Genome-wide microRNA analysis of HPV-positive self-samples yields novel triage markers for early detection of cervical cancer.

Offering self-sampling for HPV testing improves the effectiveness of current cervical screening programs by increasing population coverage. Molecular markers directly applicable on self-samples are needed to stratify HPV-positive women at risk of cervical cancer (so-called triage) and to avoid over-referral and overtreatment. Deregulated microRNAs (miRNAs) have been implicated in the development of cervical cancer, and represent potential triage markers. However, it is unknown whether deregulated miRNA expression is reflected in self-samples. Our study is the first to establish genome-wide miRNA profiles in HPV-positive self-samples to identify miRNAs that can predict the presence of CIN3 and cervical cancer in self-samples. Small RNA sequencing (sRNA-Seq) was conducted to determine genome-wide miRNA expression profiles in 74 HPV-positive self-samples of women with and without cervical precancer (CIN3). The optimal miRNA marker panel for CIN3 detection was determined by GRridge, a penalized method on logistic regression. Six miRNAs were validated by qPCR in 191 independent HPV-positive self-samples. Classification of sRNA-Seq data yielded a 9-miRNA marker panel with a combined area under the curve (AUC) of 0.89 for CIN3 detection. Validation by qPCR resulted in a combined AUC of 0.78 for CIN3+ detection. Our study shows that deregulated miRNA expression associated with CIN3 and cervical cancer development can be detected by sRNA-Seq in HPV-positive self-samples. Validation by qPCR indicates that miRNA expression analysis offers a promising novel molecular triage strategy for CIN3 and cervical cancer detection applicable to self-samples.

Host Cell Deoxyribonucleic Acid Methylation Markers for the Detection of High-grade Anal Intraepithelial Neoplasia and Anal Cancer.

BACKGROUND: High-grade anal intraepithelial neoplasia (AIN2/3; HGAIN) is highly prevalent in human immunodeficiency virus positive (HIV+) men who have sex with men (MSM), but only a minority will eventually progress to cancer. Currently, the cancer risk cannot be established, and therefore all HGAIN is treated, resulting in overtreatment. We assessed host cell deoxyribonucleic acid (DNA) methylation markers for detecting HGAIN and anal cancer. METHODS: Tissue samples of HIV+ men with anal cancer (n = 26), AIN3 (n = 24), AIN2 (n = 42), AIN1 (n = 22) and HIV+ male controls (n = 34) were analyzed for methylation of 9 genes using quantitative methylation-specific polymerase chain reaction. Univariable and least absolute shrinkage and selection operator logistic regression, followed by leave-one-out cross-validation, were used to determine the performance for AIN3 and cancer detection. RESULTS: Methylation of all genes increased significantly with increasing severity of disease (P < 2 × 10-6). HGAIN samples revealed heterogeneous methylation patterns, with a subset resembling cancer. Four genes (ASCL1, SST, ZIC1,ZNF582) showed remarkable performance for AIN3 and anal cancer detection (area under the curve [AUC] > 0.85). ZNF582 (AUC = 0.89), detected all cancers and 54% of AIN3 at 93% specificity. Slightly better performance (AUC = 0.90) was obtained using a 5-marker panel. CONCLUSIONS: DNA methylation is associated with anal carcinogenesis. A marker panel that includes ZNF582 identifies anal cancer and HGAIN with a cancer-like methylation pattern, warrantingvalidation studies to verify its potential for screening and management of HIV+ MSM at risk for anal cancer.

Detection of hypermethylated genes as markers for cervical screening in women living with HIV.

