| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Clinical Oncology, the Sir Y.K. Pao Centre for Cancer, The Chinese University of Hong Kong, Shatin, Hong Kong, the People’s Republic of China.
2 Ciphergen Biosystems, Inc., Fremont, CA 94555.
3 Cancer Research UK Institute for Cancer Studies, University of Birmingham, Vincent Drive, Edgbaston, Birmingham B15 2TT, England.
aAddress correspondence to this author at: Department of Clinical Oncology, Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, Hong Kong. Fax 852-2649-7426. or Cancer Research UK Institute for Cancer Studies, University of Birmingham, Vincent Drive, Edgbaston, Birmingham B15 2TT, England. E-mail JohnsonP{at}cancer.bham.ac.uk.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: Proteomes in sera from 20 CLD patients with 
-fetoprotein (AFP) <500 µg/L (control group) and 38 HCC patients (disease group) were profiled by anion-exchange fractionation (first dimension), two types (IMAC3 copper and WCX2) of ProteinChip® Arrays (second dimension), and time-of-flight mass spectrometry (third dimension). Bioinformatic tests were used to identify tumor-specific proteomic features and to estimate the values of the tumor-specific proteomic features in the diagnosis of HCC. Cross-validation was performed, and we also validated the models with pooled sera from the control and disease groups, serum from a CLD patient with AFP >500 µg/L, and postoperative sera from two HCC patients.
Results: Among 2384 common serum proteomic features, 250 were significantly different between the HCC and CLD cases. Two-way hierarchical clustering differentiated HCC and CLD cases. Most HCC cases with advanced disease were clustered together and formed two subgroups that contained significantly more cases with lymph node invasion or distant metastasis. For differentiation of HCC and CLD by an artificial network (ANN), the area under the ROC curve was 0.91 (95% confidence interval, 0.82–1.01; P <0.0005) for all cases and 0.954 (95% confidence interval, 0.881–1.027; P <0.0005) for cases with nondiagnostic serum AFP (<500 µg/L). At a specificity of 90%, the sensitivity was 92%. Both cluster analysis and ANN correctly classified the pooled serum samples, the CLD serum sample with increased AFP, and the HCC patient in complete remission.
Conclusion: Tumor-specific proteomic signatures may be useful for detection and classification of hepatocellular cancers.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-fetoprotein (AFP;
1
reference interval, 0–10 µg/L) is widely used as a serologic tumor marker for hepatocellular carcinoma (HCC) (1). Approximately 80% of patients with HCC will have concentrations above the reference interval (1), but modestly increased AFP (10–500 µg/L) may also be detected in nonmalignant chronic liver disease (CLD), leading to a decrease in the specificity of the AFP test for HCC (1). This poses an important clinical problem because most HCCs arise in patients with coexisting CLD, usually at the stage of cirrhosis (2). There is therefore a need to look for new markers for the effective identification of HCC. By analyzing and comparing the proteomes of tissues or sera from cancer patients and control individuals, attempts have been made to identify cancer-associated proteins. HCC, bladder carcinoma, breast carcinoma, renal cell carcinoma, and lung carcinoma have been studied with use of traditional two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) (3). Although various approaches have been developed for protein expression analysis (4), the combination of 2D-PAGE and mass spectrometry (MS) has been the most successful (4)(5)(6)(7). There are, however, limitations in the 2D-PAGE approach. Hydrophobic, strongly acidic, or strongly basic proteins are poorly resolved by 2D-PAGE (4). The utility of this technology is also limited by an inability to accurately monitor proteins that are present at very low abundance (4).
