Tarja Rajalahti

Multivariate data analysis in pharmaceutics: A tutorial review

We provide an overview of latent variable methods used in pharmaceutics and integrated with advanced characterization techniques such as vibrational spectroscopy. The basics of the most common latent variable methods, principal component analysis (PCA), principal component regression (PCR) and partial least-squares (PLS) regression, are presented. Multiple linear regression (MLR) and methods for improved interpretation, variable selection, classification and validation are also briefly discussed. Extensive use of the methods is demonstrated by compilation of the recent literature.

Discriminating variable test and selectivity ratio plot: quantitative tools for interpretation and variable (biomarker) selection in complex spectral or chromatographic profiles.

The discriminating variable (DIVA) test and the selectivity ratio (SR) plot are developed as quantitative tools for revealing the variables in spectral or chromatographic profiles discriminating best between two groups of samples. The SR plot is visually similar to a spectrum or a chromatogram, but with the most intense regions corresponding to the most discriminating variables. Thus, the variables with highest SR represent the variables most important for interpretation of differences between groups. Regions with variables that are positively or negatively correlated to each other are displayed as corresponding negative and positive regions in the SR plot. The nonparametric DIVA test is designed for connecting SR to discriminatory ability of a variable quantified as probability for correct classification. A mean probability for a certain SR range is calculated as the mean correct classification rate (MCCR) for all variables in the same SR interval. The MCCR is thus similar to a mean sensitivity in each SR interval. In addition to the ranking of all variables according to their discriminatory ability provided by the SR plot, the DIVA test connects a probability measure to each SR interval. Thus, the DIVA test makes it possible to objectively define thresholds corresponding to mean probability levels in the SR plot and provides a quantitative means to select discriminating variables. In order to validate the approach, samples of untreated cerebrospinal fluid (CSF) and samples spiked with a multicomponent peptide standard were analyzed by matrix-assisted laser desorption ionization (MALDI) mass spectrometry. The differences in the multivariate spectral profiles of the two groups were revealed using partial least-squares discriminant analysis (PLS-DA) followed by target projection (TP). The most discriminating mass-to-charge (m/z) regions were revealed by calculating the ratio of explained to unexplained variance for each m/z number on the target-projected component and displaying this measure in SR plots with quantitative boundaries determined from the DIVA test. The results are compared to some established methods for variable selection.

Pre-analytical influence on the low molecular weight cerebrospinal fluid proteome

Cerebrospinal fluid (CSF) is a perfect source to search for new biomarkers to improve early diagnosis of neurological diseases. Standardization of pre‐analytical handling of the sample is, however, important to obtain acceptable analytical quality. In the present study, MALDI‐TOF MS was used to examine the influence of pre‐analytical sample procedures on the low molecular weight (MW) CSF proteome. Different storage conditions like temperature and duration or the addition of as little as 0.2 µL blood/mL neat CSF caused significant changes in the mass spectra. The performance of different types of MW cut‐off spin cartridges from different suppliers used to enrich the low MW CSF proteome showed great variance in cut‐off accuracy, stability and reproducibility. The described analytical method achieved a polypeptide discriminating limit of approximately 800 pM, two to three orders of magnitude lower than reported for plasma. Based on this study, we recommend that CSF is centrifuged immediately after sampling, prior to storage at –80ºC without addition of protease inhibitors. Guanidinium hydrochloride is preferred to break protein‐protein interactions. A spin cartridge with cut‐off limit above the intended analytical mass range is recommended. Our study contributes to the important task of developing standardized pre‐analytical protocols for the proteomic study of CSF.

A Multivariate Approach To Reveal Biomarker Signatures for Disease Classification: Application to Mass Spectral Profiles of Cerebrospinal Fluid from Patients with Multiple Sclerosis

Mass spectral profiles from cerebrospinal fluid (CSF) are used as input to a novel multivariate approach to select features responsible for the separation of patients with multiple sclerosis (MS) from control groups. Our targeted statistical approach makes it possible to systematically remove features in the spectral fingerprints masking the components expressing the disease pattern. The low molecular weight CSF proteome from 54 patients with MS and a range of other neurological diseases (OND), as well as neurological healthy controls (NHC), is analyzed in replicates using mass spectral profiling. Statistically validated partial least-squares discriminant analysis (PLS-DA) models are created as a first step to separate the groups. Using the group membership as a target, the most discriminatory projection in the multivariate space spanned by the spectral profiles is revealed. From the resulting target-projected component, the spectral regions most significantly contributing to group separation are identified using the nonparametric discriminating variable (DIVA) test together with the so-called selectivity ratio (SR) plot. Our approach is general and can be applied for other diseases and instrumental techniques as well.

