Publications by authors named "Luisa Beltran"

7 Publications

  • Page 1 of 1

A Machine Learning Approach for the Automated Interpretation of Plasma Amino Acid Profiles.

Clin Chem 2020 09;66(9):1210-1218

Biochemical Sciences, Viapath, Guys & St Thomas' NHS Foundation Trust, London, UK.

Background: Plasma amino acid (PAA) profiles are used in routine clinical practice for the diagnosis and monitoring of inherited disorders of amino acid metabolism, organic acidemias, and urea cycle defects. Interpretation of PAA profiles is complex and requires substantial training and expertise to perform. Given previous demonstrations of the ability of machine learning (ML) algorithms to interpret complex clinical biochemistry data, we sought to determine if ML-derived classifiers could interpret PAA profiles with high predictive performance.

Methods: We collected PAA profiling data routinely performed within a clinical biochemistry laboratory (2084 profiles) and developed decision support classifiers with several ML algorithms. We tested the generalization performance of each classifier using a nested cross-validation (CV) procedure and examined the effect of various subsampling, feature selection, and ensemble learning strategies.

Results: The classifiers demonstrated excellent predictive performance, with the 3 ML algorithms tested producing comparable results. The best-performing ensemble binary classifier achieved a mean precision-recall (PR) AUC of 0.957 (95% CI 0.952, 0.962) and the best-performing ensemble multiclass classifier achieved a mean F4 score of 0.788 (0.773, 0.803).

Conclusions: This work builds upon previous demonstrations of the utility of ML-derived decision support tools in clinical biochemistry laboratories. Our findings suggest that, pending additional validation studies, such tools could potentially be used in routine clinical practice to streamline and aid the interpretation of PAA profiles. This would be particularly useful in laboratories with limited resources and large workloads. We provide the necessary code for other laboratories to develop their own decision support tools.
View Article and Find Full Text PDF

Download full-text PDF

Source Listing
September 2020

Therapeutic thresholds for golimumab serum concentrations during induction and maintenance therapy in ulcerative colitis: results from the GO-LEVEL study.

Aliment Pharmacol Ther 2020 07 7;52(2):292-302. Epub 2020 Jun 7.

Gastroenterology, Guy's & St Thomas' Hospital, London, UK.

Background: Significant associations between serum golimumab concentrations and favourable outcomes have been observed during both induction and maintenance therapy in ulcerative colitis (UC). However, data regarding optimal therapeutic serum golimumab concentration thresholds are limited.

Aims: To identify optimal serum golimumab concentration thresholds during induction and maintenance treatment with golimumab.

Methods: GO-LEVEL was an open label, phase IV study that included a prospective cohort of UC patients commencing golimumab, as well as a cross-sectional cohort receiving maintenance treatment. Patients commencing induction for active UC (defined as a simple clinical colitis activity index [SCCAI] >5 in addition to a raised faecal calprotectin [FC] >59μg/g or, raised C-reactive protein [CRP] [>5mg/L] or, Mayo endoscopic disease activity 2 or 3) were evaluated at weeks 6, 10 and 14. Patients receiving maintenance therapy were recruited either at the point of flare or during remission. Combined clinical-biochemical remission was defined as SCCAI ≤2 and FC <250μg/g. Serum golimumab concentrations were measured using a commercially available ELISA (LISATRACKER, Theradiag).

Results: Thirty-nine patients were included in the induction cohort, of whom 15 (38%) achieved combined clinical-biochemical remission at week 6. The median serum golimumab concentration of those in combined clinical-biochemical remission was significantly higher than those who were not (5.0 vs 3.1 μg/mL, respectively, P = 0.03). Receiver operating characteristic (ROC) curve analysis demonstrated 3.8 μg/mL as the optimal threshold (sensitivity 0.71, specificity 0.65, area under curve [AUC] 0.72, positive predictive value [PPV] 0.59 and negative predictive value [NPV] 0.79). Sixty-three patients were included in the maintenance cohort; 31 (49%) were in combined remission, 32 (51%) were not. The median serum golimumab concentration of those in combined remission was significantly higher (2.9 vs 2.1 μg/mL, respectively, P = 0.01). ROC curve analysis demonstrated 2.4 μg/mL as the optimal threshold (sensitivity 0.68, specificity 0.66, AUC 0.68, PPV 0.65 and NPV 0.66).