Surface-enhanced laser desorption/ionization time-of-flight (SELDI-TOF) MS is a novel non-electrophoresis-based proteomic technology that was introduced by Hutchens and Yip (8) and forms the core technology of the SELDI ProteinChip® Biomarker System (9). In this system, proteins are retained on a solid-phase chromatographic surface (i.e., the ProteinChip Array) and are subsequently ionized and detected by TOF MS. The ProteinChip System can be used in protein profiling of various biological materials, such as serum, urine, and cell lysates (9)(10)(11)(12)(13). It has been shown to be useful in the discovery of potential diagnostic markers for cancers such as prostate (10), ovarian(11), and breast cancer (12). Profiling of low-molecular-mass proteins/peptides (<20 kDa) revealed that the relative abundances of five peptides with specific m/z values in the SELDI-TOF spectrum formed a unique "proteomic signature" for the highly sensitive and specific diagnosis of ovarian cancer (11). Similarly, profiling of low-molecular-mass proteins/peptides demonstrated that nine low-molecular-mass peptides with m/z values <10 000 formed a proteomic signature for the detection of prostate cancer with a sensitivity of 83% (10).
Recently we showed that artificial neural networks (ANNs) trained with the serum concentrations of six serologic protein markers, all of which have molecular masses >20 kDa, could improve the identification of HCC (14). It has also been shown that, coupled with significant analysis of microarray (SAM) for gene filtering, ANNs can distinguish between global gene expression profiles of Barrett esophagus and esophageal cancer (15). We have now extended these observations and the proteomic signature concept to the proteomic analysis of serum proteins with molecular masses up to 200 kDa.
The objective of this study was to determine whether comprehensive proteomic profiling (molecular mass, 0.5–200 kDa) of serum coupled with bioinformatic analysis methods originally designed for gene expression data could identify a proteomic signature for effectively differentiating HCC from CLD. To obtain comprehensive proteomic profiles of sera from 0.5 to 200 kDa, we used an automated high-throughput, multidimensional strategy involving anion-exchange fractionation, two different ProteinChip surfaces, and the SELDI ProteinChip Biomarker System. For bioinformatic analyses, we used SAM for protein filtering, two-way hierarchical clustering, and ANN. As a proof of principle, our results show that the multidimensional ProteinChip analysis strategy permits comprehensive serum proteomic profiling. It allows the identification of tumor-specific proteomic signatures for differentiation of HCC from CLD with high sensitivity and specificity and may even permit the identification of HCC subtypes.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
anion-exchange fractionation (first-dimension separation)
Anion-exchange fractionation can be regarded as a replacement of the first-dimensional separation, isoelectric focusing, in the 2D-PAGE technology. Both technologies separate proteins on the basis of their pI values. Q Ceramic HyperDF ion-exchange resin (Biosepra SA) was used to fractionate the samples. All procedures were carried out by a Biomek 2000 Automation Workstation (Beckman Coulter). We added 125 µL of ion-exchange resin (50% suspension) to each well of a 96-well filter plate on the Biomek 2000 Automation Workstation and washed the wells three times with 150 µL of U1 solution (1 mol/L urea, 2 mL/L CHAPS, 50 mmol/L Tris-HCl, pH 9). We then mixed 20 µL of serum with 30 µL of U9 solution (9 mol/L urea, 20 mL/L CHAPS, 50 mmol/L Tris-HCl, pH 9). The urea-treated serum was transferred to each well containing ion-exchange resin and mixed on a platform shaker at 4 °C for 30 min. The flow-through fraction was then collected by suction. The proteins were first desorbed from the resin by washing with 100 µL of 50 mmol/L Tris-HCl, pH 9, containing 1 mL/L n-octyl-ß-D-glucopyranoside (OGP) for 10 min at room temperature. The previously collected flow-through fraction was pooled with the eluant and regarded as fraction 1. The proteins on the resin were eluted by sequential washing with 200 µL (100 µL twice) of five different solutions with decreasing pH values: 100 mmol/L sodium phosphate, pH 7, containing 1 mL/L OGP; 100 mmol/L sodium acetate, pH 5, containing 1 mL/L OGP; 100 mmol/L sodium acetate, pH 4, containing 1 mL/L OGP; 50 mmol/L sodium citrate, pH 3, containing 1 mL/L OGP; and a mixture containing, per liter, 333 mL of isopropanol, 167 mL of acetonitrile, and 1 mL of trifluoroacetic acid in deionized water. The washings were collected as fractions 2, 3, 4, 5, and 6, respectively.
