Feature selection for OPLS discriminant analysis of cancer tissue lipidomics data

Performance evaluate of different chemometrics formalisms used for prostate cancer diagnosis by NMR-based metabolomics

INTRODUCTION

In general, two characteristics are ever present in NMR-based metabolomics studies: (1) they are assays aiming to classify the samples in different groups, and (2) the number of samples is smaller than the feature (chemical shift) number. It is also common to observe imbalanced datasets due to the sampling method and/or inclusion criteria. These situations can cause overfitting. However, appropriate feature selection and classification methods can be useful to solve this issue.

OBJECTIVES

Investigate the performance of metabolomics models built from the association between feature selectors, the absence of feature selection, and classification algorithms, as well as use the best performance model as an NMR-based metabolomic method for prostate cancer diagnosis.

METHODS

We evaluated the performance of NMR-based metabolomics models for prostate cancer diagnosis using seven feature selectors and five classification formalisms. We also obtained metabolomics models without feature selection. In this study, thirty-eight volunteers with a positive diagnosis of prostate cancer and twenty-three healthy volunteers were enrolled.

RESULTS

Thirty-eight models obtained were evaluated using AUROC, accuracy, sensitivity, specificity, and kappa's index values. The best result was obtained when Genetic Algorithm was used with Linear Discriminant Analysis with 0.92 sensitivity, 0.83 specificity, and 0.88 accuracy.

CONCLUSION

The results show that the pick of a proper feature selection method and classification model, and a resampling method can avoid overfitting in a small metabolomic dataset. Furthermore, this approach would decrease the number of biopsies and optimize patient follow-up. 1H NMR-based metabolomics promises to be a non-invasive tool in prostate cancer diagnosis.

Methodology of determining the metabolomic profile of tumor-associated macrophages and monocytes in oncological diseases

Breast cancer is the leading cause of cancer-related death among women worldwide. Tumor-associated macrophages (TAMs) constitute the primary component of innate immunity in breast cancer tissue. During the development of new approaches for breast cancer treatment aimed at editing the epigenome of TAM, precise methods for the analysis of macrophage metabolome are required to examine the effect on new approaches on macrophage metabolism. Our study aimed to develop an HPLC-MS/MS-based analytical approach to characterize the metabolome of human innate immune cells (TAMs and their precursors, monocytes). Analysis of lipid extracts was conducted on a Dionex UltiMate 3000 liquid chromatograph connected to a Maxis Impact qTOF mass analyzer with an ESI ion source. Quantitative analysis of 38 amino acids in the cells was conducted using the Jasem Amino Acids LC-MS/MS Analysis Kit and an HPLC-MS/MS chromatographic system consisting out of an Agilent 6460 triple quadrupole mass spectrometric detector (Agilent), and an Agilent 1260 II liquid chromatograph (Agilent ) with Amino acids-HPLC Column (Jasem). The modified Folch method with double extraction was found to be the optimal approached for the sample preparation, since it enables to simultaneously isolate the lipid extract and water-soluble substances, in particular, amino acids. The method of reversed-phase chromatography yielded more useful data on the cell lipid composition than the method of hydrophilic interaction liquid chromatography (HILIC). The minimum number of cells required to determine the metabolome of immune system cells (TAM and monocytes) was identified as 2 × 106. Thus, we have developed the approach to determine the lipid and amino acid composition of modelled human TAMs and primary monocytes isolated out of breast cancer patients using minimal amount of clinical material.

Rapid diagnosis of papillary thyroid carcinoma with machine learning and probe electrospray ionization mass spectrometry

Frozen section examination could provide pathological diagnosis for surgery of thyroid nodules, which is time-consuming, skill- and experience-dependent. This study developed a rapid classification method for thyroid nodules and machine learning. Total 69 tissues were collected including 43 nodules and 26 nodule-adjacent tissues. Intraoperative frozen section was first performed to give accurate diagnosis, and the rest frozen specimen were pretreated for probe electrospray ionization mass measurement. By multivariate analysis of mass scan data, a series compounds were found downregulated in the extraction solution of papillary thyroid carcinoma (PTC), but some were found upregulated by mass spectrometry imaging. m/z 758.5713 ([PC[34:2] + H]⁺ ), m/z 772.5845 ([PC[32:0] + K]⁺ ), and m/z 786.6037 ([PC[36:2] + H]⁺ ) were firstly identified as potential biomarkers for nodular goiter (NG). Machine learning was employed by means of support vector machine (SVM) and random forest (RF) algorithms. For classification of PTC from NG, SVM and RF algorithms exhibited the same performance and the concordance was 94.2% and 94.4% between prediction and pathological diagnosis with positive and negative mass dataset, respectively. For the classification of PTC from PTC adjacent tissues, SVM was better than RF and the concordance was 93.8% and 83.3% with positive and negative mass dataset, respectively. With the identified compounds as training features, the sensitivity and specificity are 87.5% and 88.9% for the test set. The developed method could also correctly predict the malignancy of one medullary thyroid carcinoma and one adenomatous goiter (benign). The diagnosis time is about 10 min for one specimen, and it is very promising for the intraoperative diagnosis of papillary thyroid carcinoma.

