• Users Online: 262
  • Home
  • Print this page
  • Email this page
Home About us Editorial board Ahead of print Current issue Search Archives Submit article Instructions Subscribe Contacts Login 


 
 Table of Contents  
ORIGINAL ARTICLE
Year : 2016  |  Volume : 14  |  Issue : 1  |  Page : 14-18

Noninvasive prediction of HCV-4 SVR by 2D US: a randomized study using data mining algorithm


1 Department of Tropical Medicine, Gastroenterology and Hepatology, Al Azhar University, Aswan, Egypt
2 Department of Immunology and Rheumatology, Aswan Hospital of Febrile, Diseases, Aswan, Egypt
3 Department of Gynecology and Obstetrics, Al Azhar University, Aswan, Egypt
4 Department of Internal Medicine, Al Azhar University, Aswan, Egypt
5 Department of General and Bariatric Surgery, Faculty of Medicine, Al Azhar University, Aswan, Egypt
6 Department of Biochemistry, Faculty of Pharmacy, Suez Canal University, Cairo, Egypt

Date of Submission10-Nov-2015
Date of Acceptance13-Dec-2015
Date of Web Publication18-Apr-2016

Correspondence Address:
Hamdy Mahfouz
Tropical Medicine, Department of Tropical Medicine, Gastroenterology and Hepatology, Faculty of Medicine, Al Azhar University, Cairo
Egypt
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/1687-1693.180454

Rights and Permissions
  Abstract 

Objective and aim
Hepatitis C virus (HCV) can cause both acute and chronic hepatitis. Antiviral therapy is the cornerstone for the treatment of chronic HCV infection once diagnosis is confirmed by PCR. The goal of antiviral therapy is to eradicate HCV RNA or attain sustained virological response (SVR). In many countries worldwide, including Egypt, HCV infection is treated with a combination of pegylated interferon α and ribavirin (RBV). Liver fibrosis/cirrhosis stage influences the response to pegylated interferon α and RBV. Even with new oral therapies such as Sovaldi many patients have to continue to be on combination regimens of interferon/RBV or RBV alone. In the current study, we aimed to use data mining analysis to determine sonographic pictures that can successfully predict SVR in HCV-4 patients before the antiviral therapy.
Methods
Eighty-two patients were enrolled in this study and they underwent two-dimensional ultrasound examination before the antiviral therapy. The sonographic data obtained were analyzed with Rapidminer version 4.6 to create a decision tree algorithm for the prediction of SVR.
Results
The absence of significant liver fibrosis was a predictive parameter of SVR mainly in those patients without a sonographic picture of cirrhosis. The resulting tree yielded an accuracy, sensitivity, and specificity of 85.82 ± 10.79, 68.75, and 96.00%, respectively, upon 10-fold cross-validation.
Conclusion
In the current study we used decision tree algorithm, one of the most important computational methods and tools for data analysis and predictive modeling in applied medicine, to predict SVR in HCV-infected patients. Two-dimensional ultrasound can give predictive information regarding the treatment outcome before interferon therapy for HCV-4.

Keywords: data mining, HCV-4, liver cirrhosis, prediction, two-dimensional ultrasound


How to cite this article:
Elrazek AE, Abdelazeem K, Elfattah MA, Foad M, Salama K, Elbanna A, Bilasy SE, Fakhry M, Mahfouz H. Noninvasive prediction of HCV-4 SVR by 2D US: a randomized study using data mining algorithm. Al-Azhar Assiut Med J 2016;14:14-8

How to cite this URL:
Elrazek AE, Abdelazeem K, Elfattah MA, Foad M, Salama K, Elbanna A, Bilasy SE, Fakhry M, Mahfouz H. Noninvasive prediction of HCV-4 SVR by 2D US: a randomized study using data mining algorithm. Al-Azhar Assiut Med J [serial online] 2016 [cited 2017 Dec 17];14:14-8. Available from: http://www.azmj.eg.net/text.asp?2016/14/1/14/180454


  Introduction Top


Hepatitis C virus (HCV) infection is a global health problem, with more than 180 million people (around 3% of the world's population) infected worldwide. HCV is an important cause of end-stage liver disease and hepatocellular carcinoma (HCC) [1],[2]. HCC is the third most common cause of cancer-related mortality worldwide [3],[4]. In Egypt, HCV infection is a major public health burden, with the highest prevalence worldwide (15%) [5],[6]. HCV genotype 4 (HCV-4) is the most common genotype in Egypt as it represents 90% of the total infections. HCV-4 infection is considered to be the major cause of chronic hepatitis and liver cirrhosis, which eventually leads to HCC in the country [7],[8],[9]. In a recent national retrospective study conducted on 1456 patients with HCC, HCV antibodies were detected in 91% [10].

