Use of lipidomics to investigate sebum dysfunction in juvenile acne
2016; Elsevier BV; Volume: 57; Issue: 6 Linguagem: Inglês
10.1194/jlr.m067942
ISSN1539-7262
AutoresEmanuela Camera, Matteo Ludovici, Sara Tortorella, Jo Linda Sinagra, B. Capitanio, Laura Goracci, Mauro Picardo,
Tópico(s)Essential Oils and Antimicrobial Activity
ResumoAcne is a multifactorial skin disorder frequently observed during adolescence with different grades of severity. Multiple factors centering on sebum secretion are implicated in acne pathogenesis. Despite the recognized role of sebum, its compositional complexity and limited analytical approaches have hampered investigation of alterations specifically associated with acne. To examine the profiles of lipid distribution in acne sebum, 61 adolescents (29 males and 32 females) were enrolled in this study. Seventeen subjects presented no apparent clinical signs of acne. The 44 affected individuals were clinically classified as mild (13 individuals), moderate (19 individuals), and severe (12 individuals) acne. Sebum was sampled from the forehead with SebutapeTM adhesive patches. Profiles of neutral lipids were acquired with rapid-resolution reversed-phase/HPLC-TOF/MS in positive ion mode. Univariate and multivariate statistical analyses led to the identification of lipid species with significantly different levels between healthy and acne sebum. The majority of differentiating lipid species were diacylglycerols (DGs), followed by fatty acyls, sterols, and prenols. Overall, the data indicated an association between the clinical grading of acne and sebaceous lipid fingerprints and highlighted DGs as more abundant in sebum from adolescents affected with acne. Acne is a multifactorial skin disorder frequently observed during adolescence with different grades of severity. Multiple factors centering on sebum secretion are implicated in acne pathogenesis. Despite the recognized role of sebum, its compositional complexity and limited analytical approaches have hampered investigation of alterations specifically associated with acne. To examine the profiles of lipid distribution in acne sebum, 61 adolescents (29 males and 32 females) were enrolled in this study. Seventeen subjects presented no apparent clinical signs of acne. The 44 affected individuals were clinically classified as mild (13 individuals), moderate (19 individuals), and severe (12 individuals) acne. Sebum was sampled from the forehead with SebutapeTM adhesive patches. Profiles of neutral lipids were acquired with rapid-resolution reversed-phase/HPLC-TOF/MS in positive ion mode. Univariate and multivariate statistical analyses led to the identification of lipid species with significantly different levels between healthy and acne sebum. The majority of differentiating lipid species were diacylglycerols (DGs), followed by fatty acyls, sterols, and prenols. Overall, the data indicated an association between the clinical grading of acne and sebaceous lipid fingerprints and highlighted DGs as more abundant in sebum from adolescents affected with acne. Acne is a complex and multifactorial skin disorder targeting the pilosebaceous unit, wherein the sebaceous gland produces and secretes a lipid-rich mixture known as sebum. Increased sebogenesis is pivotal in the pathogenesis of acne and predisposes skin to deregulated inflammatory responses (1Zouboulis C.C. Eady A. Philpott M. Goldsmith L.A. Orfanos C. Cunliffe W.C. Rosenfield R. What is the pathogenesis of acne?.Exp. Dermatol. 2005; 14: 143-152Crossref PubMed Google Scholar, 2Zouboulis C.C. Jourdan E. Picardo M. Acne is an inflammatory disease and alterations of sebum composition initiate acne lesions.J. Eur. Acad. Dermatol. Venereol. 2014; 28: 527-532Crossref PubMed Scopus (160) Google Scholar, 3Shi V.Y. Leo M. Hassoun L. Chahal D.S. Maibach H.I. Sivamani R.K. Role of sebaceous glands in inflammatory dermatoses.J. Am. Acad. Dermatol. 2015; 73: 856-863Abstract Full Text Full Text PDF PubMed Scopus (70) Google Scholar). Quantitative and qualitative modifications of sebaceous lipids likely trigger inflammatory responses causative of acne lesions commonly called comedones (4Ottaviani M. Camera E. Picardo M. Lipid mediators in acne.Mediators Inflamm. 2010; 2010: 858176Crossref PubMed Scopus (95) Google Scholar, 5Picardo M. Ottaviani M. Camera E. Mastrofrancesco A. Sebaceous gland lipids.Dermatoendocrinol. 