Conclusions: GO-LEVEL (NCT03124121) offers further evidence regarding golimumab's exposure-response relationship. Clinicians may consider using therapeutic drug monitoring to optimise golimumab dosing aiming to achieve our suggested therapeutic thresholds of 3.8 μg/mL at week 6 and 2.4 μg/mL during maintenance.
View Article and Find Full Text PDF

Download full-text PDF

Source Listing
July 2020

Global profiling of protein kinase activities in cancer cells by mass spectrometry.

J Proteomics 2012 Dec 4;77:492-503. Epub 2012 Oct 4.

Analytical Signaling Group, Centre for Cell Signaling, Barts Cancer Institute, Queen Mary University of London, Charterhouse Square, London EC1M 6BQ, UK.

Protein kinases have important functions in the control of cell biology and are implicated in several diseases including cancer. Here we describe a technique to quantify protein kinase activity in a global fashion and without preconception of the kinases that may be active in the cell or tissue under investigation. In Global Kinase Activity Profiling (GKAP), protein kinases present in experimental cell lysates phosphorylate endogenous substrates, also present in the lysate, under defined conditions. Reaction products are then quantified using standard phosphoproteomic techniques based on LC-MS/MS. The technique thus allows measuring the combined activities of kinases targeting common substrates, which are detected as phosphopeptides by LC-MS/MS. Almost four hundred kinase reactions could be quantified in a human epithelial cell line, 177 of which increased in response to EGF treatment while others decreased in cells exposed to the kinase inhibitors LY294002 or U0126. GKAP also detected marked differences in the patterns of kinase activities in human leukemia cell lines with different sensitivities to kinase inhibitors. These results reveal that GKAP detects and quantifies hundreds of kinase activities modulated by growth factors or pharmacological inhibitors, and that these activities correlate with the phenotypes of cancer cells and their responses to kinase inhibitors.
View Article and Find Full Text PDF

Download full-text PDF

Source Listing
December 2012

Advances in phosphopeptide enrichment techniques for phosphoproteomics.

Amino Acids 2012 Sep 22;43(3):1009-24. Epub 2012 Jul 22.

Analytical Signalling Group, Centre for Cell Signalling, Barts Cancer Institute-CR-UK Centre of Excellence, Queen Mary University of London, John Vane Science Centre, Charterhouse Square, London, UK.

Phosphoproteomics is increasingly used to address a wide range of biological questions. However, despite some success, techniques for phosphoproteomics are not without challenges. Phosphoproteins are present in cells in low abundance relative to their unphosphorylated counterparts; therefore phosphorylated proteins (or phosphopeptides after protein digestion) are rarely detected in standard shotgun proteomics experiments. Thus, extraction of phosphorylated polypeptides from complex mixtures is a critical step in the success of phosphoproteomics experiments. Intense research over the last decade has resulted in the development of powerful techniques for phosphopeptide enrichment prior to analysis by mass spectrometry. Here, we review how the development of IMAC, MOAC, chemical derivatization and antibody affinity purification and chromatography is contributing to the evolution of phosphoproteomics techniques. Although further developments are needed for the technology to reach maturity, current state-of-the-art techniques can already be used as powerful tools for biological research.
View Article and Find Full Text PDF

Download full-text PDF

Source Listing
September 2012

Calpain interacts with class IA phosphoinositide 3-kinases regulating their stability and signaling activity.

Proc Natl Acad Sci U S A 2011 Sep 19;108(39):16217-22. Epub 2011 Sep 19.

Analytical Signaling Group, Centre for Cell Signaling, Barts Cancer Institute, Queen Mary University of London, Charterhouse Square, London EC1M 6BQ, United Kingdom.

Class IA phosphoinositide 3-kinases (PI3Ks) are signaling enzymes with key roles in the regulation of essential cellular functions and disease, including cancer. Accordingly, their activity is tightly controlled in cells to maintain homeostasis. The formation of multiprotein complexes is a ubiquitous mechanism to regulate enzyme activity but the contribution of protein-protein interactions to the regulation of PI3K signaling is not fully understood. We designed an affinity purification quantitative mass spectrometry strategy to identify proteins interacting dynamically with PI3K in response to pathway activation, with the view that such binding partners may have a functional role in pathway regulation. Our study reveals that calpain small subunit 1 interacts with PI3K and that the association between these proteins is lower in cells stimulated with serum compared to starved cells. Calpain and PI3K activity assays confirmed these results, thus demonstrating that active calpain heterodimers associate dynamically with PI3K. In addition, calpains were found to cleave PI3K proteins in vitro (resulting in a reduction of PI3K lipid kinase activity) and to regulate endogenous PI3K protein levels in vivo. Further investigations revealed that calpains have a role in the negative regulation of PI3K/Akt pathway activity (as measured by Akt and ribosomal S6 phosphorylation) and that their inhibition promotes cell survival during serum starvation. These results indicate that the interaction between calpain and PI3K is a novel mechanism for the regulation of class IA PI3K stability and activity.
View Article and Find Full Text PDF

Download full-text PDF

Source Listing
September 2011

Characterization of a TiO₂ enrichment method for label-free quantitative phosphoproteomics.