In Egypt, treatment of chronic HCV infection with pegylated interferon α (PEG-IFN) and ribavirin (RBV) represents the gold standard for HCV management as treatment with sofosbuvir (Sovaldi), an oral nucleotide analog inhibitor of the HCV NS5B polymerase enzyme, is not well established yet. Sustained virological response (SVR) is the primary goal of the therapy; however, SVR rate is around 50% in HCV-4 infection [11]. Given the large number of people chronically infected with HCV-4 in Egypt and the expected economic and health burdens, proper selection of patients who are more likely to attain SVR will save money and effort and prevent unnecessary complications in poor responders. To date, several virological and host factors that can influence the treatment outcome have been proposed. Sequence heterogeneity within the viral NS5A in a region named Interferon Ribavirin Resistant Determining Region in HCV-4 was correlated with a positive treatment response [12],[13],[14].

Also, host factors such as the single-nucleotide polymorphism near the IL28B region were correlated with the treatment response [15],[16]. Liver cirrhosis is one of the determining factors for the treatment outcome. It is well established that the treatment response in patients with advanced fibrosis and cirrhosis is inferior to the response observed in patients with healthy liver or less severe fibrosis [17],[18]. Moreover, advanced liver cirrhosis or decompensated HCV cirrhosis affects the safety and tolerability of the IFN-based therapy.

A decision tree is the computing method of extracting valuable information and finding new correlations from analyzing current data. In this study, we used Rapidminer 4.6 algorithm to create a decision tree that predicts the treatment outcome of PEG-IFN/RBV-based therapy in Egyptian patients infected with HCV-4. Two-dimensional ultrasound (2D US) is a noninvasive technique that can be used to detect liver cirrhosis and was recently reported to be beneficial in detecting risky esophageal varices (EVs) on the basis of cirrhosis, ascites, and esophageal wall thickness while predicting the HCV treatment outcome of IFN-based therapy.


  Methods Top


Ethics statement

Ethical approval for this study was obtained from the Ethical Committee of Al Azhar University Hospitals. A written informed consent was obtained from patients enrolled in this study.

Patients

A total of 106 patients, 57 men (53.7%) and 49 women (46.3%), were prospectively examined for HCV infection by means of enzyme-linked immunosorbent assay and quantitative PCR.

HCV-positive patients underwent further investigations for HCV-related complications such as HCC and EVs. Inclusion criteria were HCV-4 infection with or without early stages of EVs (grades I and II).

Exclusion criteria were presence of alcoholic hepatitis, HCV, hepatits B virus coinfection (one patient), HCC (four patients), or advanced stages of EVs (grades III and IV; 19 patients). A total of 82 patients, between the ages of 21 and 64 years, with a mean age of 47 ± 14 years, consisting of 40 men (48.7%) and 42 women (51.3%) met the inclusion criteria of this study; 69 patients had HCV-4 infection without EVs and 13 patients had HCV-4 with grade I EVs. All patients were examined clinically and by 2D US between May and August 2013 in Egypt before being initiated on PEG-IFN/RBV.

Abdominal ultrasound

All radiological studies were performed by one sonarist with considerable experience in liver US to confirm the presence of liver cirrhosis, portal hypertension, and other important sonographic pictures of liver diseases. He was blinded to the results of the liver biopsy.

Data mining analysis

Data mining analysis is the process of examining large amounts of data by means of a computer to create an algorithm. Internal validation was performed using the test mode: 10-fold cross-validation using Naive Bayes application, which is generally applied to predict the performance of a model on a validation set using computation in place of mathematical analysis. All sonographic pictures and data obtained from each patient before PEG-IFN/RBV therapy were collected and analyzed using Rapid I, version 4.6 (Berlin, Germany), to create a decision tree algorithm. This algorithm was used to predict the treatment response in HCV-4 cirrhotic patients after PEG-IFN/RBV treatment.