2009; 1: 68-71Crossref PubMed Google Scholar, 6Jeremy A.H. Holland D.B. Roberts S.G. Thomson K.F. Cunliffe W.J. Inflammatory events are involved in acne lesion initiation.J. Invest. Dermatol. 2003; 121: 20-27Abstract Full Text Full Text PDF PubMed Scopus (392) Google Scholar, 7Akaza N. Akamatsu H. Numata S. Matsusue M. Mashima Y. Miyawaki M. Yamada S. Yagami A. Nakata S. Matsunaga K. Fatty acid compositions of triglycerides and free fatty acids in sebum depend on amount of triglycerides, and do not differ in presence or absence of acne vulgaris.J. Dermatol. 2014; 41: 1069-1076Crossref PubMed Scopus (23) Google Scholar). Juvenile acne, which affects approximately 80% of adolescents in industrialized countries, develops during puberty when androgens and other endocrine factors stimulate sebum secretion, thereby generating conditions for development of lesions (8Taylor M. Gonzalez M. Porter R. Pathways to inflammation: acne pathophysiology.Eur. J. Dermatol. 2011; 21: 323-333Crossref PubMed Scopus (59) Google Scholar). Studies aimed at the identification of lipid alterations involved in acne pathogenesis date back to the early 70s to late 80s (9Powell E.W. Beveridge G.W. Sebum excretion and sebum composition in adolescent men with and without acne vulgaris.Br. J. Dermatol. 1970; 82: 243-249Crossref PubMed Scopus (51) Google Scholar, 10Cunliffe W.J. The relationship between surface lipid composition and acne vulgaris.Br. J. Dermatol. 1971; 85: 86-89Crossref PubMed Scopus (10) Google Scholar, 11Morello A.M. Downing D.T. Strauss J.S. Octadecadienoic acids in the skin surface lipids of acne patients and normal subjects.J. Invest. Dermatol. 1976; 66: 319-323Abstract Full Text PDF PubMed Scopus (56) Google Scholar, 12Pochi P.E. Strauss J.S. Downing D.T. Skin surface lipid composition, acne, pubertal development, and urinary excretion of testosterone and 17-ketosteroids in children.J. Invest. Dermatol. 1977; 69: 485-489Abstract Full Text PDF PubMed Scopus (54) Google Scholar, 13Rebillo T. Hawk J.L. Skin surface glycerol levels in acne vulgaris.J. Invest. Dermatol. 1978; 70: 352-354Abstract Full Text PDF PubMed Scopus (17) Google Scholar, 14Nordstrom K.M. Labows J.N. McGinley K.J. Leyden J.J. Characterization of wax esters, triglycerides, and free fatty acids of follicular casts.J. Invest. Dermatol. 1986; 86: 700-705Abstract Full Text PDF PubMed Scopus (41) Google Scholar); however, the complexity of the sebum composition has represented a major limitation to understanding lipid modifications involved with acne. The recent availability of methods with improved analytical performance and high-throughput opens new perspectives to understanding sebum alterations in acne. GC-MS has been widely employed to investigate sebum lipids. GC-MS conditions are suitable for analysis of FFAs, squalene, and cholesterol and also allow for detection of wax esters and sterol oxidation products (15Girod A. Weyermann C. Lipid composition of fingermark residue and donor classification using GC/MS.Forensic Sci. Int. 2014; 238: 68-82Crossref PubMed Scopus (62) Google Scholar). The latest advances in GC-MS analysis have provided structural information and detailed fingerprints of intact sebaceous lipids (16Michael-Jubeli R. Bleton J. Baillet-Guffroy A. High-temperature gas chromatography-mass spectrometry for skin surface lipids profiling.J. Lipid Res. 2011; 52: 143-151Abstract Full Text Full Text PDF PubMed Scopus (53) Google Scholar, 17Michael-Jubeli R. Tfayli A. Bleton J. Baillet-Guffroy A. Chemometric approach for investigating the skin surface lipids (SSLs) composition: influence of geographical localization.Eur. J. Dermatol. 