagnosed asHCC with gold standard assessment

2. Comparison of diagnostic value of AFU with AFP forHCC in the same patients with prospective or retrospec-tive design

3. Sensitivity and specificity of AFU and AFP could bepresent alone, or derived by calculating the raw data inarticles

4. Patients in the control group only had benign liver dis-eases or other malignant liver cancers

5. Serum was the only sample form

6. No treatment before collection of the samples

Reasons that exclusion of studies were based on

1. The experiment was on animals2. Data on sensitivity and specificity could not be obtained

in articles3. Samples came from tissues or other body fluids4. Control groups only had healthy persons or studies had no

control group5. Abstracts, letters, editorials and expert opinions, and re-

views without original data6. Patients received therapy before samples were taken

Study selection

All searched studies were reviewed independently by tworeviewers including the titles, the abstracts, and then the fulltexts for potentially eligible studies. Disagreements betweenthe reviewers were resolved by discussion. Different articleswith the same authors were checked carefully to avoid over-lapped population. The authors of articles would be contactedfor full texts if necessary.

119 articles identifiedthrough databasesearching.

2 additional articlesidentified throughother resources.

69 articles screenedafter duplicates removed.

20 potentially eligiblearticles for full-textassessment.

12 articles includedin the meta-analysis.

8 articles exclude:*the control group were only health (n=1);*the useful data of AFP cannot obtain (n=2):*not test AFP (n=3);*no control group (n=1);*no full-text (n=1).

49 articles excluded afterreading the title andabstract:*reviews;*not clearly study on AFU for diagnosis of HCC;*language other than English.

52 duplicate articlesexclude.

Fig. 1 Study selection flow

Tumor Biol.

Data extraction

To improve reliability of the present research, two authorsindependently extracted data from the included articles. Dis-crepancies were resolved by discussion. The following datawere extracted in each article: first author, year of publication,country, number of the patients in experimental group andcontrol group, study design, type of markers, age and sex ratioof HCC patients, cutoff values, measuring methods for AFU,and raw data (the number of true positive, false positive, falsenegative, and true negative subject).

Assessment of methodological quality

The quality of each study was assessed by two reviewers byusing the new assessment tool Quality Assessment of Diag-nostic Accuracy Studies 2 (QUADAS-2). QUADAS-2 wasdeveloped based on the QUADAS [10]. Compared with

QUADAS, the QUADAS-2 tool offers additional and im-proved features. This tool comprises four domains: patientselection, index test, reference standard, and flow and timing.

Table 1 Main characteristics of studies included in the meta-analysis

Study Country HCC/control Gender (F/M HCC) Average age (HCC) TNM

Bukofzer et al. [21] South Africa 72/64 NK NK NK

El-Houseini et al. [22] Egypt 44/20 10/34 55±1.5 8/28/8/0

El-Houseini [23] Egypt 50/50 11/37 63.7±8.3 NK

El-Tayeh et al. [24] Egypt 37/59 8/29 55 (41–70) 7/11/9/10

Fawzy Montaser et al. [25] Egypt 40/40 7/33 54.88±7.93 NK

Giardina et al. [26] Italy 21/76 6/15 59.76±10.4 NK

Hutchinson et al. [17] England 35/35 8/27 20–73 NK

Marotta et al. [28] Japan 19/30 NK NK NK

Shao [29] China 30/30 4/26 51.2 NK

Takahashi et al. [30] Japan 67/47 14/53 63.7±8.3 NK

Tangkijvanich et al. [31] Thailand 60/150 10/50 56.5 3/48/9/0

Zhu et al. [32] China 113/72 18/95 55.3 NK

Cut-off value Assay type TP FP FN NP

820 nmol/mL/h PNPF 54 19 18 45

213 nmol/mL/h PNPF 36 9 8 11

10 μmol/L/mina PNPF 35 7 15 43

347.4 nmol/mL/h PNPF 27 8 10 51

2.3005 μmol/L/min PNPF 36 1 4 39

443 nmol/mL/h PNPF 16 7 5 69

121 nkat/Lb PNPF 21 12 14 23

740 nmol/mL/h PNPF 16 3 3 27

40 U/Lc CNPF 20 12 10 18

616 nmol/mL/h PNPF 52 10 15 37

870 nmol/mL/h PNPF 49 44 11 106

636.5 nmol/mL/h CNPF 64 18 49 54

TP true positive, FP false positive, FN false negative, TN true negative, NK unknown, F/M female/male, PNPF PNPF end-point method, CNPFCNPFkinetic rate assay methodaMicromoles per liter per minute=60 nmol/mL/hbKatal=6×107 UcUnits per liter=1 μmol/L/min

Table 2 Main characteristics of studies excluded in the meta-analysis

Study Reasons for excluding

Chen et al. [13] Control group without related liverdiseases, all healthy people

Wang and Cao [14] Data of AFP not clearly reported

Ishizuka et al. [15] Not test AFP

Giardina et al. [16] No control group

Hutchinson et al. [17] Not test AFP

Zheng and Ren [18] Not test AFP

Yamamoto et al. [19] Not find the full text

Deugnier et al. [20] Excluded the healthy patients; the usefuldata of AFP cannot be obtained

Tumor Biol.

