PEDIATRICS Vol. 120 No. 1 July 2007, pp. e56-e60 (doi:10.1542/peds.2006-1364)
ARTICLE |
Detection and Significance of Serum Protein Marker of Hirschsprung Disease
a Pediatric Surgery Department, First Affiliated Hospital, Zhengzhou University, Zhengzhou, China
b Cancer Institute, Second Affiliated Hospital, College of Medicine, Zhejiang University, Hangzhou, Zhejiang, China
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
OBJECTIVE. The objective of this study was to identify a specific fingerprint chromatogram model of serum proteins for early screening and diagnosis of Hirschsprung disease.
METHODS. To detect the protein mass spectrograms of 78 serum specimens (42 specimens of Hirschsprung disease, 16 specimens of adhesive ileus including appendicitis and Meckel diverticulum after operation and inflammatory bowel disease, and 20 specimens of normal control subjects), we used surface-enhanced laser desorption/ionization time of flight mass spectrometry technology, combined with bioinformatics methods (support vector machine) to develop and compare protein mass spectrograms from serum samples.
RESULTS. We identified 3 protein markers, the mass-to-charge ratio of which is positioned at 3221.7, 5639.2, and 6884.2 from the fingerprint chromatogram model of serum protein for early screening and diagnosis of Hirschsprung disease. The markers had 100% sensitivity and specificity.
CONCLUSION. The fingerprint chromatogram model of serum protein using surface-enhanced laser desorption/ionization time of flight mass spectrometry technology combining support vector machine is a new method of early screening and diagnosis of Hirschsprung disease that is worthy of additional research and application.
Key Words: Hirschsprung disease • diagnosis • SELDI • support vector machine • fingerprint chromatogram of protein
Abbreviations: HSCR—Hirschsprung disease • SELDI-TOF-MS—surface-enhanced laser desorption-ionization time-of-flight mass spectrometry • m/z—mass-to-charge ratio
Hirschsprung disease (HSCR) is a common intestinal disease that is characterized by abnormal neurologic development of the intestine. Successful early screening and diagnosis and timely treatment significantly improve prognosis. Current clinical practice that is based on clinical suspicion alone infrequently provides early diagnosis. Advances in microprocessing and microelectronic technologies have allowed construction of microbe biochemical analysis systems on solid chip surfaces that can be used to detect quantitatively disease markers. The "protein chip" is a means of examining for the presence or absence of thousands of proteins in serum samples. The chip can be an ideal method to discover protein markers of disease.1,2 We used this technology and bioinformatic techniques to identify markers to detect whether presence or absence of specific serum proteins would distinguish children who had HSCR from normal children. Successful identification of characteristic patterns of presence or absence of specific protein peaks in this pilot study would justify future investigations using this technology in larger, prospective, population-based samples, as well as provide insights for investigations that could lead to a better understanding of the pathophysiology of HSCR.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Materials
Seventy-eight serum specimens were obtained from the Pediatric Surgery Department of the First Affiliated Hospital of Zhengzhou University, among those, 42 case specimens were of patients with HSCR (including long-segment type [7], short-segment type [30], super short–segment type [4], and leap type [1]), 16 specimens were of adhesive ileus including appendicitis (8) and Meckel diverticulum (3) after operation and inflammatory bowel disease (5), and the 20 controls are from healthy children who had a physical checkup in Pediatric Surgery Department in our hospital. All of the specimens with HSCR were previously confirmed by pathologic results. Among the children with HSCR, 33 were male and 9 were female; their ages were 3 days to 18 years. Three cases were diagnosed when the patients presented with megacolon crisis after they were hospitalized, and they were urgently treated with colostomy. All of the blood specimens were drawn on the morning when the patients had fasted. The specimens were placed at room temperature for 1 to 2 hours and turned centrifugally for 10 minutes, and then the serum was extracted and preserved under –80°C. Consent for collecting serum specimens was obtained from the Ethics Committee of Honan province and also from the patients and normal control subjects.
Main Reagents and Devices
CHAPS, urea, dithiothreitol, sodium acetate, and sinapinic acid were purchased from Promega (Madison, WI). PBS II+ surface-enhanced laser desorption-ionization time-of-flight mass spectrometry (SELDI-TOF-MS) and WCX2 protein chips were purchased from Ciphergen (Fremont, CA).
