Pooling of Case Specimens to Create Standard Serum Sets for Screening Cancer Biomarkers

Estimating the AUC with a Graphical Lasso Method for High-dimensional Biomarkers with LOD

This manuscript estimates the area under the receiver operating characteristic curve (AUC) of combined biomarkers in a high-dimensional setting. We propose a penalization approach to the inference of precision matrices in the presence of the limit of detection. A new version of expectation-maximization algorithm is then proposed for the penalized likelihood, with the use of numerical integration and the graphical lasso method. The estimated precision matrix is then applied to the inference of AUCs. The proposed method outperforms the existing methods in numerical studies. We apply the proposed method to a data set of brain tumor study. The results show a higher accuracy on the estimation of AUC compared with the existing methods.

Large-scale Identification of N-linked Intact Glycopeptides in Human Serum using HILIC Enrichment and Spectral Library Search

Large-scale identification of N-linked intact glycopeptides by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) in human serum is challenging because of the wide dynamic range of serum protein abundances, the lack of a complete serum N-glycan database and the existence of proteoforms. In this regard, a spectral library search method was presented for the identification of N-linked intact glycopeptides from N-linked glycoproteins in human serum with target-decoy and motif-specific false discovery rate (FDR) control. Serum proteins were firstly separated into low-abundance and high-abundance proteins by acetonitrile (ACN) precipitation. After digestion, the N-linked intact glycopeptides were enriched by hydrophilic interaction liquid chromatography (HILIC) and a portion of the enriched N-linked intact glycopeptides were processed by Peptide-N-Glycosidase F (PNGase F) to generate N-linked deglycopeptides. Both N-linked intact glycopeptides and deglycopeptides were analyzed by LC-MS/MS. From N-linked deglycopeptides data sets, 764 N-linked glycoproteins, 1699 N-linked glycosites and 3328 unique N-linked deglycopeptides were identified. Four types of N-linked glycosylation motifs (NXS/T/C/V, X≠P) were used to recognize the N-linked deglycopeptides. The spectra of these N-linked deglycopeptides were utilized for N-linked deglycopeptides library construction and identification of N-linked intact glycopeptides. A database containing 739 N-glycan masses was constructed and utilized during spectral library search for the identification of N-linked intact glycopeptides. In total, 526 N-linked glycoproteins, 1036 N-linked glycosites, 22,677 N-linked intact glycopeptides and 738 N-glycan masses were identified under 1% FDR, representing the most in-depth serum N-glycoproteome identified by LC-MS/MS at N-linked intact glycopeptide level.

A tumor specific antibody to aid breast cancer screening in women with dense breast tissue

Screening for breast cancer has predominantly been done using mammography. Unfortunately, mammograms miss 50% cancers in women with dense breast tissue. Multi-modal screenings offer the best chance of enhancing breast cancer screening effectiveness. We evaluated the use of TAB004, an antibody that recognizes the tumor form of the glycoprotein MUC1 (tMUC1), to aid early detection of breast cancer. Our experimental approach was to follow tMUC1 from the tissue into circulation. We found that 95% of human breast cancer tissues across all subtypes stained positive for TAB004. In breast cancer cell lines, we showed that the amount of tMUC1 released from tumor cells is proportional to the cell's tMUC1 expression level. Finally, we showed that TAB004 can be used to assess circulating tMUC1 levels, which when monitored in the context of cancer immunoediting, can aid earlier diagnosis of breast cancer regardless of breast tissue density. In a blinded pilot study with banked serial samples, tMUC1 levels increased significantly up to 2 years before diagnosis. Inclusion of tMUC1 monitoring as part of a multi-modal screening strategy may lead to earlier stage diagnosis of women whose cancers are missed by mammography.

Creation of a Human Secretome: A Novel Composite Library of Human Secreted Proteins: Validation Using Ovarian Cancer Gene Expression Data and a Virtual Secretome Array

PURPOSE

To generate a comprehensive "Secretome" of proteins potentially found in the blood and derive a virtual Affymetrix array. To validate the utility of this database for the discovery of novel serum-based biomarkers using ovarian cancer transcriptomic data.