2011; 21: 63-71Crossref PubMed Scopus (8) Google Scholar). Coupling the separation power of HPLC with high-resolution MS has also proved to meet the difficult task of analyzing intact structures of sebaceous mixture components (18Rolim A.E. Henrique-Araújo R. Ferraz E.G. de Araújo Alves Dultra F.K. Fernandez L.G. Lipidomics in the study of lipid metabolism: current perspectives in the omic sciences.Gene. 2015; 554: 131-139Crossref PubMed Scopus (61) Google Scholar). On the other hand, no separative spectroscopic methods based on NMR have been used for the quantitative analysis of sebum lipids independent of their fatty acid composition (19Robosky L.C. Wade K. Woolson D. Baker J.D. Manning M.L. Gage D.A. Reily M.D. Quantitative evaluation of sebum lipid components with nuclear magnetic resonance.J. Lipid Res. 2008; 49: 686-692Abstract Full Text Full Text PDF PubMed Scopus (52) Google Scholar). We have defined a method for analysis of intact lipids in sebum that requires minimal manipulation of the sample and is both efficient and accurate. The HPLC-MS method enabled detection of intact neutral lipids belonging to chemical classes known to enrich sebum, including esters of glycerol, cholesterol, and wax, as well as squalene and FFAs (20Camera E. Ludovici M. Galante M. Sinagra J.L. Picardo M. Comprehensive analysis of the major lipid classes in sebum by rapid resolution high-performance liquid chromatography and electrospray mass spectrometry.J. Lipid Res. 2010; 51: 3377-3388Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar). For the purposes of the current study, this analytical method offered a suitable basis for semi-targeted analysis of characterized lipids in acne sebum and an untargeted approach to address the sebaceous lipidome in its entirety. The primary focus of the current study was to identify differences in the lipid composition of healthy and acne sebum in order to understand the role of sebum lipid alterations in this disease. Discovery of new disease markers and clues on the enzymatic pathways involved in acne pathogenesis are crucial to characterizing new targets for successful acne management in the near future. Methanol (MeOH), isopropyl alcohol (iPrOH), acetone (Me2O), and ethanol (EtOH) were of HPLC-MS grade and were purchased from Merck (Darmstadt, Germany), whereas HPLC-MS grade ammonium formate (HCO2NH4) was purchased in granular form from Fluka (Buchs, St. Gallen, Switzerland). Deuterated cholesterol-2,2,3,4,4,6-d6 (d6CH) and glyceryl-d5-trihexadecanoate (d5TG 48:0), selected as the internal standards (ISTDs), were purchased from CDN Isotopes Inc., Pointe-Claire, Quebec, Canada. The study was approved by the Institutional Ethics Committee and was conducted in agreement with the principles expressed in the Declaration of Helsinki. Female and male Caucasian adolescents were enrolled consecutively in the first 4 months of the year. Acne patients were recruited among subjects at their first access to the Acne Unit. The control group was enrolled among healthy subjects undergoing mole check in the same Institution. Informed consent was obtained from their parents or a legal representative. The acne severity scores and the clinical grading were assessed by dermatologists. Criteria of inclusions were absence of systemic diseases and of skin disorders other than acne. Oral or topic pharmacological treatments at the first visit or up to 2 weeks before the examination, smoking, and use of sun beds were exclusion criteria. Sixty-one subjects (32 females, age 14.1 ± 1.7 years; 29 males, age 14.4 ± 1.4 years) resulted as eligible for the study. None of the enrolled patients had a history of isotretinoin use. The intensity of clinical manifestations was evaluated on the face areas according to the Global European Acne (GEA) severity (GEAS) criteria (21Dréno B. Poli F. Pawin H. Beylot C. Faure M. Chivot M. Auffret N. Moyse D. Ballanger F. Revuz J. Development and evaluation of a Global Acne Severity Scale (GEA Scale) suitable for France and Europe.J. Eur. Acad. Dermatol. Venereol. 2011; 25: 43-48Crossref PubMed Scopus (106) Google Scholar). Clinical classification of the subjects was based on four grades scales adapted from the reported GEAS scores. Briefly, subjects were classified as non-affected (clear) when no or very few lesions were clinically observable (0 ≤ GEAS < 1). Acne patients were subdivided into mild, moderate, and severe groups. Patients with GEA scores between 1 and 2 (1 ≤ GEAS ≤ 2) were referred to as the mild group, whereas patients with GEAS 3 and 4–5 were classified as moderate and severe groups, respectively. Seventeen, 13, 19, and 12 subjects were included in the clear (10 female/7 male), mild (8 female/5 male), moderate (12 female/7 male), and severe (2 female/10 male) groups, respectively. Noninvasive sampling of sebum lipids was performed from central foreheads as previously described (20Camera E. Ludovici M. Galante M. Sinagra J.L. Picardo M. Comprehensive analysis of the major lipid classes in sebum by rapid resolution high-performance liquid chromatography and electrospray mass spectrometry.J. Lipid Res. 2010; 51: 3377-3388Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar). Briefly, the sampling surface was cleansed with a gauze pad wetted with 70% EtOH solution beforehand. After allowing the surface to dry, two sebum-adsorbent tapes (SebutapesTM; CuDerm Corp., Dallas, TX), which had been preweighed beforehand, were held onto the skin for 30 min. The tapes were reweighted for the gravimetric assessment of sebum excretion rates (SERs, micrograms per square centimeter per minute). Supplementary Table 1 and supplementary Fig. 1 summarize demographics, distribution of subjects according to the GEAS, and average SER for the respective groups. Extraction of tapes was performed according to the previously defined procedure (20Camera E. Ludovici M. Galante M. Sinagra J.L. Picardo M. Comprehensive analysis of the major lipid classes in sebum by rapid resolution high-performance liquid chromatography and electrospray mass spectrometry.J. Lipid Res. 2010; 51: 3377-3388Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar). Briefly, tapes were extracted with 10 ml of absolute EtOH containing 0.025% butylated hydroxytoluene. EtOH was evaporated almost to dryness under nitrogen. The concentrated EtOH solution was subjected to liquid-liquid extraction with ethyl acetate to abate matrix-derived material. The final lipid extract was dissolved in Me2O/MeOH/iPrOH (40/40/20) to obtain a solution of sebum at the final weight per volume concentration of 5 mg/ml. Before analysis, the concentrated solutions were diluted 1:5 in Me2O/MeOH/iPrOH (40/40/20). A mixture of ISTDs, namely d6CH and d5TG 48:0, were added at a final concentration of 50 and 10 μM, respectively, to control the analytical performance and to calculate the relative abundance of detected lipids. The chromatographic apparatus consisted of a 1200 series rapid-resolution (RR)-HPLC (Agilent Technologies, Santa Clara, CA), equipped with a degasser, a binary pump, an autosampler, and a thermostated column compartment from the same manufacturer. For the RR-reversed-phase (RP)-HPLC separation, two columns were connected in series. The first one was a Zorbax SB-C8 RR cartridge 2.1 × 30 mm 3.5 μm particle size (Agilent Technologies, Santa Clara, CA) and the second one was a Zorbax SB-C8 RR HT 2.1 × 100 mm 1.8 μm particle size with a maximal operational backpressure at 600 Bar (from the same manufacturer). Sebum samples and authentic standards were eluted with a binary gradient of (A) 5 mM ammonium formate in MilliQ water (18.2 Ω) and (B) MeOH/iPrOH 95/5. The mobile phases were filtered through 0.45 μm glass filters and continuously degassed under vacuum. The elution program was as follows: 0–1 min 70% B, 20 min 99% B, 20–32 min 99% B, 34 min 100% B, 34–44 min 100% B, 56 min 70% B. A post run of 4 min at 70% B was included. The flow rate was maintained at 0.25 ml/min during the entire HPLC run and post-run time (4 min). The column was thermostated at 60°C. The injection volume was 0.5 μl. The injector needle was washed with the mobile phase in the wash port during the HPLC runs. The eluent outlet was connected to two different MS analyzers for the detection and characterization. Accurate mass measurements were conducted with a G6220A series TOF-MS (Agilent Technologies, Santa Clara, CA) equipped with an ESI interface operating in the positive ion mode. Analytes eluted from the RR-RP-HPLC system were introduced into the TOF-MS apparatus at the operating chromatographic flow rate (see chromatographic conditions described in the HPLC section). Nitrogen was used as the nebulizing and desolvation gas. The temperature and the flow of the drying gas were 350°C and 10 l/min, respectively. The capillary and the cone voltage were 4,000 and 60 V, respectively. Scan mode TOF mass spectra were acquired in the positive and negative ion mode by using the TOF at 10,000 mass resolving power for scans over the range from m/z 100 to 1,000. To enhance accurate mass measurement for the ion species, a reference solution of two compounds with m/z 121.050873 and 922.009798, respectively, was vaporized in continuum