Protein Chip Techniques
Serum specimens were thawed in an ice bath and centrifuged at 10000 rpm under 4°C for 2 minutes. A 96-well plate was placed on an ice box, 10 µL of U9 (9M Urea, 2% CHAPS, and 1% dithiothreitol) and 5 µL of serum were added to each well, and the plate was vibrated at 600 rpm under 4°C for 30 minutes in a cold laboratory chamber. The chip was prepared by placing it in the bioprocessor, noting the chip number, adding 200 µL of sodium acetate (100 Mm, pH 4) to each well, vibrating at 600 rpm for 2 minutes in a cold laboratory chamber, and repeating the operation once. The 96-well plate being processed by U9 was placed on the ice, and a medical gun was used to add 185 µL of sodium acetate. The plate was then vibrated at 600 rpm under 4°C for 2 minutes in a cold laboratory chamber. A total of 100 µL of processed specimens was added to the chip, which was then placed in the cold laboratory chamber under 4°C combining 600 rpm for 60 minutes. The remaining liquid was swung off, and the sample was dried rapidly. A total of 200 µL of sodium acetate was added, and after vibrating at 600 rpm for 5 minutes was swung off and dried. This was repeated 3 times. Each well was washed twice using 200 µL of deionized water, and excessive water was swung off. After the chip was air-dried, each 50% saturated sinapinic acid (1 µL) was added in 2 stages. After drying, they were placed on the device for testing.
Data Collection and Processing
A protein chip whose molecular weight was known was used to adjust the SELDI-TOF-MS system, until the tolerance of molecular weight was <0.1%. A mass spectrum reader was used to analyze the WCX2 protein chip combined with protein. Analysis parameters included laser strength of 170 and sensitivity of 6; the total number of each specimen collection was repeated 140 times. The scope of data collection was 1000 to 30000 Da; the optimized scope was 2000 to 20000 Da. Quality control serum was used to make repeated tests; the coefficients of variation of the peak value and the strength were 0.05% and 19.7%, respectively. All of the data used Protein Chip Software 3.1 (Ciphergen) to adjust and to make the strength and molecular weight of the total ions uniform. The ZUCI-Protein Chip Data Analyze System software package (Zhejiang University) was used to analyze the results. Peaks with mass-to-charge ratio (m/z) of <2000 were left out, and discrete wavelength analysis was used to eliminate noise and subtract the baseline. The method of local extremum was used to find out the respective peaks of the specimens and filter out the peaks with signal-to-noise ratio <2. Clustering analysis regards 10% as the minimum threshold, to cluster all of the specimens' peaks with m/z variation <0.3% to 1 class.
Support Vector Machine
A linear support vector machine (SVM)3 classifier was used to identify peaks. Radial-based kernel function was adopted with its 
value set at 0.6 and penalty function C at 19. The selection of feature vector used the method of statistical filtration combined with model-dependent screening to build a discrimination model. The method of leave-1-out crossing verification was used to assess the discrimination result of the model. This experiment also used discrimination analysis methods to process mass spectrum data and to verify the result being processed by SVM.
Statistical Analysis
After the noise was filtered out of the original mass spectrum, data were filtered, and after clustering analysis, the m/z peaks data analysis was conducted using the Wilcoxon rank sum test; the testing standard was set at 
= .01 to account for multiple testing.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
HSCR Group and Normal Control Group
After the mass spectrum data of the HSCR group and the normal control group were filtered and screened, 213 m/z peaks were attained. After carrying out Wilcoxon rank sum tests to test relative signal strength, 13 m/z peaks with P < .01 were obtained. From the random combination of protein peaks with remarkable variation, SVM screened out the combination model with the maximum Youden index of the predicted value, identifying 3 markers positioned at 3221.7, 5639.2, and 6884.2. In the HSCR group, the proteins were not significantly expressed, whereas in normal control group, they were noted to have high expression (Figs 1–3). Combining 3 potential markers, using the method of leave-1-out to make crossing detection, in the test collection of 62 patients, the specificity of discrimination model was 100%, and its sensitivity was 100%.