EXPERIMENTAL DESIGN

The secretome was constructed by aggregating the data from databases of known secreted proteins, transmembrane or membrane proteins, signal peptides, G-protein coupled receptors, or proteins existing in the extracellular region, and the virtual array was generated by mapping them to Affymetrix probeset identifiers. Whole-genome microarray data from ovarian cancer, normal ovarian surface epithelium, and fallopian tube epithelium were used to identify transcripts upregulated in ovarian cancer.

RESULTS

We established the secretome from eight public databases and a virtual array consisting of 16,521 Affymetrix U133 Plus 2.0 probesets. Using ovarian cancer transcriptomic data, we identified candidate blood-based biomarkers for ovarian cancer and performed bioinformatic validation by demonstrating rediscovery of known biomarkers including CA125 and HE4. Two novel top biomarkers (FGF18 and GPR172A) were validated in serum samples from an independent patient cohort.

CONCLUSIONS

We present the secretome, comprising the most comprehensive resource available for protein products that are potentially found in the blood. The associated virtual array can be used to translate gene-expression data into cancer biomarker discovery. A list of blood-based biomarkers for ovarian cancer detection is reported and includes CA125 and HE4. FGF18 and GPR172A were identified and validated by ELISA as being differentially expressed in the serum of ovarian cancer patients compared with controls.

Construction and analysis of the NCI-EDRN breast cancer reference set for circulating markers of disease

BACKGROUND

Many circulating biomarkers have been reported for the diagnosis of breast cancer, but few, if any, have undergone rigorous credentialing using prospective cohorts and blinded evaluation.

METHODS

The NCI Early Detection Research Network (EDRN) has created a prospective, multicenter collection of plasma and serum samples from 832 subjects designed to evaluate circulating biomarkers for the detection and diagnosis of breast cancer. These samples are available to investigators who wish to evaluate their biomarkers using a set of blinded samples. The breast cancer reference set is composed of blood samples collected using a standard operating procedure at four U.S. medical centers from 2008 to 2010 from women undergoing either tissue diagnosis for breast cancer or routine screening mammography. The reference set contains samples from women with incident invasive cancer (n = 190), carcinoma in situ (n = 55), benign pathology with atypia (n = 63), benign disease with no atypia (n = 231), and women with no evidence of breast disease by screening mammography (BI-RADS 1 or 2, n = 276). Using a subset of plasma samples (n = 505) from the reference set, we analyzed 90 proteins by multiplexed immunoassays for their potential utility as diagnostic markers.

RESULTS

We found that none of these markers is useful for distinguishing cancer from benign controls. However, elevated CA-125 does appear to be a candidate marker for estrogen receptor-negative cancers.

CONCLUSIONS

Markers that can distinguish benign breast conditions from invasive cancer have not yet been found.

IMPACT

Availability of prospectively collected samples should improve future validation efforts.

Determination of CA-125 levels in the serum, cervical and vaginal secretions, and endometrium in Chinese women with precancerous disease or endometrial cancer

Background

Serum CA-125 has been used as a biomarker of gynecological tumors. In this study, we investigated the CA-125 levels in cervical and vaginal secretions from Chinese patients with endometrial polyps, hyperplasia and carcinoma in comparison with those in endometrium and serum.

Material/Methods

An electro-chemiluminescent immunoassay was utilized to determine the levels of CA-125 in 51 healthy Chinese women and 97 patients with polyps, hyperplasia or endometrial cancer. An immunohistochemistry method was used to detect endometrial CA-125 expression in 242 subjects.

Results

Our study demonstrated that serum CA-125 levels were much lower than those in cervical and vaginal secretions in healthy and diseased women. The levels of CA-125 in serum, and cervical and vaginal secretions were significantly increased in complex hyperplasia and endometrial cancer. The increase of CA-125 content in serum, cervical and vaginal secretions was lesser significant in grade 3 cancer than that in grade 1 and 2 cancer. Generally, serum CA-125 levels correlated with those in cervical and vaginal secretions and CA-125 content in cervical secretion correlated with that in vaginal secretion. There was only a weak CA-125 expression in normal endometrium and simple endometrial hyperplasia. There was a significant difference in CA-125 expression among patients with pathological grade 1, 2 and 3 of endometrial carcinoma.