Each domain is assessed in terms of risk of bias, and the firstthree domains are also assessed in terms of concerns regardingapplicability. Signaling questions are included to help judgerisk of bias as “high” or “low” [11].

Data analysis

The statistical analysis was all performed by Meta-Disc 1.4analysis software [12]. Firstly, discussion about heterogeneity

is going on. Heterogeneity caused by the threshold effect wasanalyzed by spearman correlation. If the spearman correlationcoefficient is about 1 and P≤0.05, then threshold effect exists.Heterogeneity caused by the non-threshold effect was checkedby a test of inconsistency (I2). Value of I2>0 shows statisticalsignificance. If there is heterogeneity caused by non-thresholdeffect, meta-regression is used to analyze the sources. Avalueof P≤0.05 shows statistical significance. A summary receiveroperating characteristic (sROC) curve was drawn with Moseslinear model, and then AUC and Q value were calculated.Pooled sensitivities, specificities, and diagnostic odds ratio(DOR) were obtained using DerSimonian–Laird method.

Results

Research results

The detailed flow diagram of article selection is shown inFig. 1. A total of 70 articles were found after duplicationswere removed, of which 20 potentially eligible articles werefor full-text assessment. Based on the inclusion and exclusioncriteria, eight articles [13–20] were excluded, and the rest 12articles [21–32] were included in our meta-analysis. Thecharacteristics of the included articles are shown in Table 1,and those of the excluded articles are shown in Table 2. Intotal, there are 1,261 patients in the meta-analysis including588 patients with HCC as experimental group and 673 pa-tients without HCC as control group.

Quality of studies

The result of the QUADAS-2 assessment to the includedarticles is shown in Fig. 2. The quality was not satisfactory.As shown in Fig. 2, the major bias of the studies was focused

Fig. 2 Risk of bias and applicability concerns summary: review ofauthors’ judgments about each domain for each included article

Fig. 3 Forest plot of sensitivityfor AFU

Tumor Biol.

on “patient selection” and “index test”. In the domain ofpatient selection, only three studies used prospective design,of which two studies about the exclusion criteria of patientselection were unclear, and nine articles did not avoid case–control design. In the domain of index test, almost all studiesdid not use a blind method, and one study was unclear. In fivestudies, the threshold was not prespecified. As for the domainof “flow and timing,” the patients from four studies did notreceive the same reference standards.

Data analysis

Summary diagnostic accuracy of AFU and AFP for HCC

The forest plots of sensitivity and specificity for AFU andAFP in the diagnosis of HCC are shown in Figs. 3, 4, 5, and 6.

The sensitivity that these studies observed ranged from 0.57 to0.90 for AFU and from 0.34 to 0.88 for AFP, while thespecificity ranged from 0.55 to 0.91 and from 0.63 to 1.00,respectively. The DORs for AFU and AFP ranged from 2.88to 351 and from 4.06 to 116.28. The results of pooled sensi-tivity, specificity, and DOR for AFU and AFP were calculated(Table 3). We also obtained the sROC curves for AFU andAFP. The AUC is 0.8125, and the Q index is 0.7489 for AFU,while 0.8081 and 0.7429 for AFP, respectively.

Test of heterogeneity for AFU

Studying on the effects of heterogeneity is the key to know thepossible influencing factors on accurate estimates in statisticsand to evaluate whether merging different studies is appropri-ate. In diagnostic tests, threshold effect is the most important

Fig. 4 Forest plot of specificityfor AFU

Fig. 5 Forest plot of sensitivityfor AFP

Tumor Biol.

cause of heterogeneity. Threshold effect can be observed insROC figure. If plane scatter distribution presents typical“shoulder arm shape,” then threshold effect exists. In our anal-ysis, the sROC figure of AFU did not show a typical shoulderarm shape (Fig. 7). The Spearman correlation coefficient was−0.378, and the P value was 0.226. These indicated that therewas no threshold effect. I2 of sensitivity and specificity forAFU was 64.8 and 77.7 %, implicating that there is significantheterogeneity caused by non-threshold effect.