INTRODUCTION: To evaluate the performance of hypermethylation analysis of ASCL1, LHX8 and ST6GALNAC5 in physician-taken cervical scrapes for detection of cervical cancer and cervical intraepithelial neoplasia (CIN) grade 3 in women living with HIV (WLHIV) in South Africa. METHODS: Samples from a prospective observational cohort study were used for these analyses. Two cohorts were included: a cohort of WLHIV who were invited for cervical screening (n = 321) and a gynaecologic outpatient cohort of women referred for evaluation of abnormal cytology or biopsy proven cervical cancer (n = 108, 60% HIV seropositive). Cervical scrapes collected from all subjects were analysed for hypermethylation of ASCL1, LHX8 and ST6GALNAC5 by multiplex quantitative methylation specific PCR (qMSP). Histology endpoints were available for all study subjects. RESULTS: Hypermethylation levels of ASCL1, LHX8 and ST6GALNAC5 increased with severity of cervical disease. The performance for detection of CIN3 or worse (CIN3+ ) as assessed by the area under the receiver operating characteristic (ROC) curves (AUC) was good for ASCL1 and LHX8 (AUC 0.79 and 0.81 respectively), and moderate for ST6GALNAC5 (AUC 0.71). At a threshold corresponding to 75% specificity, CIN3+ sensitivity was 72.1% for ASCL1 and 73.8% for LHX8 and all samples from women with cervical cancer scored positive for these two markers. CONCLUSIONS: Hypermethylation analysis of ASCL1 or LHX8 in cervical scrape material of WLHIV detects all cervical carcinomas with an acceptable sensitivity and good specificity for CIN3+ , warranting further exploration of these methylation markers as a stand-alone test for cervical screening in low-resource settings.

Host-cell DNA methylation patterns during high-risk HPV-induced carcinogenesis reveal a heterogeneous nature of cervical pre-cancer.

Cervical cancer development following a persistent infection with high-risk human papillomavirus (hrHPV) is driven by additional host-cell changes, such as altered DNA methylation. In previous studies, we have identified 12 methylated host genes associated with cervical cancer and pre-cancer (CIN2/3). This study systematically analyzed the onset and DNA methylation pattern of these genes during hrHPV-induced carcinogenesis using a longitudinal in vitro model of hrHPV-transformed cell lines (n = 14) and hrHPV-positive cervical scrapings (n = 113) covering various stages of cervical carcinogenesis. DNA methylation analysis was performed by quantitative methylation-specific PCR (qMSP) and relative qMSP values were used to analyze the data. The majority of genes displayed a comparable DNA methylation pattern in both cell lines and clinical specimens. DNA methylation onset occurred at early or late immortal passage, and DNA methylation levels gradually increased towards tumorigenic cells. Subsequently, we defined a so-called cancer-like methylation-high pattern based on the DNA methylation levels observed in cervical scrapings from women with cervical cancer. This cancer-like methylation-high pattern was observed in 72% (38/53) of CIN3 and 55% (11/20) of CIN2, whereas it was virtually absent in hrHPV-positive controls (1/26). In conclusion, hrHPV-induced carcinogenesis is characterized by early onset of DNA methylation, typically occurring at the pre-tumorigenic stage and with highest DNA methylation levels at the cancer stage. Host-cell DNA methylation patterns in cervical scrapings from women with CIN2 and CIN3 are heterogeneous, with a subset displaying a cancer-like methylation-high pattern, suggestive for a higher cancer risk.

Triage of high-risk HPV-positive women in population-based screening by miRNA expression analysis in cervical scrapes; a feasibility study.