Conclusions

Endometrial CA-125 expression together with its levels in the serum and cervical and vaginal secretions can be used as a potential biomarker in the diagnosis of precancerous diseases and endometrial carcinoma

Prediction of a high-risk group based on postoperative nadir CA-125 levels in patients with advanced epithelial ovarian cancer

Objective

We aimed to determine the ideal cut-off of nadir serum CA-125 level for prediction of progression free survival.

Methods

Among 267 patients who achieved complete remission after chemotherapy, the correlation between nadir CA-125 and progression free survival were compared among the subgroups classified according to the distribution of CA-125. The diagnostic odds ratio and area under the receiver operator characteristics curve were compared at various cut-off points.

Results

The nadir CA-125 levels did not have prognostic value under 12 U/mL (to 75 percentile). In contrast, they were significantly correlated with progression free survival only when the CA-125 level was greater than 12, which was 75 percentile (p=0.034). In predicting progression free survival <6 and 12 months, the cut-off value of 18 (90 percentile) showed superior diagnostic performance over 10 or 12 U/mL. Compared with patients who showed nadir levels between 0 and 12 U/mL (0 to 75 percentile), those with nadir >18 U/mL showed a hazard ratio of 2.85 (95% confidence interval, 1.70 to 4.76; p<0.001); patients with nadir levels between 18 and 12 U/mL showed a the hazard ratio of 1.68 (95% confidence interval, 1.11 to 2.56; p=0.015) compared with those whose nadir levels were under 12 U/mL.

Conclusion

The predictive power of the traditional cut-off of 10 U/mL to classify a risk group or to identify high risk patients was unsatisfactory. The optimal diagnostic performance was observed at the cut-off of 18 U/mL and this can be proposed to dichotomize cut-off values to predict outcomes among individual patients.

SGO White Paper on ovarian cancer: etiology, screening and surveillance

Ovarian cancer is a heterogeneous, rapidly progressive, highly lethal disease of low prevalence. The etiology remains poorly understood. Numerous risk factors have been identified, the most prominent involving an inherited predisposition in 10% of cases. Women with germline mutations associated with Hereditary Breast/Ovarian Cancer and Lynch syndromes have dramatically elevated risks (up to 46% and 12%, respectively). Risk-reducing salpingo-oophorectomy is the best method to prevent ovarian cancer in these high-risk women. Significant risk reduction is also seen in the general population who use oral contraceptives. Since up to 89% patients with early-stage disease have symptoms prior to diagnosis, increased awareness of the medical community may facilitate further workup in patients who otherwise would have had a delay. Despite enormous effort, there is no proof that routine screening for ovarian cancer in either the high-risk or general populations with serum markers, sonograms, or pelvic examinations decreases mortality. Further evaluation is needed to determine whether any novel biomarkers, or panels of markers, have clinical utility in early detection. Prospective clinical trials have to be designed and completed prior to offering of any of these new diagnostic tests. CA125 is currently the only biomarker recommended for monitoring of therapy as well as detection of recurrence. This commentary provides an overview on the background, screening and surveillance of ovarian cancer.

Levels of metabolites of organophosphate pesticides, phthalates, and bisphenol A in pooled urine specimens from pregnant women participating in the Norwegian Mother and Child Cohort Study (MoBa)