Shotgun Lipidomics for Differential Diagnosis of HPV-Associated Cervix Transformation

A dramatic increase in cervical diseases associated with human papillomaviruses (HPV) in women of reproductive age has been observed over the past decades. An accurate differential diagnosis of the severity of cervical intraepithelial neoplasia and the choice of the optimal treatment requires the search for effective biomarkers with high diagnostic and prognostic value. The objective of this study was to introduce a method for rapid shotgun lipidomics to differentiate stages of HPV-associated cervix epithelium transformation. Tissue samples from 110 HPV-positive women with cervicitis (n = 30), low-grade squamous intraepithelial lesions (LSIL) (n = 30), high-grade squamous intraepithelial lesions (HSIL) (n = 30), and cervical cancers (n = 20) were obtained. The cervical epithelial tissue lipidome at different stages of cervix neoplastic transformation was studied by a shotgun label-free approach. It is based on electrospray ionization mass spectrometry (ESI-MS) data of a tissue extract. Lipidomic data were processed by the orthogonal projections to latent structures discriminant analysis (OPLS-DA) to build statistical models, differentiating stages of cervix transformation. Significant differences in the lipid profile between the lesion and surrounding tissues were revealed in chronic cervicitis, LSIL, HSIL, and cervical cancer. The lipids specific for HPV-induced cervical transformation mainly belong to glycerophospholipids: phosphatidylcholines, and phosphatidylethanolamines. The developed diagnostic OPLS-DA models were based on 23 marker lipids. More than 90% of these marker lipids positively correlated with the degree of cervix transformation. The algorithm was developed for the management of patients with HPV-associated diseases of the cervix, based on the panel of 23 lipids as a result. ESI-MS analysis of a lipid extract by direct injection through a loop, takes about 25 min (including preparation of the lipid extract), which is significantly less than the time required for the HPV test (several hours for hybrid capture and about an hour for PCR). This makes lipid mass spectrometric analysis a promising method for express diagnostics of HPV-associated neoplastic diseases of the cervix.

Alterations in lipid profile upon uterine fibroids and its recurrence

Uterine fibroids (UF) is the most common (about 70% cases) type of gynecological disease, with the recurrence rate varying from 11 to 40%. Because UF has no distinct symptomatology and is often asymptomatic, the specific and sensitive diagnosis of UF as well as the assessment for the probability of UF recurrence pose considerable challenge. The aim of this study was to characterize alterations in the lipid profile of tissues associated with the first-time diagnosed UF and recurrent uterine fibroids (RUF) and to explore the potential of mass spectrometry (MS) lipidomics analysis of blood plasma samples for the sensitive and specific determination of UF and RUF with low invasiveness of analysis. MS analysis of lipid levels in the myometrium tissues, fibroids tissues and blood plasma samples was carried out on 66 patients, including 35 patients with first-time diagnosed UF and 31 patients with RUF. The control group consisted of 15 patients who underwent surgical treatment for the intrauterine septum. Fibroids and myometrium tissue samples were analyzed using direct MS approach. Blood plasma samples were analyzed using high performance liquid chromatography hyphened with mass spectrometry (HPLC/MS). MS data were processed by discriminant analysis with projection into latent structures (OPLS-DA). Significant differences were found between the first-time UF, RUF and control group in the levels of lipids involved in the metabolism of glycerophospholipids, sphingolipids, lipids with an ether bond, triglycerides and fatty acids. Significant differences between the control group and the groups with UF and RUF were found in the blood plasma levels of cholesterol esters, triacylglycerols, (lyso) phosphatidylcholines and sphingomyelins. Significant differences between the UF and RUF groups were found in the blood plasma levels of cholesterol esters, phosphotidylcholines, sphingomyelins and triacylglycerols. Diagnostic models based on the selected differential lipids using logistic regression showed sensitivity and specificity of 88% and 86% for the diagnosis of first-time UF and 95% and 79% for RUF, accordingly. This study confirms the involvement of lipids in the pathogenesis of uterine fibroids. A diagnostically significant panel of differential lipid species has been identified for the diagnosis of UF and RUF by low-invasive blood plasma analysis. The developed diagnostic models demonstrated high potential for clinical use and further research in this direction.

Comparison of the effectiveness of variable selection method for creating a diagnostic panel of biomarkers for mass spectrometric lipidome analysis

Hundreds of compounds are detected during untargeted lipidomics analysis. The potential efficacy of lipids as disease markers makes it important to select the species with the most discriminative potential. Datasets based on a selected class of lipids allow the development of a high-quality diagnostic model using orthogonal projection on latent structure. The combination of selection of lipids by variable importance in projection and by Akaike information criteria makes it possible to build a reliable diagnostic model based on logistic regression.