  Results Top


Screening 106 patients for HCV infection and applying both the inclusion and exclusion criteria for this study yielded 82 patients positive for HCV-4 infection, 40 men (48.7%) and 42 women (51.3%), by means of enzyme-linked immunosorbent assay and quantitative PCR. All patients presented variable degrees of hepatitis, ranging from mild hepatitis activity to advanced cirrhosis; 69 patients had HCV-4 infection without EVs and 13 patients had HCV-4 with portal hypertension-induced small EVs (grades I and II).

Patients with large EVs (grades III and IV) were excluded or postponed for prophylactic measures. All patients underwent 2D US, and sonographic data obtained were used to create a decision tree algorithm. Ascites, esophageal wall thickness, and liver cirrhosis were the predictive criteria for SVR [Figure 1]. The resulting tree yielded an accuracy, sensitivity, and specificity of 85.82 ± 10.79, 68.75, and 96.00%, respectively, upon 10-fold cross-validation.
Figure 1: Intra-abdominal portion of the esophagus, just below left lobe of the liver (arrow). Anterior and posterior walls appearing hypoechoic, although the lumen in-between is hyperechoic.

Click here to view



  Discussion Top


Six common genotypes of HCV have been discovered around the world. All HCV genotypes can cause both acute and chronic hepatitis [19],[20],[21],[22],[23]. HCV-4 is the predominant strain in Egypt, Middle East, and Central and West Africa, with increasing rates in European countries such as France, Spain, Greece, and Italy related to immigration [24]. In Egypt, HCV infection is treated with PEG-IFN/RBV, and the goal of treatment is to eradicate viral RNA, which is predicted by the achievement of an SVR defined by the absence of HCV RNA by PCR 6 months after stopping treatment [25],[26].

There is a direct correlation between the degree of hepatic cirrhosis/fibrosis and outcome of hepatitis therapy. Furthermore, the degree of hepatic cirrhosis should be taken into consideration before HCV therapy [27]. Therefore, the accurate estimation of the degree of liver fibrosis/cirrhosis is important before the therapy for ascertaining prognosis, surveillance, and HCC development. Although liver biopsy is the gold standard for diagnosis of liver cirrhosis, it is subject to sampling error, complications, and seeding of malignancy. 2D US is a simple and easy technique available at all health centers and aids in detection of risky EVs [28] [Figure 1]. In the current study, we designed a decision tree algorithm using the baseline sonographic data obtained during 2D US screening before PEG-IFN-based therapy for the prediction of SVR. Our study shows that abdominal US has high specificity for predicting patients who will benefit from PEG-IFN/RBV therapy (96.00%) (the positive group) and is considerably reliable in determining those who will not benefit (68.75%) (the negative group). Accordingly, we can expedite treatment for the positive patient group.

Sonographicaly ascites, estimation of esophageal wall thickness, and degree of liver cirrhosis were the only significant predictive parameters, whereas the other sonographic parameters, such as splenomegaly, portal or splenic vein diameter, gall bladder wall thickness, presence of collaterals, and others, were not correlated to prediction of PEG-IFN/RBV therapy outcome. However, the degree of liver cirrhosis depends on the experience of sonarists. The presence of liver coarseness, attenuation of hepatic veins, irregular borders, shrunken size, hypertrophic caudate lobe, nodular margins, and presence of periportal fibrotic bands in sonographic pictures represent an advanced cirrhotic picture. If no sonographic pictures of cirrhosis are observed in HCV-positive patients, transient elastography and acoustic radiation force impulse imaging are two promising methods that can assess liver stiffness noninvasively [29],[30],[31].

We created a modified practical algorithm to predict the degree of liver fibrosis/cirrhosis in patients infected with HCV-4 who are candidates for therapy. Hence we can predict the therapy outcome [Figure 2]. Those with no cirrhosis sonographically, with ISHK score 0 or 1, had more than 70% SVR. However, those with advancing cirrhosis sonographically, corresponding to ISHK score 5 or 6, showed poor treatment response; and those with early sonographic cirrhotic pictures, with ISHK score 2–4, showed 40–70% SVR.
Figure 2: Flow chart algorithm for SVR prediction therapy using 2D US. 2D US, two-dimensional ultrasound; EVs, esophageal varices; SVR, sustained virological response.