seldi analysis of serum fractions (second- and third-dimensional analyses)
Each serum fraction was analyzed on immobilized metal ion affinity capture (IMAC3) ProteinChip Arrays (Ciphergen), loaded with copper, and weak cationic exchange (WCX2) ProteinChip Arrays (Ciphergen) in duplicate. Different ProteinChip surfaces (second dimension) helped to identify very low abundance proteins. The IMAC3 copper and WCX2 ProteinChip surfaces preferentially retain different groups of proteins according to their physiochemical properties (9). The IMAC3 copper and WCX2 arrays were equilibrated twice with 150 µL of binding buffer (0.5 mol/L NaCl buffered with 100 mmol/L sodium phosphate, pH 7, for IMAC3 copper arrays and with 100 mmol/L sodium acetate, pH 4, for WCX2 arrays). Each fraction was diluted in the appropriate binding buffer (5-fold for IMAC copper arrays; 10-fold for WCX2 arrays). We applied 100 µL of the diluted fraction to each ProteinChip Array and incubated the array with shaking at room temperature for 30 min. After the incubation, each array was washed three times with binding buffer and rinsed twice with deionized water. After the arrays were air-dried, we added sinapinic acid matrix in 500 mL/L acetonitrile–5 mL/L trifluoroacetic acid to each array. The ProteinChip Arrays were read on a ProteinChip PBS II reader of the ProteinChip Biomarker System to measure the masses and intensities of the protein peaks (Ciphergen). The mass spectrometric analysis (third dimension) with the ProteinChip PBS II reader can be regarded as a replacement of the second dimensional separation, sodium dodecyl sulfate-PAGE, in the 2D-PAGE technology. Both technologies separate proteins on the basis of their molecular masses. We averaged 235 laser shots for each array with masses ranging from 0.5 to 200 kDa. All mass spectra were normalized to have the same total ion current. The CVs of the peak intensities were <15% (manufacturer’s information). Common protein peaks were picked by the Biomarker WizardTM function of the ProteinChip Software (Ciphergen). Average peak intensities of the duplicate measurements were used in later bioinformatic analyses.
sam protein filtering
The protein filtering process was performed with SAM (17). Given the group identities, SAM was used to compare the peak intensities of the proteomic features between the cancer (38 HCC cases) and control (20 CLD cases with AFP <500 µg/L) groups and to identify the proteomic features that were significantly different at a median false significant value <0.000005. The control group was defined as "1", whereas the tumor group was defined as "2". "Two classes unpaired data" was selected as the data type, and permutations were performed 1000 times.
two-way hierarchical clustering analysis
The significant differential proteomic features that were identified from SAM were subjected to two-way hierarchical clustering analysis. Before the analysis, the median intensity of each significant proteomic feature was normalized to equal 1, and then all the normalized intensity data were subtracted by 1. After this data processing, the intensity data would be positive when they were greater than the median intensity and negative when they were lower. The processed data for the significant proteomic features and the serum samples were subjected to two-way hierarchical clustering analysis with use of Cluster and TreeView (18). Spearman rank correlation was used to calculate the distance, and complete linkage clustering was performed.