Meta-regression for heterogeneity

We explained this heterogeneity by exploring the study char-acteristics using meta-regression. Races, number of HCCpatients and controls, assay type, and cutoff value were ex-amined (Table 4). The P value of the assay type was 0.0206,indicating that the assay type might be the source of hetero-geneity. Subgroup analysis was performed for each quantita-tion method of AFU. We divided these articles into twocategories based on assay types: PNPF end-point methodand CNPF kinetic rate assay. When using the end-point meth-od, the AFU activity is determined by using p-nitrophenyl-a-L-fucoside (PNPF) as colorimetric substrate [33–35]. The kinet-ic rate assay method usually uses 2-chloro-4-nitrophenyl-a-L-fucopyranoside (CNPF) as a substrate and analyzes with asemiautomatic or full automatic biochemistry analyzer [14].The results are presented in Table 5.

Discussion

Hepatocellular carcinoma is a common malignancy which isalso the main reason of mortality in patients with chronic liverdiseases [36]. HCC generally develops following an orderlyprogression from hepatitis to cirrhosis to early cancer, and it can

be cured if detected before the development of vascular inva-sion [37]. Serum markers show many advantages comparedwith histopathology examination. Now, ultrasonography andserum AFP are used widely to detect early HCC in the clinic,but because of a low sensitivity, the clinic effect of AFP fordisease diagnosis is compromised. Therefore, there is an in-creased need for finding more sensitive and specific tumormarkers for HCC. AFU has been proposed as a serum markerfor HCC in many studies. Healthy and patients with benignliver diseases have low concentration of AFU, but the level ofAFU in patients with HCC is significantly higher than theirs.Compared with ultrasonography, AFU is an earlier marker fordetecting HCC. Before ultrasonographic imaging can depict thecase, AFU significantly rises 6–9 months [25].

In this study, the diagnostic value of AFU compared withAFP was explored. To avoid substantial confounding factors incomparisons, we only selected studies that measured AFU andAFP in the same patients. And for increasing accuracy, we justobtained data frompatients with chronic liver diseases excludingthose from healthy persons. The pooled sensitivity of AFU washigher than that of AFP (0.72 vs. 0.61). Meanwhile, we alsonoted lower specificity (0.78 vs. 0.90). AUC was an index toassess the discriminating ability of a marker [38]. The AUC ofAFU was slightly higher than that of AFP (0.8125 vs. 0.8081),indicating that AFU had a better diagnosing accuracy. There

Fig. 6 Forest plot of specificityfor AFP

Table 3 Summary of diagnostic accuracy of AFP and GPC-3 usingMeta-Disc 1.4

Summary AFU (95 % CI) AFP (95 % CI)

Sensitivity 0.72 (0.69–0.76) 0.61 (0.56–0.65)

Specificity 0.78 (0.74–0.81) 0.90 (0.87–0.92)

DOR 10.26 (5.99–17.59) 14.60 (7.90–27.00)

DOR diagnostic odds ratio, CI confidence interval

Tumor Biol.

were studies finding that a combination of these two markersmight improve all the previous parameters [22, 24, 25, 31, 32].Some studies even reported that the specificity of combinedassessment could reach to 100 %. However, not all studiesincluded did combine the assessment, so we did not analyzethis in our meta-analysis. Recently, an article reported that AFP-L3 % had a higher specificity than AFP [39], we thought that acombination of AFU with AFP-L3 % might improve the diag-nostic accuracy, but further studies are still needed.

Large heterogeneity was observed in the meta-analysis. Weremoved any article from the study, but no remarkable changewas observed. It indicated that the heterogeneity was notcaused by any individual study. Then we explained this het-erogeneity by exploring the study characteristics by usingmeta-regression. We found that the assay type contributed tothis heterogeneity. Then subgroup analysis was performedbased on detection methods. The value of I2 decreased re-markably, but heterogeneity still existed. Many studies lacked

information on design, conduct, the data of the number ofhepatic focal lesions, Child’s class, and size of hepatic focal.We could not estimate whether these caused bias.

Fawzy et al. [25] noticed that the levels of AFU droppedafter a successful intervention for HCC patients, which provedthat AFU could be used as a prognostic factor to follow up theefficacy of intervention for the treatment of HCC patients, butmore studies to support it are needed. Many studies did notpresent detailed data about the stage of HCC. Therefore, wecould not conclude the diagnostic accuracy of AFU for earlyHCC.