Click here to view


Similar results were observed in a few cases with HCV-4 infection treated with the new oral therapy sofosbuvir [6].

Results obtained from this study will be especially valuable for patients with advanced cirrhosis who cannot benefit from IFN-based therapy.

In addition, our results will prompt patients showing early signs of liver cirrhosis and those in the most favorable disease stage to start therapy and will thus expedite treatment. Screening with 2D US will be extremely beneficial in developing countries such as Egypt especially when advanced cirrhosis before therapy and therapy success rate, also viral relapse increased with advanced cirrhosis. Further, although universal application of screening is controversial, selective antenatal HCV screening in high-risk populations is highly recommended; women included in our study with HCV-4 were screened for esophageal wall thicknesses during pregnancy before the therapy.

Data mining offers new insights into the analysis of medical data and aids in the construction of predictive models [32],[31],[32],[3],[33],[34],[35],[36]. Further, predictive data mining is becoming instrumental for researchers and clinical practitioners in making more informed treatment decisions that will ultimately benefit the patients. To conclude, through data mining analysis, we found that 2D US can give predictive information on the treatment outcome before interferon therapy for HCV-4 even with the combination regimen of sofosbuvir/RBV with or without interferon therapy for those with HCV G4.

Addressed in details in future studies especially for those recommended for all seven oral free therapy through ultrasound or transient elastography/acoustic radiation force impulse technology.

Acknowledgements

Author contributions: Abd Elrazek M. Ali Abd Elrazek: planning and conducting the study, collecting and interpreting data, drafting the manuscript, helped in statistical analysis, he performed the medical data mining analysis using Rapidminer version 4.6, he approved the final draft submitted; Khaled Abd Elazeem: approved the final draft submitted; Mohamed Abd Elfattah: interpreting data and drafting the manuscript, followed patients up during the study; he approved the final draft submitted; Mahmoud Foad: drafting the manuscript, he approved the final draft submitted; Khaled salama: drafting the manuscript, he approved the final draft submitted; Abduh Elbanna: drafting the manuscript, he approved the final draft submitted; Shymaa E. Bilasy: conducting the study, critical reading and revising the manuscript, she approved the final draft; Mohamed Fakhry: drafting the manuscript; Hamdy Mahfouz: drafting the manuscript, he approved the final draft submitted.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

 
  References Top

1.
Poynard T, Yuen MF, Ratziu V, Lai CL, Poynard T, Yuen MF, et al. Viral hepatitis C. Lancet 2003; 362 (9401):2095–2100.  Back to cited text no. 1
    
2.
Maekawa S, Enomoto N. Viral factors influencing the response to the combination therapy of peginterferon plus ribavirin in chronic hepatitis C. J Gastroenterol 2009; 44 (10):1009–1015.  Back to cited text no. 2
    
3.
Kamangar F, Dores GM, Anderson WF. Patterns of cancer incidence, mortality, and prevalence across five continents: defining priorities to reduce cancer disparities in different geographic regions of the world. J Clin Oncol 2006; 24 (14):2137–2150.  Back to cited text no. 3
    
4.
Davila JA, Morgan RO, Shaib Y, McGlynn KA, El-Serag HB. Hepatitis C infection and the increasing incidence of hepatocellular carcinoma: a population-based study. Gastroenterology 2004; 127 (5):1372–1380.  Back to cited text no. 4
    
5.
Ray SC, Arthur RR, Carella A, Bukh J, Thomas DL. Genetic epidemiology of The degree of liver cirrhosis predict the outcome of therapy, recently, we hepatitis C virus throughout Egypt. J Infect Dis 2000; 182:698–707.  Back to cited text no. 5
    
6.
Abd Elrazek AE, Bilasy SE, Elbanna AE, Elsherif AE. Prior to the oral therapy, what do we know about HCV-4 in Egypt: a randomized survey of prevalence and risks using data mining computed analysis. Medicine (Baltimore) 2014;93(28):e204.  Back to cited text no. 6
    