Concerns about reproductive and developmental health risks of exposure to organophosphate (OP) pesticides, phthalates, and bisphenol A (BPA) among the general population are increasing. Six dialkyl phosphate (DAP) metabolites, 3,5,6-trichloro-2-pyridinol (TCPy), BPA, and fourteen phthalate metabolites were measured in 10 pooled urine samples representing 110 pregnant women who participated in the Norwegian Mother and Child Birth Cohort (MoBa) study in 2004. Daily intakes were estimated from urinary data and compared with reference doses (RfDs) and daily tolerable intakes (TDIs). The MoBa women had a higher mean BPA concentration (4.50 microg/L) than the pregnant women in the Generation R Study (Generation R) in the Netherlands and the National Health and Nutrition Examination Survey (NHANES) in the United States. The mean concentration of total DAP metabolites (24.20 microg/L) in MoBa women was higher than that in NHANES women but lower than that in Generation R women. The diethyl phthalate metabolite mono-ethyl phthalate (MEP) was the dominant phthalate metabolite in all three studies, with the mean concentrations of greater than 300 microg/L. The MoBa and Generation R women had higher mean concentrations of mono-n-butyl phthalate (MnBP) and mono-isobutyl phthalate (MiBP) than the NHANES women. The estimated average daily intakes of BPA, chlorpyrifos/chlorpyrifos-methyl and phthalates in MoBa (and the other two studies) were below the RfDs and TDIs. The higher levels of metabolites in the MoBa participants may have been from intake via pesticide residues in food (organophosphates), consumption of canned food, especially fish/seafood (BPA), and use of personal care products (selected phthalates).

Do serum biomarkers really measure breast cancer?

Background

Because screening mammography for breast cancer is less effective for premenopausal women, we investigated the feasibility of a diagnostic blood test using serum proteins.

Methods

This study used a set of 98 serum proteins and chose diagnostically relevant subsets via various feature-selection techniques. Because of significant noise in the data set, we applied iterated Bayesian model averaging to account for model selection uncertainty and to improve generalization performance. We assessed generalization performance using leave-one-out cross-validation (LOOCV) and receiver operating characteristic (ROC) curve analysis.

Results

The classifiers were able to distinguish normal tissue from breast cancer with a classification performance of AUC = 0.82 ± 0.04 with the proteins MIF, MMP-9, and MPO. The classifiers distinguished normal tissue from benign lesions similarly at AUC = 0.80 ± 0.05. However, the serum proteins of benign and malignant lesions were indistinguishable (AUC = 0.55 ± 0.06). The classification tasks of normal vs. cancer and normal vs. benign selected the same top feature: MIF, which suggests that the biomarkers indicated inflammatory response rather than cancer.

Conclusion

Overall, the selected serum proteins showed moderate ability for detecting lesions. However, they are probably more indicative of secondary effects such as inflammation rather than specific for malignancy.

Higher prevalence of secretory CSE1L/CAS in sera of patients with metastatic cancer

Metastatic markers are highly useful diagnostic and prognostic indicators of cancer metastasis. Herein, we report that secretory CSE1L/CAS, a cellular apoptosis susceptibility protein, is a new marker for metastatic cancer. CAS was colocalized with matrix metalloproteinase-2 in vesicles surrounding the outside of MCF-7 cell membranes, and the COOH-terminal domain of CAS was associated with matrix metalloproteinase-2-containing vesicles. Immunohistochemical staining for CAS was positive in the stroma and gland lumens of human metastatic cancer tissues. CAS was also detected in conditioned medium from B16-F10 melanoma cells and more frequently in the sera of patients with metastatic cancer than in sera from patients with primary cancer. Specifically, the prevalence of serum CAS in serum samples from 146 patients was 58.2% (32 of 55), 32.0% (8 of 25), and 12.1% (8 of 66) for patients with metastatic, invasive, and primary cancers, respectively. Our results suggest that CAS is a secretory protein associated with cancer metastasis, which may have clinical utility in metastatic cancer screening and diagnosis.

[-2]proenzyme prostate specific antigen for prostate cancer detection: a national cancer institute early detection research network validation study

PURPOSE

This study evaluated the [-2]proenzyme prostate specific antigen serum marker using a blinded reference specimen set from 3 National Cancer Institute Early Detection Research Network centers from men with an indication for prostate biopsy.

MATERIALS AND METHODS

Serum was collected before biopsy from 123 men with no prior biopsy or prostate cancer history. Specimens (cancer cases 51%, noncancer controls 49%) were selected equally from the 3 sites, and analyzed for prostate specific antigen, free prostate specific antigen, [-2]proenzyme prostate specific antigen, benign prostate specific antigen and testosterone (Beckman Coulter ACCESS(R) analyzer).