|
|
Adhesive Ileus Group and Normal Control Group
Through deleting 16 specimens of adhesive ileus including appendicitis and Meckel diverticulum after operation and inflammatory bowel disease and normal control group, there is no significant difference between the 2 groups positioned at the 3 markers.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
HSCR is a congenital intestinal disease. The pathologic change is that a portion of the intestinal wall lacks ganglion cells. Its incidence rate is 1:5000; the proportion between male and female is 4:1. Its main cause is that ganglion cells cannot cluster and locate in the intestinal wall. Genetic predisposition and intestinal microenvironment changes may contribute to development of HSCR.4 Diagnosis of HSCR mainly relies on clinical manifestations noted by an observant caregiver and subsequent diagnostic testing using variously invasive techniques, including barium enema reduction, rectal mucus biopsy, and anorectal pressure measurement. Some newborns with HSCR are missed because their clinical signs are not typical when they are born. In newborn infants, The medical literature reports that 20% of the rectal contrasts of barium enema reduction are mistakenly diagnosed.5 This is most likely because in newborns, the dividing lines among the spasm section, transfer section, and extension section are not clear; that is, the expression of HSCR X line is not typical in this period. In addition, when carrying out barium enema reduction, because of the improper operation in the preparation of purifying intestine and injecting bubbles into the anus, "false megacolon" may be found. The barium enema is also not without risk. If the barium is not diluted properly, then water intoxication may develop; if before examination the child has developed enterocolitis to the contrast, then enema may result in intestinal perforation and consequently lead to barium peritonitis. Another frequently used diagnostic test, rectal mucus biopsy, is a traumatic and imperfect examination. If the scope of biopsy is extensive, then the trauma is relatively large; if the scope of biopsy is narrow, then the probability of missed diagnosis increases, especially for patients with short-segment and super short–segment HSCR. Besides, transfer mucus receiving histochemical examination is at the risks of hemorrhage and perforation, and the pathologic technology requirements are relatively high. The positive rate is 83%.6 The diagnostic accuracy rate for anorectal pressure measurement in newborn infants with HSCR is approximately 64.3% to 71.43%7,8; false-positives that lead to missed diagnosis may occur as a result of improper operation of the measuring device.
Susceptibility to HSCR has been attributed to the following 9 genes: RET, GDNF, NTN, EDNRB, EDN3, ECE-1, SOX-10, ZFHX1B, and PHOX2B9,10; the ones that are more widely researched are those that are associated with abnormalities of GDNF/RET signal transmission pathways. Takahashi et al11 were the first to discover RET gene in the medullary thyroid cancer and discovered that it is closely related to the growth, division, transfer, and locating of intestinal ganglion cells.12 The research by Bordeaux et al13 confirmed that GDNF, which is the ligand of RET, can inhibit the cell-decaying process that is caused by RET. Warnovara et al14 used in situ hybridization technology and reverse transcription–polymerase chain reaction method and discovered that there exist GDNF expression in the colon of both fetuses and newborn infants, but it was not discovered in other parts of intestine. Bar et al15 adopted an immunohistochemistry technique and discovered that GDNF was mainly present in neuroglial cells and Schwann cells, in muscular layer, and these 2 compositions are much more than that in mucosa. The theory of colonic gene expression does not explain the extremely rare sectional aganglionosis (also called "leap" HSCR), which is a rare special type.16 Because the clinical diagnostic tools are imperfect and a scanning of the specific candidate genes will not identify 100% of infants with HSCR, identification of ideal serum biological markers would be extremely useful. Ideally, the serum markers would identify the typical and difficult-to-diagnose nontypical, short-segment type, super short–segment type, and leap HSCR. SELDI-TOF-MS technology is a new proteome technology that was developed in 2002. It uses the principles of gene chip, reasonably combining chromatography and mass spectrum technology with protein chip, and it is able to detect proteins and peptides that are hard to verify with the traditional methods, although it does not specifically identify the proteins other than by size. It has the characteristics of rapidity, sensitivity, and high throughput,17 and it has made major breakthroughs in the marker screening and early diagnosis of ovarian cancer,18 prostate cancer,19 breast cancer,20 lung cancer,21 liver cancer,22 and colon cancer.23
This experiment used SELDI-TOF-MS technology combining the method of bioinformatics. It is the first to apply proteome technology to the detection of HSCR serum protein markers and discovered specific marker proteins in the children with HSCR. These marker proteins are based on the SVM combination to build a fingerprint chromatogram model of serum protein and to discriminate successfully children with HSCR from normal children with 100% sensitivity and specificity. This diagnostic model may successfully distinguish children with HSCR from healthy children, but more testing in larger populations with well-matched case patients and control subjects and, ideally, affected and unaffected siblings is needed. It may have important practical value, particularly in the diagnosis of HSCR with nonfamilial and the nontypical types, short-segment type, super short–segment type, and the leap type. The 3 markers at the positions of 3221.7, 5639.2, and 6884.2 are of low expression in the HSCR group and of high expression in the normal control and adhesive ileus groups. This shows that we can use SELDI-TOF-MS technology to find out protein markers for the early screening and diagnosis of HSCR.