Predictive associations between serum fatty acids and lipoproteins in healthy non-obese Norwegians: implications for cardiovascular health

A battery of methods for multivariate data analysis has been used to assess the associations between concentrations of fatty acids (FAs) and lipoprotein subclasses and particle size in serum for a normolipidemic population of ethnic Norwegians living in the rural Fjord region. Significant gender differences were found in the lipoprotein and FA patterns. Predictive FA patterns were revealed for lipoprotein features of importance for cardiovascular (CV) health. Thus, the subclasses of atherogenic small and very small low density lipoprotein (LDL) particles and the same subclasses of high density lipoprotein (HDL) particles were associated with a pattern of saturated FAs and mono-unsaturated C16-C18 FAs. Eicosapentaenoic acid (EPA) and the ratio of EPA to arachidonic acid (AA) had strongest associations to features that promotes CV health: (i) large average size of HDL and LDL particles, and, (ii) small average size of very low density lipoprotein (VLDL) particles. Total concentration of HDL in both genders correlated to EPA, but docosahexaenoic acid (DHA) correlated just as strongly for women. For men, docosapentaenoic acid (DPA) showed stronger association to HDL concentration than EPA. For both genders, concentration of large LDL particles showed associations to levels of EPA, but stronger to DHA and DPA. High values of EPA/AA seem to be the strongest single biomarker for good CV health in both men and women.

Aerobic fitness and metabolic health in children: A clinical validation of directly measured maximal oxygen consumption versus performance measures as markers of health

High aerobic fitness is consistently associated with a favorable metabolic health profile in children. However, measurement of oxygen uptake, regarded as the gold standard for evaluating aerobic fitness, is often not feasible. Thus, the aim of the present study was to perform a clinical validation of three measures of aerobic fitness (peak oxygen consumption [VO2peak] and time to exhaustion [TTE] determined from a graded treadmill protocol to exhaustion, and the Andersen intermittent running test) with clustered metabolic health in 10-year-old children. We included 93 children (55 boys and 38 girls) from Norway during 2012–2013 in the study. Associations between aerobic fitness and three different composite metabolic health scores (including lipoprotein subgroup particle concentrations, triglyceride, glucose, systolic blood pressure, and waist-to-height ratio) were determined by regression analyses adjusting for sex. The relationships among the measures of aerobic fitness were r = 0.78 for VO2peak vs. TTE, r = 0.63 for VO2peak vs. the Andersen test, and r = 0.67 for TTE vs. the Andersen test. The Andersen test showed the strongest associations across all markers of metabolic health (r = − 0.45 to − 0.31, p < 0.002), followed by VO2peak (r = − 0.35 to − 0.12, p < 0.256), and TTE (r = − 0.28 to − 0.10, p < 0.334). Our findings indicate that indirect measures of aerobic fitness do not stand back as markers of metabolic health status in children, compared to VO2peak. This is of great importance as good field tests provide opportunities for measuring aerobic fitness in many settings where measuring VO2peak are impossible.

Changes in serum fatty acid and lipoprotein subclass concentrations from prepuberty to adulthood and during aging.

Concentrations in serum were determined for 18 fatty acids (FAs) and 21 lipoprotein main and subclasses by chromatographic analyses and the average size was calculated for very low density (VLDL), low density (LDL) and high density (HDL) particles. 283 ethnic Norwegian children and adults from the rural Fjord region of Western Norway were compared with the objectives to reveal patterns and gender differences during the development from prepuberty to adulthood and during aging in adults. Both genders showed a large increase in eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) from child to adult. Males, but not females, show a significant increase in most C16–C18 FAs from prepuberty to adulthood. These changes in males correlate to a pattern of increased concentrations of triglycerides, VLDL and LDL particles, especially the atherogenic subclasses of small and very small LDL particles. Furthermore, concentrations of medium, large and very large HDL particles decrease, while concentration of very small HDL particles increase leading to reduced average size of HDL particles. Females only showed significant increase in concentrations of small and very small LDL particles, very small HDL particles and apolipoprotein B. While EPA and DHA continued to increase during aging in women, no validated model for connecting age to FA profile was obtained for men. Women showed significant increase in concentrations of all subclasses of LDL particles during aging, while men exhibited a more complex pattern with increase also in apolipoprotein A1 and HDL particles.