Methods 2011 Aug 18;54(4):370-8. Epub 2011 Feb 18.

Analytical Signalling Laboratory, Centre for Cell Signalling, Barts Cancer Institute, Bart's and the London School of Medicine, Queen Mary University of London, Charterhouse Square, London EC1M 6BQ, UK.

Phosphorylation is a protein post-translational modification with key roles in the regulation of cell biochemistry and signaling. In-depth analysis of phosphorylation using mass spectrometry is permitting the investigation of processes controlled by phosphorylation at the system level. A critical step of these phosphoproteomics methods involves the isolation of phosphorylated peptides from the more abundant unmodified peptides produced by the digestion of cell lysates. Although different techniques to enrich for phosphopeptides have been reported, there are limited data on their suitability for direct quantitative analysis by MS. Here we report a TiO(2) based enrichment method compatible with large-scale and label-free quantitative analysis by LC-MS/MS. Starting with just 500 μg of protein, the technique reproducibly isolated hundreds of peptides, >85% of which were phosphorylated. These results were obtained by using relatively short LC-MS/MS gradient runs (45 min) and without any previous separation step. In order to characterize the performance of the method for quantitative analyses, we employed label-free LC-MS/MS using extracted ion chromatograms as the quantitative readout. After normalization, phosphopeptides were quantified with good precision (coefficient of variation was 20% on average, n=900 phosphopeptides), linearity (correlation coefficients >0.98) and accuracy (deviations <20%). Thus, phosphopeptide ion signals correlated with the concentration of the respective phosphopeptide in samples, making the approach suitable for in-depth relative quantification of phosphorylation by label-free LC-MS/MS.
View Article and Find Full Text PDF

Download full-text PDF

Source Listing
August 2011

Serum total prolactin and monomeric prolactin reference intervals determined by precipitation with polyethylene glycol: evaluation and validation on common immunoassay platforms.

Clin Chem 2008 Oct 21;54(10):1673-81. Epub 2008 Aug 21.

Department of Clinical Chemistry, Southend Hospital, Westcliff-on-Sea, Essex SSOORY, UK.

Background: Macroprolactin is an important source of immunoassay interference that commonly leads to misdiagnosis and mismanagement of hyperprolactinemic patients. We used the predominant immunoassay platforms for prolactin to assay serum samples treated with polyethylene glycol (PEG) and establish and validate reference intervals for total and monomeric prolactin.

Methods: We used the Architect (Abbott), ADVIA Centaur and Immulite (Siemens Diagnostics), Access (Beckman Coulter), Elecsys (Roche Diagnostics), and AIA (Tosoh) analyzers with samples from healthy males (n = 53) and females (n = 93) to derive parametric reference intervals for total and post-PEG monomeric prolactin. Concentrations of immunoreactive prolactin isoforms in serum samples from healthy individuals were established by gel filtration chromatography (GFC). We then used samples from 22 individuals whose hyperprolactinemia was entirely attributable to macroprolactin and 32 patients with true hyperprolactinemia to compare patient classifications and prolactin concentrations measured by GFC with the newly derived post-PEG reference intervals.

Results: Parametric reference intervals for post-PEG prolactin in male and female serum samples, respectively, were (in mIU/L): 61-196, 66-278 (Centaur); 63-245, 75-381 (Elecsys); 70-301, 92-469 (Access); 72-229, 79-347 (Architect); 73-247, 83-383 (AIA); and 78-263, 85-394 (Immulite). Concordance between GFC and immunoassay-specific post-PEG reference intervals was observed in 311 of 324 cases and for 31 of 32 patients with true hyperprolactinemia and 17 of 22 patients with macroprolactinemia. Results leading to misclassification occurred in a few analyzers for 5 macroprolactinemia patient samples with relatively minor increases in post-PEG prolactin (mean 61 mIU/L).

Conclusions: Our validated normative reference data for sera pretreated with PEG and analyzed on the most commonly used immunoassay platforms should facilitate the more widespread introduction of macroprolactin screening by clinical laboratories.
View Article and Find Full Text PDF

Download full-text PDF

Source Listing
October 2008