Background: Primary testing for high-risk HPV (hrHPV) is increasingly implemented in cervical cancer screening programs. Many hrHPV-positive women, however, harbor clinically irrelevant infections, demanding additional disease markers to prevent over-referral and over-treatment. Most promising biomarkers reflect molecular events relevant to the disease process that can be measured objectively in small amounts of clinical material, such as miRNAs. We previously identified eight miRNAs with altered expression in cervical precancer and cancer due to either methylation-mediated silencing or chromosomal alterations. In this study, we evaluated the clinical value of these eight miRNAs on cervical scrapes to triage hrHPV-positive women in cervical screening. Results: Expression levels of the eight candidate miRNAs in cervical tissue samples (n = 58) and hrHPV-positive cervical scrapes from a screening population (n = 187) and cancer patients (n = 38) were verified by quantitative RT-PCR. In tissue samples, all miRNAs were significantly differentially expressed (p < 0.05) between normal, high-grade precancerous lesions (CIN3), and/or cancer. Expression patterns detected in cervical tissue samples were reflected in cervical scrapes, with five miRNAs showing significantly differential expression between controls and women with CIN3 and cancer. Using logistic regression analysis, a miRNA classifier was built for optimal detection of CIN3 in hrHPV-positive cervical scrapes from the screening population and its performance was evaluated using leave-one-out cross-validation. This miRNA classifier consisted of miR-15b-5p and miR-375 and detected a major subset of CIN3 as well as all carcinomas at a specificity of 70%. The CIN3 detection rate was further improved by combining the two miRNAs with HPV16/18 genotyping. Interestingly, both miRNAs affected the viability of cervical cancer cells in vitro. Conclusions: This study shows that miRNA expression analysis in cervical scrapes is feasible and enables the early detection of cervical cancer, thus underlining the potential of miRNA expression analysis for triage of hrHPV-positive women in cervical cancer screening.

Identification and Validation of a 3-Gene Methylation Classifier for HPV-Based Cervical Screening on Self-Samples.

Purpose: Offering self-sampling of cervico-vaginal material for high-risk human papillomavirus (hrHPV) testing is an effective method to increase the coverage in cervical screening programs. Molecular triage directly on hrHPV-positive self-samples for colposcopy referral opens the way to full molecular cervical screening. Here, we set out to identify a DNA methylation classifier for detection of cervical precancer (CIN3) and cancer, applicable to lavage and brush self-samples.Experimental Design: We determined genome-wide DNA methylation profiles of 72 hrHPV-positive self-samples, using the Infinium Methylation 450K Array. The selected DNA methylation markers were evaluated by multiplex quantitative methylation-specific PCR (qMSP) in both hrHPV-positive lavage (n = 245) and brush (n = 246) self-samples from screening cohorts. Subsequently, logistic regression analysis was performed to build a DNA methylation classifier for CIN3 detection applicable to self-samples of both devices. For validation, an independent set of hrHPV-positive lavage (n = 199) and brush (n = 287) self-samples was analyzed.Results: Genome-wide DNA methylation profiling revealed 12 DNA methylation markers for CIN3 detection. Multiplex qMSP analysis of these markers in large series of lavage and brush self-samples yielded a 3-gene methylation classifier (ASCL1, LHX8, and ST6GALNAC5). This classifier showed a very good clinical performance for CIN3 detection in both lavage (AUC = 0.88; sensitivity = 74%; specificity = 79%) and brush (AUC = 0.90; sensitivity = 88%; specificity = 81%) self-samples in the validation set. Importantly, all self-samples from women with cervical cancer scored DNA methylation-positive.Conclusions: By genome-wide DNA methylation profiling on self-samples, we identified a highly effective 3-gene methylation classifier for direct triage on hrHPV-positive self-samples, which is superior to currently available methods. Clin Cancer Res; 24(14); 3456-64. ©2018 AACR.

Meta-analysis approach as a gene selection method in class prediction: does it improve model performance? A case study in acute myeloid leukemia.

BACKGROUND: Aggregating gene expression data across experiments via meta-analysis is expected to increase the precision of the effect estimates and to increase the statistical power to detect a certain fold change. This study evaluates the potential benefit of using a meta-analysis approach as a gene selection method prior to predictive modeling in gene expression data. RESULTS: Six raw datasets from different gene expression experiments in acute myeloid leukemia (AML) and 11 different classification methods were used to build classification models to classify samples as either AML or healthy control. First, the classification models were trained on gene expression data from single experiments using conventional supervised variable selection and externally validated with the other five gene expression datasets (referred to as the individual-classification approach). Next, gene selection was performed through meta-analysis on four datasets, and predictive models were trained with the selected genes on the fifth dataset and validated on the sixth dataset. For some datasets, gene selection through meta-analysis helped classification models to achieve higher performance as compared to predictive modeling based on a single dataset; but for others, there was no major improvement. Synthetic datasets were generated from nine simulation scenarios. The effect of sample size, fold change and pairwise correlation between differentially expressed (DE) genes on the difference between MA- and individual-classification model was evaluated. The fold change and pairwise correlation significantly contributed to the difference in performance between the two methods. The gene selection via meta-analysis approach was more effective when it was conducted using a set of data with low fold change and high pairwise correlation on the DE genes. CONCLUSION: Gene selection through meta-analysis on previously published studies potentially improves the performance of a predictive model on a given gene expression data.