Determinants of New Product Launch Success in the Pharmaceutical Industry

PurposeThis study identifies key determinants of new product launch success, examines their role and impact on launch performance and links them to the different stages of product life cycle in the pharmaceutical new product launch context.MethodsSurvey data from pharmaceutical industry was analysed with multivariate data analysis using latent variable regression modelling followed by the calculation of selectivity ratios to reveal the most informative determinants.ResultsThe results distinguish between the determinants driving financial new product launch success and those driving customer acceptance. Whereas financial success is driven by strategic choices and tactical decisions, the relationship approach is vital in fostering customer acceptance at different phases of the innovation diffusion. Product advantage and relationship marketing activities contribute to achieving key opinion leaders’ acceptance in the early phase, while the accumulated market-based assets largely determine acceptance of a majority of other target customers in the later phase. Furthermore, launch performance is enhanced by a relationship-oriented company culture.ConclusionsThe study emphasises the significance of relational aspects in new product launches and provides both important theoretical insights and managerial implications for commercialising new pharmaceutical products.

Determination of optimum number of components in partial least squares regression from distributions of the root-mean-squared error obtained by Monte Carlo resampling: Determination of optimum number of components in PLS regression

Monte Carlo resampling is utilized to determine the number of components in partial least squares (PLS) regression. The data are randomly and repeatedly divided into calibration and validation samples. For each repetition, the root‐mean‐squared error (RMSE) is determined for the validation samples for a = 1, 2, … , A PLS components to provide a distribution of RMSE values for each number of PLS components. These distributions are used to determine the median RMSE for each number of PLS components. The component (Amin) having the lowest median RMSE is located. The fraction p of the RMSE values of Amin exceeding the median RMSE for the preceding component is determined. This fraction p represents a probability measure that can be used to decide if the RMSE for the Amin PLS component is significantly lower than the RMSE for the preceding component for a preselected threshold (pupper). If so, it defines the optimum number of PLS components. If not, the process is repeated for the previous components until significance is achieved. The pupper = 0.5 implies that the median is used for selecting the optimum number of components. The RMSE is approximately normally distributed on the smallest components. This can be utilized to relate p to a fraction of a standard deviation. For instance, p = 0.308 corresponds to half a standard deviation if RMSE is normally distributed.

Strong association between cardiorespiratory fitness and serum lipoprotein subclass pattern in prepubertal healthy children

This study reveals the lipoprotein subclass pattern associated with cardiorespiratory fitness (CRF) in healthy children. Serum concentrations of lipoprotein subclasses and concentrations and average particle size of their main classes were quantified in 94 ethnic Norwegian children using high‐performance liquid chromatography (HPLC). Twenty‐four lipoprotein features were used as input to multivariate regression analysis with CRF measured either by peak oxygen consumption (VO2peak) using a continuous treadmill protocol or indirectly by the 10‐minute Andersen intermittent running field test. By including BMI and gender as descriptors, a predictive cross‐validated multivariate regression model was obtained for both CRF measures. CRF correlated positively with average particle size for high‐density lipoprotein (HDL) and its subclasses of large HDL particles and negatively with very small HDL particles, chylomicrons, triglycerides, and average size and concentration of very low‐density lipoprotein (VLDL) particles and VLDL subclasses of large particles (P<.05). BMI correlated negatively with both measures of CRF, but exhibited a stronger association with VO2peak than with the Andersen test. Our data showed a strong association between CRF measured either by VO2peak or by the Andersen test and a subclass lipoprotein pattern that is associated with cardiovascular (CV) health. Thus, our results show why high levels of CRF are beneficial for childrens CV health. The Andersen test, being a practical field test that involves minimal equipment and, being less influenced by BMI than VO2peak, represents a good measure of CRF, and, accordingly, a proxy measure of cardiovascular health status in children.