Use of SELDI-TOF-MS technology to analyze protein will produce tremendous mass spectra data that challenge the traditional data processing and analysis.24 SVM used in this experiment is a kind of classification technology proposed by Vapnik25 and others. It is a new machine learning method developed on the basis of statistical theory. In the model discrimination, the popularization, model selection, overfitting, latitude disaster, and other problems of the small specimen model have been solved successfully in SVM.26–28 In the data processing of this experiment, we eliminated noise by discrete wavelength, found out mass-charge peaks of the specimens using the method of local extremum, and clustered mass-charge peaks by setting 10% as the minimum threshold. Wilcoxon rank sum test analysis assesses the relative importance of each peak in the discrimination of 2 kinds of specimens according to P values. To combine randomly the remarkably different mass-charge peaks and to input them into SVM, screen out the markers, and build the discrimination model, use the method of leave 1 out to assess the model by means of crossing verification; that is, to regard 1 specimen as testing and the other specimens as training, make repeated tests until the result is stable. Because each test collection is independent from the training specimens, it can be a totally blind trial. Going through these procedures and combining many methods to process the data ensures the popularization of the model built and the accuracy of the prediction.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The diagnostic model established in this experiment using SELDI-TOF-MS technology combined with SVM to build fingerprint chromatogram models of serum protein of patients with HSCR and normal control subjects shows its potential value in the screening and diagnosis of a difficult-to-diagnose, complex disease. Additional characterization of the identity and function of the 3 identified peptides may contribute to understanding of HSCR pathophysiology. Future studies using cases of varying phenotype and well-matched controls and prospective studies comparing the sensitivity and specificity of existing methods with this new technique are indicated.
|
|
ACKNOWLEDGMENTS |
|---|
We acknowledge the financial support of the National Natural Science Foundation of China (30430730).
|
FOOTNOTES |
|---|
Accepted Dec 12, 2006.
Address correspondence to Shu Zheng, MD, Cancer Institute, Zhejiang University, Hangzhou Zhejiang 310009, China. E-mail: zhengshu{at}mail.hz.zj.cn
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
1. Zhang L, Yu W, He T, et al. Contribution of human alpha-defensin 1, 2, and 3 to the anti-HIV-1 activity of CD8 antiviral factor. Science. 2002;298 :995 –1000
2. Spencer D, Robson L, Purdy D, et al. A strategy for mapping and neutralizing conformational immunogenic sites on protein therapeutics. Proteomics. 2002;2 :271 –279[CrossRef][Web of Science][Medline]
3. Liu Y. Active learning with support vector machine applied to gene expression data for cancer classification. J Chem Inf Comput Sci. 2004;44 :1936 –1941[CrossRef][Web of Science][Medline]
4. Parisi MA, Kapur RP. Genetics of Hirschsprung disease. Curr Opin Pediatr. 2000;12 :610 –617[CrossRef][Web of Science][Medline]
5. Raffenserger JG. Swenson's Pediatric Surgery. Chicago, IL: University of Chicago Press; 1980:507 –531
6. De Brito IA, Maksoud JG. Evolution with age of the acetylcholinesterase activity in rectal suction biopsy in Hirschsprung disease. J Pediatr Surg. 1987;22 :425 –430[CrossRef][Web of Science][Medline]
7. Morais MB, Sdepanian VL, Tahan S, et al. Effectiveness of anorectal manometry using the balloon method to identify the inhibitory recto-anal reflex for diagnosis of Hirschsprung's disease [in Portuguese]. Rev Assoc Med Bras. 2005;51 :313 –317[Medline]