Better diagnostic signatures from RNAseq data through use of auxiliary co-data.

SUMMARY: Our aim is to improve omics based prediction and feature selection using multiple sources of auxiliary information: co-data. Adaptive group regularized ridge regression (GRridge) was proposed to achieve this by estimating additional group-based penalty parameters through an empirical Bayes method at a low computational cost. We illustrate the GRridge method and software on RNA sequencing datasets. The method boosts the performance of an ordinary ridge regression and outperforms other classifiers. Post-hoc feature selection maintains the predictive ability of the classifier with far fewer markers. AVAILABILITY AND IMPLEMENTATION: GRridge is an R package that includes a vignette. It is freely available at ( https://bioconductor.org/packages/GRridge/ ). All information and R scripts used in this study, including those on retrieval and processing of the co-data, are available from http://github.com/markvdwiel/GRridgeCodata . CONTACT: mark.vdwiel@vumc.nl. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Genome-wide DNA Methylation Profiling Reveals Methylation Markers Associated with 3q Gain for Detection of Cervical Precancer and Cancer.

Purpose: Epigenetic host cell changes involved in cervical cancer development following a persistent high-risk human papillomavirus (hrHPV) infection, provide promising markers for the management of hrHPV-positive women. In particular, markers based on DNA methylation of tumor suppressor gene promoters are valuable. These markers ideally identify hrHPV-positive women with precancer (CIN2/3) in need of treatment. Here, we set out to identify biologically relevant methylation markers by genome-wide methylation analysis of both hrHPV-transformed cell lines and cervical tissue specimens.Experimental Design and Results: Genome-wide discovery by next-generation sequencing (NGS) of methyl-binding domain-enriched DNA (MBD-Seq) yielded 20 candidate methylation target genes. Further verification and validation by multiplex-targeted bisulfite NGS and (quantitative) methylation-specific PCR (MSP) resulted in 3 genes (GHSR, SST, and ZIC1) that showed a significant increase in methylation with severity of disease in both tissue specimens and cervical scrapes (P < 0.005). The area under the ROC curve for CIN3 or worse varied between 0.86 and 0.89. Within the group of CIN2/3, methylation levels of all 3 genes increased with duration of lesion existence (P < 0.0005), characterized by duration of preceding hrHPV infection, and were significantly higher in the presence of a 3q gain (P < 0.05) in the corresponding tissue biopsy.Conclusions: By unbiased genome-wide DNA methylation profiling and comprehensive stepwise verification and validation studies using in vitro and patient-derived samples, we identified 3 promising methylation markers (GHSR, SST, and ZIC1) associated with a 3q gain for the detection of cervical (pre)cancer. Clin Cancer Res; 23(14); 3813-22. ©2017 AACR.

Selecting a classification function for class prediction with gene expression data.

MOTIVATION: Class predicting with gene expression is widely used to generate diagnostic and/or prognostic models. The literature reveals that classification functions perform differently across gene expression datasets. The question, which classification function should be used for a given dataset remains to be answered. In this study, a predictive model for choosing an optimal function for class prediction on a given dataset was devised. RESULTS: To achieve this, gene expression data were simulated for different values of gene-pairs correlations, sample size, genes' variances, deferentially expressed genes and fold changes. For each simulated dataset, ten classifiers were built and evaluated using ten classification functions. The resulting accuracies from 1152 different simulation scenarios by ten classification functions were then modeled using a linear mixed effects regression on the studied data characteristics, yielding a model that predicts the accuracy of the functions on a given data. An application of our model on eight real-life datasets showed positive correlations (0.33-0.82) between the predicted and expected accuracies. CONCLUSION: The here presented predictive model might serve as a guide to choose an optimal classification function among the 10 studied functions, for any given gene expression data. AVAILABILITY AND IMPLEMENTATION: The R source code for the analysis and an R-package 'SPreFuGED' are available at Bioinformatics online. CONTACT: v.l.jong@umcutecht.nl SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

An Application of Sequential Meta-Analysis to Gene Expression Studies.