in the spray chamber by means of a separate nebulizer. LC-MS-ESI TOF data were acquired and deconvoluted into individual chemical peaks using the Mass HunterTM acquisition software (B.01.03 version; Agilent Technologies). Untargeted and semi-targeted mining of the LC-MS-ESI TOF data were performed with the molecular feature extraction algorithm in the Mass Hunter software. Due to the nature of the sebum samples, quality controls could not be prepared from the study samples. Thus, instrumental drifts and responses were controlled with the help of lipid standard mixtures and the ISTDs. The features extracted were exported into a compound exchange format reporting the retention time, the accurate mass, and the absolute abundance for each entity to be processed in the subsequent chemometric analysis. Alignment of both m/z and retention time was performed against the two ISTDs, d6CH and d5TG, whereas normalization of peak intensities was performed against d6CH. Mass Profiler Professional (MPP) (version 12.6.1; Agilent Technologies) followed by a statistical procedure developed in-house (22Baroni M. Costantino G. Cruciani G. Riganelli D. Valigi R. Clementi S. Generating optimal linear PLS estimations (GOLPE): an advanced chemometric tool for handling 3D-QSAR problems.Quantitative Structure-Activity Relationships. 1993; 12: 9-20Crossref Scopus (444) Google Scholar) were used to process the above data. Retention times were aligned by setting a retention time window of 0.6 min, whereas mass-to-charge ratio (m/z) binning was performed by setting windows at 100 ppm. Grouping of biological replicates was set according to the clinical classification of patients with acne. Absolute abundance of each entity was normalized by the absolute abundance of the d6CH ISTD. Data were filtered by frequency of detection, which reflects the number of samples that presented particular features. A frequency filter was applied to data extracted from MPP and only entities present in 75% of samples belonging to at least one of the investigated groups were retained for the statistical analysis. Fold changes of filtered entities were compared between subsets of the studied population by volcano plots in the MPP tools. Fold changes with P < 0.05 after the Bonferroni's correction were considered as significant. Identification of entities within the MPP workflow was performed based on the METLIN Metabolomics Database (http://metlin.scripps.edu/). Golpe procedure was used to perform partial least squares (PLS) algorithm (23Wold S. Sjostrom M. PLS-regression: a basic tool of chemometrics.Chemom. Intell. Lab. Syst. 2001; 58: 109-130Crossref Scopus (6841) Google Scholar). After a preliminary screening focused on enlightening the main differences in the expression of lipid species among the investigated grades of acne (see Data analysis and Results for further details), clear and moderate acne samples were taken as representative of non-affected and affected condition, respectively. Thus, PLS algorithm was performed assigning a categorical activity of −1 for clear and 1 for moderate samples and considering the log2 of the obtained peaks area. Pareto scaling (24Eriksson, L., E. Johansson, N. Kettaneh-Wold, and S. Wold, . 1999. Scaling. In Introduction to Multi- and Megavariate Data Analysis using Projection Methods (PCA & PLS). Umetrics, Sweden. 213–225.Google Scholar) was used to pretreat the data. PLS weights plot has been analyzed in order to select the most important entities responsible for the discrimination between unaffected and moderate acne samples. The selected entities were then identified searching for a correspondence of exact mass in the LIPID MAPS (http://www.lipidmaps.org/) structure database (25Sud M. Fahy E. Cotter D. Brown A. Dennis E.A. Glass C.K. Merrill Jr., A.H. Murphy R.C. Raetz C.R. Russell D.W. et al.LMSD: LIPID MAPS structure database.Nucleic Acids Res. 2007; 35: D527-D532Crossref PubMed Scopus (769) Google Scholar). Statistical significance was calculated by Student's t-test or ANOVA following post hoc Tukey's honestly significant difference (HSD) test using XLSTAT version 2015.6.01.24027 (Addinosoft, Paris, France). A statistical probability of P < 0.05 was considered significant. Two letter abbreviations for lipids were used, consistent with the literature (26Fahy E. Subramaniam S. Brown H.A. Glass C.K. Merrill Jr., A.H. Murphy R.C. Raetz C.R. Russell D.W. Seyama Y. Shaw W. et al.A comprehensive classification system for lipids.J. Lipid Res. 2005; 46: 839-861Abstract Full Text Full Text PDF PubMed Scopus (1127) Google Scholar, 27Fahy E. Subramaniam S. Murphy R.C. Nishijima M. Raetz C.R. Shimizu T. Spener F. van Meer G. Wakelam M.J. Dennis E.A. Update of the LIPID MAPS comprehensive classification system for lipids.J. Lipid Res. 2009; 50: S9-S14Abstract Full Text Full Text PDF PubMed Scopus (1054) Google Scholar). Notation of fatty acids, monoacylglycerols, diacylglycerols (DGs), and TGs included the total number of carbon atoms arising from the fatty acyls and the total number of carbon-carbon double bonds in the fatty acyls, independent of their distribution. Participants in the current study included adolescent subjects with juvenile acne, the most common type of the disease. Supplementary Table 1 shows the distribution of age, gender, and clinical grades (clear, mild, moderate, and severe ones) of participants. Acne severity groups were similarly distributed in terms of age and gender. Mild and severe acne were observed with higher frequency in females and males, respectively. To evaluate sebometry, a parameter associated with acne pathogenesis, the SER, was determined for each participant. Distribution of SER values in the clinical groups is provided in a scattergram in supplementary Fig. 1. The SER was not significantly different in mild and moderate acne compared with unaffected sebum. In contrast, severe acne was associated with a significantly higher SER in both genders. To minimize the effects of SER values on sebum chemometrics, all samples were analyzed with the same weight to volume concentration of sebum in order to compare lipid abundance profiles. The applied analytical procedure allowed for the untargeted detection of lipid species expressed in the sebum samples. The minimal number of investigated lipid species was obtained by retaining those species that were detected in 75% of samples in at least one condition, which was coincident with the acne severity grade. The applied filtering retrieved 1,492 entities. Measurement of intensity ratios was used to establish relative abundance within a single sample. Intensity ratios were obtained by normalizing the peak area of detected species by the peak area of the d6CH ISTD. To show differences in the expression of untargeted lipid species in sebum from subjects with different acne grades, we plotted fold-changes in the relative abundance of multiple lipids versus sebum from unaffected donors. Thus, we generated volcano plots of normalized intensities of the 1,492 entities in pair-wise comparisons of the sebum lipidome in mild, moderate, and severe acne with that of unaffected subjects (i.e., clear) as a function of estimated effect size (in log2 fold-change) and the statistical significance (P < 0.05, Bonferroni correction; Fig. 1). Gray-colored entities on both sides indicate no significant change with respect to control sebum. Red-colored entities on the left and the right indicate that the normalized intensities of the corresponding entities were statistically lower and higher, respectively, in affected sebum. Changes in the relative abundance of a small number of lipid species were statistically different in sebum from mild grades of acne. In contrast, moderate acne showed significantly greater expression of several lipids compared with control sebum. In sebum from severe acne, we detected both significant decreases and increases in the relative abundance of numerous sebum lipids. The differences highlighted in the volcano plots were consistent with the extent of acne severity. An Eulero-Venn diagram of entities significantly perturbed in acne sebum showed commonalities between the different acne grades (supplementary Fig. 2). In particular, among lipids with a fold-change ≥2 with respect to unaffected sebum, moderate acne shared one lipid in common with mild acne and nine with severe acne. Class assignment of these lipids was completed using an in-house algorithm designed to match the exact mass of each entity with entries of the LIPID MAPS structure database. Annotation of the shared entities is reported in supplementary Table 2. The epidemiology of moderate acne (28Bhate K. Williams H.C. Epidemiology of acne vulgaris.Br. J. Dermatol. 