Identification of novel neuroblastoma biomarkers in urine samples

Urine is a complex liquid containing numerous small molecular metabolites. The ability to non-invasively test for cancer biomarkers in urine is especially beneficial for screening child patients. This study attempted to identify neuroblastoma biomarkers by comprehensively analysing urinary metabolite samples from children. A total of 87 urine samples were collected from 54 participants (15 children with neuroblastoma and 39 without cancer) and used to perform a comprehensive analysis. Urine metabolites were extracted using liquid chromatography/mass spectrometry and analysed by Metabolon, Inc. Biomarker candidates were extracted using the Wilcoxon rank sum test, random forest method (RF), and orthogonal partial least squares discriminant analysis (OPLS-DA). RF identified three important metabolic pathways in 15 samples from children with neuroblastoma. One metabolite was selected from each of the three identified pathways and combined to create a biomarker candidate (3-MTS, CTN, and COR) that represented each of the three pathways; using this candidate, all 15 cases were accurately distinguishable from the control group. Two cases in which known biomarkers were negative tested positive using this new biomarker. Furthermore, the predictive value did not decrease in cases with a low therapeutic effect. This approach could be effectively applied to identify biomarkers for other cancer types.

Metabolomics for Diagnosis and Prognosis of Uterine Diseases? A Systematic Review

This systematic review analyses the contribution of metabolomics to the identification of diagnostic and prognostic biomarkers for uterine diseases. These diseases are diagnosed invasively, which entails delayed treatment and a worse clinical outcome. New options for diagnosis and prognosis are needed. PubMed, OVID, and Scopus were searched for research papers on metabolomics in physiological fluids and tissues from patients with uterine diseases. The search identified 484 records. Based on inclusion and exclusion criteria, 44 studies were included into the review. Relevant data were extracted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) checklist and quality was assessed using the QUADOMICS tool. The selected metabolomics studies analysed plasma, serum, urine, peritoneal, endometrial, and cervico-vaginal fluid, ectopic/eutopic endometrium, and cervical tissue. In endometriosis, diagnostic models discriminated patients from healthy and infertile controls. In cervical cancer, diagnostic algorithms discriminated patients from controls, patients with good/bad prognosis, and with/without response to chemotherapy. In endometrial cancer, several models stratified patients from controls and recurrent from non-recurrent patients. Metabolomics is valuable for constructing diagnostic models. However, the majority of studies were in the discovery phase and require additional research to select reliable biomarkers for validation and translation into clinical practice. This review identifies bottlenecks that currently prevent the translation of these findings into clinical practice.

Validation of Breast Cancer Margins by Tissue Spray Mass Spectrometry

Current methods for the intraoperative determination of breast cancer margins commonly suffer from the insufficient accuracy, specificity and/or low speed of analysis, increasing the time and cost of operation as well the risk of cancer recurrence. The purpose of this study is to develop a method for the rapid and accurate determination of breast cancer margins using direct molecular profiling by mass spectrometry (MS). Direct molecular fingerprinting of tiny pieces of breast tissue (approximately 1 × 1 × 1 mm) is performed using a home-built tissue spray ionization source installed on a Maxis Impact quadrupole time-of-flight mass spectrometer (qTOF MS) (Bruker Daltonics, Hamburg, Germany). Statistical analysis of MS data from 50 samples of both normal and cancer tissue (from 25 patients) was performed using orthogonal projections onto latent structures discriminant analysis (OPLS-DA). Additionally, the results of OPLS classification of new 19 pieces of two tissue samples were compared with the results of histological analysis performed on the same tissues samples. The average time of analysis for one sample was about 5 min. Positive and negative ionization modes are used to provide complementary information and to find out the most informative method for a breast tissue classification. The analysis provides information on 11 lipid classes. OPLS-DA models are created for the classification of normal and cancer tissue based on the various datasets: All mass spectrometric peaks over 300 counts; peaks with a statistically significant difference of intensity determined by the Mann–Whitney U-test (p < 0.05); peaks identified as lipids; both identified and significantly different peaks. The highest values of Q2 have models built on all MS peaks and on significantly different peaks. While such models are useful for classification itself, they are of less value for building explanatory mechanisms of pathophysiology and providing a pathway analysis. Models based on identified peaks are preferable from this point of view. Results obtained by OPLS-DA classification of the tissue spray MS data of a new sample set (n = 19) revealed 100% sensitivity and specificity when compared to histological analysis, the “gold” standard for tissue classification. “All peaks” and “significantly different peaks” datasets in the positive ion mode were ideal for breast cancer tissue classification. Our results indicate the potential of tissue spray mass spectrometry for rapid, accurate and intraoperative diagnostics of breast cancer tissue as a means to reduce surgical intervention.