Most of the discoveries from gene expression data are driven by a study claiming an optimal subset of genes that play a key role in a specific disease. Meta-analysis of the available datasets can help in getting concordant results so that a real-life application may be more successful. Sequential meta-analysis (SMA) is an approach for combining studies in chronological order while preserving the type I error and pre-specifying the statistical power to detect a given effect size. We focus on the application of SMA to find gene expression signatures across experiments in acute myeloid leukemia. SMA of seven raw datasets is used to evaluate whether the accumulated samples show enough evidence or more experiments should be initiated. We found 313 differentially expressed genes, based on the cumulative information of the experiments. SMA offers an alternative to existing methods in generating a gene list by evaluating the adequacy of the cumulative information.

Factors affecting the accuracy of a class prediction model in gene expression data.

BACKGROUND: Class prediction models have been shown to have varying performances in clinical gene expression datasets. Previous evaluation studies, mostly done in the field of cancer, showed that the accuracy of class prediction models differs from dataset to dataset and depends on the type of classification function. While a substantial amount of information is known about the characteristics of classification functions, little has been done to determine which characteristics of gene expression data have impact on the performance of a classifier. This study aims to empirically identify data characteristics that affect the predictive accuracy of classification models, outside of the field of cancer. RESULTS: Datasets from twenty five studies meeting predefined inclusion and exclusion criteria were downloaded. Nine classification functions were chosen, falling within the categories: discriminant analyses or Bayes classifiers, tree based, regularization and shrinkage and nearest neighbors methods. Consequently, nine class prediction models were built for each dataset using the same procedure and their performances were evaluated by calculating their accuracies. The characteristics of each experiment were recorded, (i.e., observed disease, medical question, tissue/cell types and sample size) together with characteristics of the gene expression data, namely the number of differentially expressed genes, the fold changes and the within-class correlations. Their effects on the accuracy of a class prediction model were statistically assessed by random effects logistic regression. The number of differentially expressed genes and the average fold change had significant impact on the accuracy of a classification model and gave individual explained-variation in prediction accuracy of up to 72% and 57%, respectively. Multivariable random effects logistic regression with forward selection yielded the two aforementioned study factors and the within class correlation as factors affecting the accuracy of classification functions, explaining 91.5% of the between study variation. CONCLUSIONS: We evaluated study- and data-related factors that might explain the varying performances of classification functions in non-cancerous datasets. Our results showed that the number of differentially expressed genes, the fold change, and the correlation in gene expression data significantly affect the accuracy of class prediction models.

Exploring homogeneity of correlation structures of gene expression datasets within and between etiological disease categories.

The literature shows that classifiers perform differently across datasets and that correlations within datasets affect the performance of classifiers. The question that arises is whether the correlation structure within datasets differ significantly across diseases. In this study, we evaluated the homogeneity of correlation structures within and between datasets of six etiological disease categories; inflammatory, immune, infectious, degenerative, hereditary and acute myeloid leukemia (AML). We also assessed the effect of filtering; detection call and variance filtering on correlation structures. We downloaded microarray datasets from ArrayExpress for experiments meeting predefined criteria and ended up with 12 datasets for non-cancerous diseases and six for AML. The datasets were preprocessed by a common procedure incorporating platform-specific recommendations and the two filtering methods mentioned above. Homogeneity of correlation matrices between and within datasets of etiological diseases was assessed using the Box's M statistic on permuted samples. We found that correlation structures significantly differ between datasets of the same and/or different etiological disease categories and that variance filtering eliminates more uncorrelated probesets than detection call filtering and thus renders the data highly correlated.