8. Zhou XL, Chen FB, Ou BY. Clinical significance of anorectal manometry in newborn in Hirschsprung disease. Chin J Pediatr. 2004;9 :681 –683
9. Childs B, Beaudet AL, Valle D, Kinzler KW, Vogelstein B. Hirschsprung disease. In: Scriver CR, Sly WS, eds. The Metabolic and Molecular Bases of Inherited Disease. New York, NY: The McGraw-Hill Companies; 2004:1 –47
10. Amiel J, Laudier B, Attie-Bitach T, et al. Polyalanine expansion and frameshift mutations of the paired-like homeobox gene PHOX2B in congenital central hypoventilation syndrome. Nat Genet. 2003;33 :459 –461[CrossRef][Web of Science][Medline]
11. Takahashi M, Ritz J, Cooper GM. Activation of a novel human transforming gene, ret, by DNA rearrangement. Cell. 1985;42 :281 –288[CrossRef][Web of Science][Medline]
12. Kapur RP. Hirschsprung disease and other enteric dysganglionoses. Crit Rev Clin Lab Sci. 1999;36 :225 –273[CrossRef][Web of Science][Medline]
13. Bordeaux MC, Forcet C, Granger L, et al. The RET proto-oncogene induces apoptosis: a novel mechanism for Hirschsprung's disease. EMBO J. 2000;19 :4056 –4063[CrossRef][Web of Science][Medline]
14. Warnovara K, Salo M, Sainio K, et al. Distribution of glial cell line-derived neurotrophic factor mRNA in human colon suggests roles for muscularis mucosae in innervation. Pediatr Surg Int. 1998;13 :581 –583[CrossRef][Web of Science][Medline]
15. Bar KJ, Facer P, Williams NS, Tam PK, Anand P. Glial-derived neurotrophic factor in human adult and fetal intestine and in Hirschsprung's disease. Gastroenterology. 1997;112 :1381 –1385[CrossRef][Web of Science][Medline]
16. Yang HY, Liu QL, Wang JX, Xu HF. Clinical study of multiple zonal aganglionosis in long segment Hirschsprung disease. Natl Med J China. 2005;39 :2772 –2774
17. Weinberger SR, Boschetti E, Santambien P, Brenac V. Surface-enhanced laser desorption-ionization retentate chromatography mass spectrometry (SELDI-RC-MS): a new method for rapid development of process chromatography conditions. J Chromatogr B Analyt Technol Biomed Life Sci. 2002;782 :307 –316[CrossRef][Web of Science][Medline]
18. Vlahou A, Schorge JO, Gregory BW, Coleman RL. Diagnosis of ovarian cancer using decision tree classification of mass spectral data. Biomed Biotechnol. 2003;2003 :308 –314[CrossRef]
19. Wagner M, Naik DN, Pothen A, et al. Computational protein biomarker prediction: a case study for prostate cancer. BMC Bioinformatics. 2004;5 :26A[CrossRef]
20. Vlahou A, Laronga C, Wilson L, et al. A novel approach toward development of a rapid blood test for breast cancer. Clin Breast Cancer. 2003;4 :203 –209[Medline]
21. Xiao XY, Tang Y, Wei XP, He DC. A preliminary analysis of non-small cell lung cancer biomarkers in serum. Biomed Environ Sci. 2003;16 :140 –147[Web of Science][Medline]
22. Wang JX, Zhang B, Yu JK, et al. Using ANN and serum protein pattern models in liver cancer diagnosis [in Chinese]. Zhonghua Yi Xue Za Zhi. 2005;85 :189 –192[Medline]
23. Chen YD, Zheng S, Yu JK, Hu X.
Clin Cancer Res. 2004;10
:8380
–8385
24. Somorjai RL, Dolenko B, Baumgartner R. Class prediction and discovery using gene microarray and proteomics mass spectroscopy data: curses, caveats, cautions.
Bioinformatics. 2003;19
:1484
–1491
25. Vapnik V. Statistical Learning Theory. New York, NY: Wiley; 1998
26. Wang XD, Wang JQ. A survey on support vector machines training and testing algorithms. Computer Engineering and Application. 2004;13 :75 –79
27. Byvatov E, Schneider G. Support vector machine applications in bioinformatics. Appl Bioinformatics. 2003;2 :67 –77[Medline]
28. Jorissen RN, Gilson MK. Virtual screening of molecular databases using a support vector machine. Chem Inf Model. 2005;45 :549 –561[CrossRef]
PEDIATRICS (ISSN 1098-4275). ©2007 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