Conclusion

Our results showed that AFU had a potential value for thediagnosis of HCC as a serummarker with good sensitivity and

Fig. 7 sROC for AFU

Table 4 Results of various factors in meta-regression

Variables P value RDOR (95 % CI)

HCC/control 0.0768 0.32 (0.09–1.16)

No. of HCC and control 0.9680 1.00 (0.99–1.01)

Assay type 0.0223 0.25 (0.08–0.78)

Cutoff value 0.7630 1.00 (1.00–1.00)

Race 0.1729 0.52 (0.19–1.44)

RDOR relative diagnostic odds ratio, CI confidence interval

Table 5 Results of subgroup analysis

Sensitivity(95 % CI)

Specificity(95 % CI)

DOR (95 % CI)

Heterogeneity (I2) Heterogeneity (I2) Heterogeneity (I2)

PNFP 0.77 (0.73–0.81) 0.79 (0.75–0.82) 13.08 (7.38–23.18)

32.7 % 79.4 % 65.1 %

CNPF 0.59 (0.50–0.67) 0.71 (0.61–0.79) 3.64 (2.09–6.33)

0.2 % 55.1 % 0.0 %

PNPF PNPF end-point method, CNPFCNPF kinetic rate assay method,CI confidence interval, DOR diagnostic odds ratio

Tumor Biol.

considerable moderate specificity. AFU could be complemen-tary to AFP. The combination value of AFU and AFP stillneeds further research.

Conflict of interest None

References

1. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM.Estimates of worldwide burden of cancer in 2008: GLOBOCAN2008. Int J Cancer. 2010;127(12):2893–917.

2. Gish RG. Hepatocellular carcinoma: overcoming challenges indisease management. Clin Gastroenterol Hepatol. 2006;4(3):252–61.

3. El-Serag HB. Epidemiology of viral hepatitis and hepatocellularcarcinoma. Gastroenterology. 2012;142(6):1264–73.

4. Fattovich G, Stroffolini T, Zagni I, Donato F. Hepatocellular carci-noma in cirrhosis: incidence and risk factors. Gastroenterology.2004;127(5 \ 1):S35–50.

5. Bertino G, Ardiri A, Malaguarner M, et al. Hepatocellualar carcino-ma serum markers. Semin Oncol. 2012;39(4):410–33.

6. Masuzaki R, Karp SJ, Omata M. New serum markers of hepatocel-lular carcinoma. Semin Oncol. 2012;39(4):434–9.

7. Ba MC, Long H, Tang YQ, Cui SZ. GP73 expression and itssignificance in the diagnosis of hepatocellular carcinoma: a review.Int J Clin Exp Pathol. 2012;5(9):874–81.

8. Stefaniuk P, Cianciara J, Wiercinska-Drapalo A. Present and futurepossibilities for early diagnosis of hepatocellular carcinoma. World JGastroenterol. 2010;16(4):418–24.

9. Zhou L, Liu J, Luo F. Serum tumor markers for detection of hepato-cellular carcinoma. World J Gastroenterol. 2006;12(8):1175–81.

10. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. Thedevelopment of QUADAS: a tool for the quality assessment ofstudies of diagnostic accuracy included in systematic reviews.BMC Med Res Methodol. 2003;10:3–25.

11. Whiting PF, Rutjes AW, Westwood ME. QUADAS-2: a revised toolfor the quality assessment of diagnostic accuracy studies. Ann InternMed. 2011;155:529–36.

12. Zamora J, Abraira V, Muriel A, Khan K, Coomarasamy A. Meta-DiSc: a software for meta-analysis of test accuracy data. BMC MedRes Methodol. 2006;6:31.

13. Chen K, Zhang H, Zhang LN, et al. Value of circulating cell-freeDNA in diagnosis of hepatocelluar carcinoma.World J Gastroenterol.2013;19(20):3143–9.

14. Wang JJ, Cao EH. Rapid kinetic rate assay of the serum alpha-L-fucosidase in patients with hepatocellular carcinoma by using a novelsubstrate. Clin Chim Acta. 2004;347(1–2):103–9.

15. Ishizuka H, Nakayama T, Matsuoka S, et al. Prediction of the devel-opment of hepato-cellular-carcinoma in patients with liver cirrhosisby the serial determinations of serum alpha-L-fucosidase activity.Intern Med. 1999;38(12):927–31.

16. Giardina MG, Matarazzo M, Morante R, et al. Serum alpha-L-fucosidase activity and early detection of hepatocellular carcinoma:a prospective study of patients with cirrhosis. Cancer. 1998;83(12):2468–74.

17. Hutchinson WL, Du MQ, Johnson PJ, Williams R. Fucosyltransferases:differential plasma and tissue alterations inhepatocellular carci-noma and cirrhosis. Hepatol. 1991;13(4):683–8.