ann model development
An ANN model was developed as described previously (14). The ANN algorithm applies artificial intelligence to classification, pattern recognition, and prediction (14)(15). An ANN model consists of processing elements (neurones), which are organized in layers. From a training data set, an ANN model can "learn" the association patterns between the input variables and outcomes and then apply these patterns to new cases. The ANN model was developed with EasyNN (Ver. 8.1; Stephen Wolstenholme). The development method was of the feed-forward type, and the networks were trained by weighted back-propagation. Both learning rate and momentum were optimized automatically by the software. The ANN model was composed of three layers, one input layer, one hidden layer, and one output layer. There were seven nodes in the middle, hidden layer. The input variables for the development of the ANN model were the relative levels of the significant differential proteomic features, whereas the output variable was the diagnostic score (range, 0–1.0000) of each case. During training of the ANN model, the diagnostic scores were defined as 0.0000 and 1.0000 for the CLD cases and HCC cases, respectively. The training was stopped when all output errors were <0.02. The maximum number of training cycles was restricted to 300 to prevent overtraining. Using the developed ANN model, we performed 10-fold cross-validation to estimate the ANN diagnostic scores for each HCC and CLD case. The whole data set (38 HCC and 20 CLD cases) was randomly divided into 10 subsets of equal size. The model was then trained 10 times, each time leaving out one of the subsets from training but using the omitted subset as an independent blinded test set to assess the performance of the model. ROC curves were constructed by calculating the sensitivities and specificities of tests at different cutoff points of the ANN diagnostic scores for differentiating HCC cases from CLD cases. The developed ANN model was also used to classify the unseen cases, including the two cases with resected HCC (RHCC1 and RHCC2), the CLD case with AFP = 905 µg/L (CLD21), and the unseen pooled sera (HCCP1, HCCP2, and CLDP1).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. Because the chemistries of the binding surfaces of the IMAC3 copper ProteinChip Array and the WCX2 copper ProteinChip Array were different, different groups of polypeptides were preferentially bound to the IMAC3 copper ProteinChip Array or the WCX2 ProteinChip Array. As illustrated in Fig. 1
, different specific proteomic features were found within the range of m/z 8500–9500 when different types of ProteinChip Arrays were used. Among these 250 potential markers, 109 and 141 were identified with the IMAC3 copper ProteinChip Arrays and the WCX2 ProteinChip Arrays, respectively.
|
|
identification of hcc and its subtypes by two-way hierarchical clustering
We performed two-way hierarchical clustering analysis (two-dimensional complete linkage), based on the 250 differential proteomic features. The clustering algorithm separated HCC and CLD cases into two major clusters (Fig. 2
). Most of the typical CLD cases with AFP <500 µg/L (19 of 20 cases) and case CLD21 (with increased serum AFP) were grouped together. The pooled serum samples HCCP1, HCCP2, and CLDP1 were also aligned correctly together with the HCC and CLD cases. The resected case with residual tumor (RHCC1) was aligned with the HCC cases, whereas the complete remission case (RHCC2) was aligned with the CLD cases, indicating its nontumor nature. This also suggests that the identified serum proteomic signatures in the HCC patients, which disappeared when a tumor was completely removed, were tumor specific.
|
The majority of HCC cases with advanced disease (stage III and IV) were clustered together, forming two HCC subgroups, A and B (Fig. 2
and Table 2
). There were significantly more cases with lymph node invasion or distant metastasis in these two subgroups (Fig. 2
and Table 2
). Our results therefore indicate that advanced HCCs and metastatic tumors could be identified on the basis of the serum proteomic profiles, without knowledge of the clinical information of the cases, and suggest the potential of using comprehensive serum proteomic profiling to classify HCC into different subtypes.
|
ann model for differentiation of hcc from cld
To integrate the clinical values of the 250 differential proteomic features for discrimination of HCC from CLD, we developed an ANN model from these proteomic features of the cancer and control cases for diagnostic score calculation. We performed 10-fold cross-validation analysis to estimate the diagnostic performance of the resulting ANN model. The diagnostic scores of the HCC cases [mean (SD), 0.922 (0.358)] were significantly higher (P <0.0005, Mann-Whitney test) than those of CLD cases [0.177 (0.246)]. ROC curve analyses showed that the ANN diagnostic score was useful in differentiating HCC and CLD cases regardless of serum AFP concentrations (Fig. 3
). The area under ROC curve was 0.914 [95% confidence interval, 0.823–1.006; P <0.0005] for all cases (Fig. 3A
), whereas the area was 0.954 (95% confidence interval, 0.881–1.027; P <0.0005) for cases with nondiagnostic serum AFP concentrations (<500 µg/L; Fig. 3B
).