2013; 168: 474-485Crossref PubMed Scopus (475) Google Scholar) combined with the outcomes of above comparisons were the basis for its selection as a representative of overall changes associated with average juvenile acne. Thus, the sebum lipidome in moderate acne was further investigated with multivariate approaches. PLS algorithm was performed in order to reinforce the separation between controls and moderate acne in terms of sebum lipid profile. The PLS training set was built selecting samples from 11 clear and 13 moderate subjects. The remaining 12 samples, half from controls and half from the moderate acne group, were then used as a test set to assess the goodness-of-fit and predictive ability of the proposed model. The PLS score plot of the obtained model [PLS first latent variable (LV1) versus the PLS second latent variable (LV2)] is reported in Fig. 2A, which shows obvious separation between control and moderate acne sebum samples, suggestive of significant differences in the lipidome of the two groups. The correlation coefficient of recomputed versus experimental PLS activity reached 0.92 at the LV2. As an external validation, the test set was projected on the model and found to agree perfectly with the experimental results. Indeed, the six samples belonging to unaffected subjects were correctly projected among clear samples of the training set, while the six samples belonging to subjects with moderate acne were projected among the moderate samples (Fig. 2B). Having assessed both the goodness and predictive ability of the obtained PLS model, a new global model was built in that was comprised of both training and test sets in order to extract the desired sebum lipidome-activity relationship. For the obtained global model (Fig. 3A), a recalculated/experimental correlation coefficient of 0.88 was achieved. The PLS weights plot is reported in Fig. 3B. In general, PLS weights plot analysis allows a straightforward individuation of variables that are most correlated with the proposed activity. In our case, among the entities responsible for the differences in sebum lipidome profiles between the examined groups, the entities highlighted in green in Fig. 3B were those more positively correlated to the acne activity (red square). The pie charts in Fig. 4 illustrate the results of this first identification by showing the distribution of the assigned lipid class. The majority of differentiating lipid species were glycerolipids, followed by fatty acyls, sterols, and prenols. Wax esters and squalene-related compounds (triterpenes) accounted for a small percentage of the total discriminating lipids. The fatty acyl family also included FFAs that are normally present in sebum. Consistent with the literature, squalene and members of the FFA family were found expressed at higher levels in moderate acne (data not shown) (13Rebillo T. Hawk J.L. Skin surface glycerol levels in acne vulgaris.J. Invest. Dermatol. 1978; 70: 352-354Abstract Full Text PDF PubMed Scopus (17) Google Scholar, 29Pappas A. Johnsen S. Liu J.C. Eisinger M. Sebum analysis of individuals with and without acne.Dermatoendocrinol. 2009; 1: 157-161Crossref PubMed Google Scholar). Because glycerolipids were predominant among the differentiating lipids (Fig. 4A) and DGs were found to be the most significantly affected glycerolipids in acne (Fig. 4B), we focused our attention on this group of compounds. The identity of DGs found to be expressed at significantly different levels in moderate acne were tentatively assigned using the LIPID MAPS database based on mass and isotope pattern, and then confirmed according to diagnostic ions in the MS spectrum (20Camera E. Ludovici M. Galante M. Sinagra J.L. Picardo M. Comprehensive analysis of the major lipid classes in sebum by rapid resolution high-performance liquid chromatography and electrospray mass spectrometry.J. Lipid Res. 2010; 51: 3377-3388Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar). The relative abundance of annotated DGs in clear and moderate acne groups are reported in Table 1. Sixteen out of the 17 annotated DGs were expressed at significantly higher levels in moderate acne (P < 0.05). On average, the levels of these DGs
Referência(s)