RESULTS

There was no difference in total prostate specific antigen concentrations (noncancer 6.80 +/- 5.20 ng/ml, cancer 6.94 +/- 5.12 ng/ml) among the groups. Overall %[-2]proenzyme prostate specific antigen had the greatest area under the curve (AUC 0.69) followed by percent free prostate specific antigen (AUC 0.61). For %[-2]proenzyme prostate specific antigen maximal sensitivity was 60% and specificity was 70%. A logistic regression model combining prostate specific antigen, benign prostate specific antigen, percent free prostate specific antigen, %[-2]proenzyme prostate specific antigen, [-2]proenzyme prostate specific antigen/benign prostate specific antigen and testosterone had an AUC of 0.73. In the 2 to 10 ng/ml prostate specific antigen range %[-2]proenzyme prostate specific antigen and the model had the largest AUC (0.73). The AUC for percent free prostate specific antigen was 0.53. Specificities for %[-2]proenzyme prostate specific antigen, the logistic regression model and percent free prostate specific antigen at 90% sensitivity were 41%, 32% and 18%, and at 95% sensitivity were 31%, 26% and 16%, respectively.

CONCLUSIONS

%[-2]proenzyme prostate specific antigen was the best predictor of prostate cancer detection compared to percent free prostate specific antigen, particularly in the 2 to 10 ng/ml total prostate specific antigen range. These findings provide a rationale for broader validation studies to determine whether %[-2]proenzyme prostate specific antigen alone can replace other molecular prostate specific antigen assays (such as percent free prostate specific antigen) for improving the accuracy of prostate cancer early detection. These findings also support the usefulness of well characterized, carefully collected reference sets to evaluate new biomarkers.

How to improve reliability and efficiency of research about molecular markers: roles of phases, guidelines, and study design

BACKGROUND AND OBJECTIVE

The search for molecular markers for cancer, using "discovery-based" techniques, has resulted in claims of a very high degree of discrimination both for cancer diagnosis (e.g., serum proteomics patterns) and prognosis (e.g., RNA expression genomic signatures). However, many promising initial results have been found to be unreliable or not reproducible, and the larger process of discovery can seem slow and inefficient. To improve the process to develop molecular markers, proposals to use "phases" and "guidelines" have been made, based on experience with the process of drug development and randomized controlled clinical trials. The objective is to help improve the reliability and efficiency of development of molecular markers for cancer diagnosis.

STUDY DESIGN AND SETTING

The literature was searched to identify important current problems (in serum proteomics for cancer diagnosis and RNA expression genomics for cancer prognosis) are identified, and the roles of tools ("phases," "guidelines," and "study design") to address those problems are considered. Based on lessons learned, approaches for the future are discussed, some of which may seem "radical" compared with drug development.

RESULTS

Phases identify and organize questions to be addressed by individual studies. Guidelines identify features of design and conduct to be reported so that each study's reliability can be judged. Study design involves the myriad details and choices involved in actual planning and conduct of a study. Study design is most important in the sense of determining whether a study is reliable or not. Studies that are unreliable, because of problems from chance and bias, constitute a major current problem leading to inflated expectations, wasted effort, and inefficiency in the larger process of development. By considering fundamental principles, it may be possible to identify approaches that are different than those used in drug development, while preserving reliability and efficiency.

CONCLUSION

Phases and guidelines have important roles, but issues in study design address the fundamental problems that compromise reliability and efficiency. Tools to study markers are underdeveloped and will evolve over time, perhaps to include seemingly radical approaches.

Biomarker Discovery from Uveal Melanoma Secretomes: Identification of gp100 and Cathepsin D in Patient Serum

There is a necessity to better characterize uveal melanoma (UM) tumors according to their metastasis potential at an early stage. In this study we report the identification of potential biomarkers by a combination of proteomics-related approaches: the characterization of UM cell secretomes, the analysis of UM autoantibodies, and the differential depleted serum proteome analysis. We describe a possible role of cathepsin D, syntenin, and gp100 in UM as potential biomarkers.