|
At an ANN diagnostic score cutoff of 0.8500, the mean (SE) sensitivity and specificity were 92 (4)% (35 of 38 HCC cases) and 90 (7)% (18 of 20 control cases), respectively. For HCC cases with nondiagnostic AFP, 95 (5)% of HCC cases (19 of 20 cases) were correctly classified. Furthermore, case CLD21 [mean (SE) ANN diagnostic score, 0.0186 (0.0017)], which on the basis of its AFP concentration was falsely positive for HCC, and all three pooled sera, HCCP1 [ANN diagnostic score, 0.9989 (0.0001)], HCCP2 [ANN diagnostic score, 0.9993 (0.0001)], and CLDP1 [ANN diagnostic score, 0.0028 (0.0002)], were also correctly classified. Furthermore, the resected case with residual tumor [RHCC1; ANN diagnostic score, 0.9989 (0.0001)] and the complete remission case [RHCC2; ANN diagnostic score, 0.1117 (0.0132)] were also correctly classified as a HCC case and a noncancer case, respectively.
A negative control ANN model was developed using 250 randomly selected insignificant proteomic features as input variables and was examined by 10-fold cross-validation. As expected, the ROC curve analysis showed that the negative control ANN model was much less useful in differentiating between HCC and CLD cases (0.780; 95% confidence interval, 0.643–0.916; P = 0.001). At an ANN diagnostic score cutoff of 0.8500, the mean (SE) sensitivity and specificity of the negative control ANN model were only 68 (8)% (26 of 38 HCC cases) and 65 (11)% (13 of 20 control cases), respectively. Our results strongly indicate that (a) an ANN algorithm can be used to combine the diagnostic values of the 250 differential proteomic features, (b) combined use of the 250 differential proteomic features allows the detection of HCC at a sensitivity and a specificity of 
90%, and (c) the resulting ANN model can identify HCC cases with nondiagnostic AFP concentrations at a sensitivity of 95%.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Various high-throughput technologies have been developed for measurement of the expression profiles of numerous genes simultaneously and have been used to study gene expression patterns in various cancer tissues, including leukemia, colon carcinoma, breast cancer, and HCC (19)(20). On the basis of such gene expression profiles, tumor subtypes with distinctive prognostic features have been identified (21). Unique gene expression profiles have been observed in metastatic HCC (22). Because proteins are gene products, it is logical to expect that specific proteomic profiles/patterns may be also associated with specific types or even subtypes of tumors. In the present study, comprehensive serum proteomic profiling enabled us to identify HCC patient subgroups with significantly more advanced HCCs and metastatic tumors, without knowledge of the clinical information. This suggests that protein expression profiles may differ between early and advanced HCCs and between metastatic and nonmetastatic tumors, leading to different serum proteomic profiles. Our study therefore suggests that comprehensive serum proteomic profiling can classify cancers into different subtypes.
The identification of cancer subtypes is important in estimating prognosis and in the selection of patients who are likely to be responsive to a particular treatment. Comprehensive serum proteomic profiling may be particularly invaluable when tumor tissue is not available for histologic examination. Furthermore, the correct classification of two resected HCC cases, one with residual tumor and one without, as a HCC case and a noncancer case suggests the possibility of using comprehensive serum proteomic profiling for monitoring treatment response and tumor recurrence. In the future, proteomic profiling, gene expression profiling, and single-nucleotide polymorphism profiling may complement each other in unraveling an individual’s response to therapy (23)(24).