Evaluation of gene expression classification studies: factors associated with classification performance.

Classification methods used in microarray studies for gene expression are diverse in the way they deal with the underlying complexity of the data, as well as in the technique used to build the classification model. The MAQC II study on cancer classification problems has found that performance was affected by factors such as the classification algorithm, cross validation method, number of genes, and gene selection method. In this paper, we study the hypothesis that the disease under study significantly determines which method is optimal, and that additionally sample size, class imbalance, type of medical question (diagnostic, prognostic or treatment response), and microarray platform are potentially influential. A systematic literature review was used to extract the information from 48 published articles on non-cancer microarray classification studies. The impact of the various factors on the reported classification accuracy was analyzed through random-intercept logistic regression. The type of medical question and method of cross validation dominated the explained variation in accuracy among studies, followed by disease category and microarray platform. In total, 42% of the between study variation was explained by all the study specific and problem specific factors that we studied together.

Estimation of between-trial variance in sequential meta-analyses: a simulation study.

Estimators for the variance between treatment effects from randomized clinical trials (RCTs) in a meta-analysis may yield divergent or even contradictory results. In a sequential meta-analysis (SMA), their properties are even more important, as they influence the point in time at which definite conclusions are drawn. In this study, we evaluated the properties of estimators of heterogeneity to be used in an SMA. We conducted an extensive simulation study with dichotomous and continuous outcome data and applied the estimators in real life examples. Bias and variance of the estimators were used as primary evaluation criteria, as well as the number of RCTs and patients from the accumulating trials needed to get stable estimates. The simulation studies showed that the well-known DerSimonian-Laird (DL) estimator largely underestimates the true value for dichotomous outcomes. The two-step DL (DL2) significantly improves this behavior. In general, the DL2 and Paule-Mandel (PM) estimators are recommended for both dichotomous and continuous outcome data for use in an SMA.

Diagnostic accuracy of optical coherence tomography parameters in predicting in-stent hemodynamic severe coronary lesions: validation against fractional flow reserve.

BACKGROUND: By means of optical coherence tomography (OCT), coronary dimensions can be assessed accurately. However, whether OCT can identify hemodynamic significant coronary lesions as determined by fractional flow reserve (FFR) in patients with an in-stent lesion is not known. Therefore, we tried to assess the predictive value of OCT parameters in this setting as compared to FFR. METHODS AND RESULTS: In patients who underwent a percutaneous coronary intervention for an in-stent restenotic lesion, pre-, post-procedural and 6-month follow-up OCT and FFR acquisitions were performed. In case of an FFR ≤ 0.80, a lesion was classified as hemodynamically severe. Diagnostic efficiency of several OCT parameters were assessed with receiver operating characteristic curves. In 27 patients, 66 coupled OCT and FFR segments were analyzed and compared. The diagnostic efficiencies of OCT-based minimal luminal diameter (MLD) and minimal luminal area (MLA) are good, with an area under the curve (AUC) of 0.83 (95% confidence interval: 0.74-0.93) and 0.83 (0.73-0.93), and a best cutoff value of 1.77 mm (sensitivity 74% and specificity 78%) and 2.54 mm(2) (sensitivity 71% and specificity 84%), respectively. The diagnostic efficiency of OCT-based maximum neointimal area is moderate [AUC 0.73 (0.61-0.85)], and regarding maximum neointimal area stenosis, it is poor [0.39 (0.25-0.53)]. The corresponding best cutoff values are 5.01 mm(2) (sensitivity 66% and specificity 72%) and 49% (sensitivity 40% and specificity 66%), respectively. CONCLUSIONS: With OCT, a good diagnostic efficiency can be achieved in identifying coronary severity in in-stent lesions in a per-group analysis. This hallmark provides an extra dimension, next to morphological information, when acquiring OCT images in scientific studies. However, OCT seems limited in a per-patient clinical decision making process due to reasonable but limited sensitivity and specificity in predicting coronary severity.