7.
Hassan MM, Zaghloul AS, El-Serag HB, Soliman O, Patt YZ, Chappell CL, et al. The role of hepatitis C in hepatocellular carcinoma: a case control study among Egyptian patients. J Clin Gastroenterol 2001; 33 (2):123–126.  Back to cited text no. 7
    
8.
el-Zayadi AR, Badran HM, Barakat EM, Attia Mel-D, Shawky S, Mohamed MK, et al. Hepatocellular carcinoma in Egypt: a single center study over a decade. World J Gastroenterol 2005; 11 (33):5193–5198.  Back to cited text no. 8
    
9.
Guerra J, Garenne M, Mohamed MK, Fontanet A. HCV burden of infection in Egypt: results from a nationwide survey. J Viral Hepat 2012; 19 (8):560–567.  Back to cited text no. 9
    
10.
Shaker MK, Abdella HM, Khalifa MO, El Dorry AK. Epidemiological characteristics of hepatocellular carcinoma in Egypt: a retrospective analysis of 1313 cases. Liver Int 2013; 33 (10):1601–1606.  Back to cited text no. 10
    
11.
Fried MW, Shiffman ML, Reddy KR, Smith C, Marinos G, Gonçales FL Jr, et al. Peginterferon alfa-2a plus ribavirin for chronic hepatitis C virus infection. N Engl J Med 2002; 347 (13):975–982.  Back to cited text no. 11
    
12.
El-Shamy A, Nagano-Fujii M, Sasase N, Imoto S, Kim SR, Hotta H. Sequence variation in hepatitis C virus nonstructural protein 5A predicts clinical outcome protein and clinical outcome of pegylated-interferon/ribavirin combination therapy. Intervirology 2012; 55:1–11.  Back to cited text no. 12
    
13.
El-Shamy A, Kim SR, Ide YH, Sasase N, Imoto S, Deng L, et al. Polymorphisms of hepatitis C virus non-structural protein 5A and core protein and clinical outcome of pegylated-interferon/ribavirin combination therapy. Intervirology 2012; 55 (1):1–11.  Back to cited text no. 13
    
14.
El-Shamy A, Shoji I, El-Akel W, Bilasy SE, Deng L, El-Raziky M, et al. NS5A sequence heterogeneity of hepatitis C virus genotype 4a predicts clinical outcome of pegylated-interferon-ribavirin therapy in Egyptian patients. J Clin Microbiol 2012; 50 (12):3886–3892.  Back to cited text no. 14
    
15.
Asselah T, De Muynck S, Broët P, Masliah-Planchon J, Blanluet M, Bièche I, et al. IL28B polymorphism is associated with treatment response in patients with genotype 4 chronic hepatitis C. J Hepatol 2012; 56 (3):527–532.  Back to cited text no. 15
    
16.
De Nicola S, Aghemo A, Rumi MG, Galmozzi E, Valenti L, Soffredini R, et al. Interleukin 28B polymorphism predicts pegylated interferon plus ribavirin treatment outcome in chronic hepatitis C genotype 4. Hepatology 2012; 55 (2):336–342.  Back to cited text no. 16
    
17.
Everson GT. Treatment of hepatitis C in patients who have decompensated cirrhosis. Clin Liver Dis 2005; 9 (3):473–861.  Back to cited text no. 17
    
18.
Ali Hussein AE, Mahfouz H, Elazeem KA, Fakhry M, Elrazek EA, Foad M, et al. The Value of U/S to Determine Priority for Upper Gastrointestinal Endoscopy in Emergency Room, Medicine (Baltimore) 2015;94(49):e2241  Back to cited text no. 18
    
19.
El-Zanaty F, Way A. Egypt demographic and health survey 2008. Egyptian: Ministry of Health. Cairo: El-Zanaty and Associates, and Macro International; 2009.  Back to cited text no. 19
    
20.
Lavanchy D. Evolving epidemiology of hepatitis C virus. Clin Microbiol Infect 2011; 13:107–115.  Back to cited text no. 20
    
21.
Alter MJ. Epidemiology of hepatitis C virus infection. World J Gastroenterol 2007; 13 (17):2436–2441.  Back to cited text no. 21
    