The SELDI ProteinChip Biomarker System has been used previously to identify single or multiple tumor-specific biomarkers in various cancers, including prostate (10), ovarian(11), and breast cancer (12). When peptides and proteins below m/z 20 000 (i.e., molecular mass <20 kDa) were analyzed, increased concentrations of five peptides with specific m/z values were found in the sera of patients with ovarian cancer (11). These five peptides formed a proteomic signature, which allowed detection of ovarian cancer at 100% sensitivity and 95% specificity. Similarly, profiling of low-molecular-mass proteins/peptides identified a proteomic signature comprising nine low-molecular-mass peptides with m/z values <10 000 (i.e., molecular mass <10 kDa) for the detection of prostate cancer at a sensitivity of 83% (10). Our multidimensional strategy allowed us to identify >200 potential tumor markers in the sera of patients with HCC. The number of potential tumor markers thus may have been underestimated in previous studies that used SELDI-TOF MS technology. Consistent with the studies of ovarian (11) and prostate cancer (10), we also found tumor-specific peptides/polypeptides with molecular masses <20 kDa. Furthermore, our study indicates that there are significant numbers of tumor-specific proteins with molecular masses >20 kDa. More experiments are needed to identify the nature of the 250 proteins. Nevertheless, because each protein has a unique m/z value, these proteins can be unambiguously detected and quantified in patient sera, even without knowledge of their protein identities. These markers form tumor-specific proteomic signatures for the detection of HCC at a sensitivity and specificity of 90%.
The clinical applications of proteomics have been hindered by limitations of both proteomic technologies (4)(23) and bioinformatic tools (13)(25). In this study, we have shown that a multidimensional strategy involving anion-exchange fractionation and the SELDI ProteinChip Biomarker System allows comprehensive proteomic profiling. We have also shown that cDNA/oligonucleotide microarray data analysis tools, including SAM gene filtering, two-way hierarchical clustering, and ANNs, are useful in identifying proteomic signatures associated with HCC.
Compared with gene expression microarray data, the computational analysis tools for proteomic data are still underdeveloped (13)(25). Genetic algorithms(11), self-organizing cluster analysis (11), and discrete wavelet transform analysis (13) have all been used to compare ProteinChip Array data between the control and test groups, but the results generated from all these analytical tools are difficult to interpret. The present study shows that comprehensive ProteinChip Array data can be readily analyzed with the algorithms commonly used for gene expression data analysis, including SAM gene filtering and two-way hierarchical clustering. SAM uses repeated permutations of the data to determine whether the expression of any gene/protein is significantly related to a response. The cutoff for significance is determined by the user based on false discovery rate. The false discovery rate is the number of falsely called genes/proteins divided by the number of differential genes/proteins in the original data. In other words, the false discovery rate provides information on the percentage of nonsignificant genes/proteins one can expect to find in the result list of differentially expressed genes/proteins (17). This allows identification of proteomic features that are "genuinely" significantly different between HCC patients and control groups and avoids multiple testing problems. The two-way hierarchical clustering dendrogram allows examination of the correlations among cases or proteomic features. Furthermore, our results indicate that the combination of ANN and data filtering with SAM is not only useful in distinguishing between global gene expression profiles (15), but also in distinguishing between comprehensive proteomic profiles of sera from cancer patients and individuals without cancer. The computation of a diagnostic score by an ANN model provides an easy way to interpret tumor-specific proteomic signatures for the detection of HCC.
In conclusion, the present study provides the first evidence that comprehensive proteomic profiling from 0.5 to 200 kDa can be performed by the SELDI ProteinChip Biomarker System and can be analyzed and interpreted in a way similar to gene expression data. Although future studies are needed for less biased elucidation of the diagnostic accuracy, this study suggests that tumor-specific proteomic signatures are present in the sera of patients with HCC and have potential as a clinical tool for the detection or classification of individual cancers or even tumor subtypes with high sensitivity and specificity. Similar strategies may be applied to analyze the proteomes of tumor tissues or biological fluids from patients with other cancer types.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
-fetoprotein; HCC, hepatocellular carcinoma; CLD, chronic liver disease; 2D-PAGE, two-dimensional polyacrylamide gel electrophoresis; MS, mass spectrometry; SELDI, surface-enhanced laser desorption/ionization; TOF, time-of-flight; ANN, artificial neural network; SAM, significant analysis of microarray; and OGP, n-octyl-ß-D-glucopyranoside. 