18. Zheng T, Ren ZY. Activities of serum enzymes in patients with viralhepatitis B, posthepatitic cirrhosis and hepatocellular carcinoma.Chin Med J (Engl). 1990;103(9):729–34.

19. YamamotoK,Yanagita T,WatanabeN, et al. The clinical significanceof serum alpha-L-fucosidase assay in hepatocellular carcinoma.Fukuoka Igaku Zasshi. 1988;79(6):463–8.

20. Deugnier Y, David V, Brissot P, et al. Serum alpha-L-fucosidase: anew marker for the diagnosis of primary hepatic carcinoma?Hepatology. 1984;4(5):889–92.

21. Bukofzer S, Stass PM, KewMC, et al. Alpha-L-fucosidase as a serummarker of hepatocellular carcinoma in southern African blacks. Br JCancer. 1989;59(3):417–20.

22. El-Houseini ME, Mohammed MS, Elshemey WM, et al. Enhanceddetection of hepatocellular carcinoma. Cancer Control. 2005;12(4):248–53.

23. El-Houseini ME. Serum alpha-L-fucosidase enzyme activity as amarker for hepatocellular carcinoma: comparison with AFU usingROC analysis. J Egyptian Nat Canc Inst. 2001;13:277–83.

24. El-Tayeh SF, Hussein TD, El-Houseini ME, et al. Serological bio-markers of hepatocellular carcinoma in Egyptian patients. DisMarkers. 2012;32(4):255–63.

25. FawzyMontaserM, Amin SakrM, Omar KM.Alpha-L-fucosidase asa tumour marker of hepatocellular carcinoma. Arab J Gastroenterol.2012;13(1):9–13.

26. Giardina MG, Matarazzo M, Varriale A, et al. Serum alpha-L-fucosidase. A useful marker in the diagnosis of hepatocellular carci-noma. Cancer. 1992;70(5):1044–8.

27. Hutchinson WL, Johnson PJ, Du MQ, et al. Serum and tissue alpha-L-fucosidase activity in the pre-clinical and clinical stages of hepa-tocellular carcinoma. Clin Sci (Lond). 1991;81(2):177–82.

28. Marotta F, Chui DH, Safran P, Zhang SC. Serum alpha-L-fucosidase.A more sensitive marker for hepatocellular carcinoma? Dig Dis Sci.1991;36(7):993–7.

29. Shao Y. Union examination of AFP, AFU, AFPL3 and γ-GT in earlydiagnosis of primary liver cancer. Academic Journal of Xi’anJiaotong University. 2009;21(3).

30. Takahashi H, Saibara T, Iwamura S. Serum a-L-fucosidase activityand tumor size in hepatocellular carcinoma. Hepatology. 1994;19(6):1414–7.

31. Tangkijvanich P, Tosukhowong P, Bunyongyod P, et al. Alpha-L-fucosidase as a serum marker of hepatocellular carcinoma in Thailand.Southeast Asian J Trop Med Public Health. 1999;30(1):110–4.

32. Zhu J, Jiang F, Ni HB, et al. Combined analysis of serum gamma-glutamyl transferase isoenzyme II, alpha-L-fucosidase and alpha-fetoprotein detected using a commercial kit in the diagnosis ofhepatocellular carcinoma. Exp Ther Med. 2013;5(1):89–94.

33. Troost J, van der Heijden MC, Staal GE. Characterisation of alpha-L-fucosidase from two different families with fucosidosis. Clin ChimActa. 1976;73(2):329–46.

34. Zielke K, Okada S, O’Brien JS. Fucosidosis: diagnosis by serumassay of a-Lfucosidase. J Lab Clin Med. 1972;79(1):164–9.

35. Van Hoof F, Hers HG. The abnormalities of lysosomal enzymes inmucopolysaccharidoses. Eur J Biochem. 1968;7(1):34–44.

36. Elshemey WM, Desouky OS, Mohammed MS. Characterization ofcirrhosis and hepatocellular carcinoma using low-angle x-ray scatter-ing signatures of serum. Phys Med Biol. 2003;48(17):N239–46.

37. Schwartz M, Roayaie S, Konstadoulakis M. Strategies for the man-agement of hepatocellular carcinoma. Nat Clin Pract Oncol.2007;4(7):424–32.

38. Jones CM, Athanasiou T. Summary receiver operating characteristiccurve analysis techniques in the evaluation of diagnostic tests. AnnThorac Surg. 2005;79(1):16–20.

39. Yi X, Yu S, Bao Y. Alpha-fetoprotein-L3 in hepatocellular carcino-ma: a meta-analysis. Clin Chim Acta. 2013;425C:212–20.

Tumor Biol.