22.
Shepard CW, Finelli L, Alter MJ. Global epidemiology of hepatitis C virus infection. Lancet Infect Dis 2005; 5 (9):558–567.  Back to cited text no. 22
    
23.
Frank C, Mohamed MK, Strickland GT, Lavanchy D, Arthur RR, Magder LS, et al. The role of parenteral antischistosomal therapy in the spread of hepatitis C virus in Egypt. Lancet 2000; 355 (9207):887–891.  Back to cited text no. 23
    
24.
Kamal SM, Nasser IA. Hepatitis C genotype 4: what we know and what we don't yet know. Hepatology 2008; 47 (4):1371-1383.  Back to cited text no. 24
    
25.
Ghany MG, Strader DB, Thomas DL, Seeff LB, American Association for the Study of Liver Diseases Diagnosis, management, and treatment of hepatitis C: an update. Hepatology 2009; 49 (4):1335-1374.  Back to cited text no. 25
    
26.
Feld JJ, Hoofnagle JH. Mechanism of action of interferon and ribavirin in treatment of hepatitis C. Nature 2005; 436 (7053):967-972.  Back to cited text no. 26
    
27.
Corchado S, López-Cortés LF, Rivero-Juárez A, Torres-Cornejo A, Rivero A, Márquez-Coello M, Girón-González JA Liver fibrosis, host genetic and hepatitis C virus related parameters as predictive factors of response to therapy against hepatitis C virus in HIV/HCV coinfected patients. PLoS One 2014; 9 (7): e101760.  Back to cited text no. 27
    
28.
Abd Elrazek AE, Mahfouz H, Afifi M, Nafady M, Fathy AE, Elazeem KA, et al. Detection of risky esophageal varices by 2D U/S: when to perform endoscopy. Am J Med Sci 2014; 347:28–33.  Back to cited text no. 28
    
29.
deLédinghen V, Vergniol J. Transient elastography for the diagnosis of liver fibrosis. Expert Rev Med Devices 2010; 7:811–823.  Back to cited text no. 29
    
30.
D'Onofrio M, Crosara S, De Robertis R, Canestrini S, Demozzi E, Gallotti A, Pozzi Mucelli R Acoustic radiation force impulse of the liver. World J Gastroenterol 2013; 19 (30):4841–4849.  Back to cited text no. 30
    
31.
Abd Elrazek M Ali, Elsherif Asim, Osama Abd elal, Khaled Abd El azeem, Samir El sherbiny, Shymaa Bilasy. Screening esophagus during routine U/S: Medical and Cost benefits. EJGH 2015;27(1):8-12.  Back to cited text no. 31
    
32.
AEM Ali, H Mahfouz. Prediction of esophageal variceal degrees using data mining: is validated in clinical medicine? Global J Comp Sci Tech 2013; 13:10.  Back to cited text no. 32
    
33.
Bellazzi R, Zupan B. Predictive data mining in clinical medicine: current issues and guidelines. Int J Med Inform 2008; 77 (2):81–97.  Back to cited text no. 33
    
34.
Abd Elrazek AE, Mahfouz HM, Metwally AM, El-Shamy AM. Mortality prediction of nonalcoholic patients presenting with upper gastrointestinal bleeding using data mining. Eur J Gastroenterol Hepatol 2014; 26 (2):187-191.  Back to cited text no. 34
    
35.
Grossi E, Buscema M. Introduction to artificial neural networks. Eur J Gastroenterol Hepatol 2007; 19 (12):1046-1054.  Back to cited text no. 35
    
36.
Aly Abd Elrazek AEM, Bilasy SE, Elbanna AE, Elsheriff AE. Mortality prediction of nonalcoholic patients presenting with upper gastrointestinal bleeding using data mining. Medicine (Baltimore) 2014; 93 (28):e204.  Back to cited text no. 36
    


    Figures

  [Figure 1], [Figure 2]



 

Top
 
 
  Search
 
Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

 
  In this article
   Abstract
  Introduction
  Methods
  Results
  Discussion
   References
   Article Figures

 Article Access Statistics
    Viewed271    
    Printed4    
    Emailed0    
    PDF Downloaded60    
    Comments [Add]    

Recommend this journal


[TAG2]
[TAG3]
[TAG4]