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-fetoprotein estimation in the diagnosis and management of hepatocellular carcinoma. Clin Liver Dis 2001;5:145-159.[CrossRef][Medline]
[Order article via Infotrieve]
fetoprotein may permit preclinical diagnosis of malignant change in patients with chronic liver disease. Br J Cancer 1997;75:236-240.[ISI][Medline]
[Order article via Infotrieve]
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  R. W. C. Pang, J. W. Joh, P. J. Johnson, M. Monden, T. M. Pawlik, and R. T. P. Poon Biology of Hepatocellular Carcinoma Ann. Surg. Oncol., April 1, 2008; 15(4): 962 - 971. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. T. Zinkin, F. Grall, K. Bhaskar, H. H. Otu, D. Spentzos, B. Kalmowitz, M. Wells, M. Guerrero, J. M. Asara, T. A. Libermann, et al. Serum Proteomics and Biomarkers in Hepatocellular Carcinoma and Chronic Liver Disease Clin. Cancer Res., January 15, 2008; 14(2): 470 - 477. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. Bennett Untargeted Metabolomic Analysis Hits the Target Clin. Chem., December 1, 2007; 53(12): 2037 - 2039. [Full Text] [PDF]
|
|
|
|
 |
|
 W. Sun, B. Xing, Y. Sun, X. Du, M. Lu, C. Hao, Z. Lu, W. Mi, S. Wu, H. Wei, et al. Proteome Analysis of Hepatocellular Carcinoma by Two-dimensional Difference Gel Electrophoresis: Novel Protein Markers in Hepatocellular Carcinoma Tissues Mol. Cell. Proteomics, October 1, 2007; 6(10): 1798 - 1808. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. K.T. Kam, T. C.W. Poon, H. L.Y. Chan, N. Wong, A. Y. Hui, and J. J.Y. Sung High-Throughput Quantitative Profiling of Serum N-Glycome by MALDI-TOF Mass Spectrometry and N-Glycomic Fingerprint of Liver Fibrosis Clin. Chem., July 1, 2007; 53(7): 1254 - 1263. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. H. Otu, H. Can, D. Spentzos, R. G. Nelson, R. L. Hanson, H. C. Looker, W. C. Knowler, M. Monroy, T. A. Libermann, S. A. Karumanchi, et al. Prediction of Diabetic Nephropathy Using Urine Proteomic Profiling 10 Years Prior to Development of Nephropathy Diabetes Care, March 1, 2007; 30(3): 638 - 643. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Aivado, D. Spentzos, U. Germing, G. Alterovitz, X.-Y. Meng, F. Grall, A. A. N. Giagounidis, G. Klement, U. Steidl, H. H. Otu, et al. From the cover: Serum proteome profiling detects myelodysplastic syndromes and identifies CXC chemokine ligands 4 and 7 as markers for advanced disease PNAS, January 23, 2007; 104(4): 1307 - 1312. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. T.K. Pang, T. C.W. Poon, K.C. A. Chan, N. L.S. Lee, R. W.K. Chiu, Y.-K. Tong, R. M.Y. Wong, S. S.C. Chim, S. M. Ngai, J. J.Y. Sung, et al. Serum Proteomic Fingerprints of Adult Patients with Severe Acute Respiratory Syndrome Clin. Chem., March 1, 2006; 52(3): 421 - 429. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. L. Hortin and B. Meilinger Cross-Reactivity of Amino Acids and Other Compounds in the Biuret Reaction: Interference with Urinary Peptide Measurements Clin. Chem., August 1, 2005; 51(8): 1411 - 1419. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Kim, P. Sun, Y.-W. Lam, P. Troncoso, A. L. Sabichi, R. J. Babaian, L. L. Pisters, C. A. Pettaway, C. G. Wood, S. M. Lippman, et al. Changes in Serum Proteomic Patterns by Presurgical {alpha}-Tocopherol and L-Selenomethionine Supplementation in Prostate Cancer Cancer Epidemiol. Biomarkers Prev., July 1, 2005; 14(7): 1697 - 1702. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Meyerson and D. Carbone Genomic and Proteomic Profiling of Lung Cancers: Lung Cancer Classification in the Age of Targeted Therapy J. Clin. Oncol., May 10, 2005; 23(14): 3219 - 3226. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Le, K. Chi, S. Tyldesley, S. Flibotte, D. L. Diamond, M. A. Kuzyk, and M. D. Sadar Identification of Serum Amyloid A as a Biomarker to Distinguish Prostate Cancer Patients with Bone Lesions Clin. Chem., April 1, 2005; 51(4): 695 - 707. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Malik, M. D. Ward, S. K. Gupta, M. W. Trosset, W. E. Grizzle, B.-L. Adam, J. I. Diaz, and O. J. Semmes Serum Levels of an Isoform of Apolipoprotein A-II as a Potential Marker for Prostate Cancer Clin. Cancer Res., February 1, 2005; 11(3): 1073 - 1085. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. C.W. Poon, A. Y. Hui, H. L.Y. Chan, I. L. Ang, S. M. Chow, N. Wong, and J. J.Y. Sung Prediction of Liver Fibrosis and Cirrhosis in Chronic Hepatitis B Infection by Serum Proteomic Fingerprinting: A Pilot Study Clin. Chem., February 1, 2005; 51(2): 328 - 335. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. J. Semmes, Z. Feng, B.-L. Adam, L. L. Banez, W. L. Bigbee, D. Campos, L. H. Cazares, D. W. Chan, W. E. Grizzle, E. Izbicka, et al. Evaluation of Serum Protein Profiling by Surface-Enhanced Laser Desorption/Ionization Time-of-Flight Mass Spectrometry for the Detection of Prostate Cancer: I. Assessment of Platform Reproducibility Clin. Chem., January 1, 2005; 51(1): 102 - 112. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. K. Drake, R. A.R. Bowen, A. T. Remaley, and G. L. Hortin Potential Interferences from Blood Collection Tubes in Mass Spectrometric Analyses of Serum Polypeptides Clin. Chem., December 1, 2004; 50(12): 2398 - 2401. [Full Text] [PDF]
|
|
|
|
 |
|
 B. A. Wetmore and B. A. Merrick Invited Review: Toxicoproteomics: Proteomics Applied to Toxicology and Pathology Toxicol Pathol, October 1, 2004; 32(6): 619 - 642. [Abstract] [PDF]
|
|
|
|
|
|
T. C.W. Poon, K.C. A. Chan, P.-C. Ng, R. W.K. Chiu, I. L. Ang, Y.-k. Tong, E. K.O. Ng, F. W.T. Cheng, A. M. Li, E. K.L. Hon, et al. Serial Analysis of Plasma Proteomic Signatures in Pediatric Patients with Severe Acute Respiratory Syndrome and Correlation with Viral Load Clin. Chem., August 1, 2004; 50(8): 1452 - 1455. [Full Text] [PDF]
|
|
|
|
|
|
S. G. Soltys, Q.-T. Le, G. Shi, R. Tibshirani, A. J. Giaccia, and A. C. Koong The Use of Plasma Surface-Enhanced Laser Desorption/Ionization Time-of-Flight Mass Spectrometry Proteomic Patterns for Detection of Head and Neck Squamous Cell Cancers Clin. Cancer Res., July 15, 2004; 10(14): 4806 - 4812. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z. Xiao, B. T. Luke, G. Izmirlian, A. Umar, P. M. Lynch, R. K. S. Phillips, S. Patterson, T. P. Conrads, T. D. Veenstra, P. Greenwald, et al. Serum Proteomic Profiles Suggest Celecoxib-Modulated Targets and Response Predictors Cancer Res., April 15, 2004; 64(8): 2904 - 2909. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Koopmann, Z. Zhang, N. White, J. Rosenzweig, N. Fedarko, S. Jagannath, M. I. Canto, C. J. Yeo, D. W. Chan, and M. Goggins Serum Diagnosis of Pancreatic Adenocarcinoma Using Surface-Enhanced Laser Desorption and Ionization Mass Spectrometry Clin. Cancer Res., February 1, 2004; 10(3): 860 - 868. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
|
