Comprehensive profiling of Asian and Caucasian meibomian gland secretions reveals similar lipidomic signatures regardless of ethnicity


Sample quality control

Meibum samples collected from human donors by digital expression of the secretion from their eyelids are typically small (usually, less than 0.5 mg, even if pooled from two or more eyelids). Moreover, meibum is composed of several hundred lipid species that belong to a large number of different lipid classes, and are present in widely differing molar ratios. The chemical nature of meibum (a mixture of extremely long chain, mostly hydrophobic, lipids) necessitates the use of strong organic solvents for collecting and preparing samples for the analyses. The most common and effective solvents for dissolving lipids are chloroform and its mixtures with other solvents such as methanol, ethanol, iso-propanol (IPA) etc.. However, these solvents are also effective in dissolving other compounds, including plastic extractives, skin lipids, skin care products etc. To assure that meibum samples were free of contaminants of that nature, collected samples initially underwent a quality check using normal phase HPLC on a Diol column and ion trap MS in PIM as described in our earlier publications27,33 and in the Materials and Methods section. The samples were screened for the presence of typical plastic extractives36 such as oleamide (molecular formula C18H35NO; analytical ion with m/z 282.2797), other FA amides, oxidized Irgafos (C42H63O4P, m/z 663.4551), di-iso-nonyl phthalate (C26H42O4, m/z 419.3150) and others. Minor presence of these contaminants in any lipid sample is virtually unavoidable as they exist in some quantities in all commercial organic solvents, but their effects were minimized during post-processing the data as described in Materials and Methods. The samples that contained objectionably high levels of these compounds, which could not be corrected in post-processing (one Asian and one Caucasian samples), were disqualified from further examination.

Selection of mass spectrometric and chromatographic techniques

Though APCI MS was the method of choice in our previous experiments due to exceptional clarity of the APCI mass spectra (in PIM, most of the analytes produced only proton (M + H)+ adducts with no discernible formation of ammonium, sodium and potassium ones), for the purpose of this study we used both APCI and ESI techniques. The latter was included in our protocols in attempt to directly compare our results with those published by Lam et al.26, who used exclusively ESI to analyze meibum of Asian patients. For lipid separation, we used a RP-UPLC protocol that had been described in our recent publications31,34 and resulted in good separation of individual lipid species within multiple lipid classes, and was compatible with ESI and APCI protocols. The chloroform–methanol eluent utilized by Shui et al.41 and Lam et al.26 was incompatible with our instrumentation, and is much more toxic.

Characterization of lipids using high resolution mass spectrometry in MS1 and MS
E modes

In our previous studies28,29,30,31,33,36,37,42,43,44,45,46, we already conducted comprehensive structural characterization and quantitation of major human Meibomian lipids using HPLC/multistage ion trap APCI MSn, which made duplication of those experiments in this project unnecessary. Instead, for analyte identification in our current project we relied primarily on the high resolution MS spectra (~ 40,000 FWHM; Fig. 1a,b), while the automatic MSE function, which is available in the Synapt/MassLynx combo, was used as a supplemental approach (Fig. 1c,d). As an example, the first two Fig. 1a,b demonstrate an experimental (1a) and theoretical (1b) mass spectra of a meibomian wax ester with a molecular formula C42H82O2 (lignoceryl oleate) which produced a proton adduct (M + H)+ with a theoretical m/z 619.6393. Note that the portions of both spectra between m/z 622 and 625 were magnified 36x  to show the forth and the fifth isotope peaks of the compound. One can see that a cluster of at least five isotopomers (M + 1) through (M + 5) is detectable and can be used for identification purposes. The wax is available as a pure chemical standard, which allowed us to compare the MSE fragmentation pattern of the natural compound with its synthetic counterpart (Fig. 1c,d). Both compounds fragmented identically to produce a series of ions with m/z 283.2603 (protonated oleic acid), m/z 265.2513 (oleic acid − H2O + H+), and m/z 247.2429 (oleic acid − 2H2O + H+). This experiment, together with identical chromatographic retention times of both analytes (not shown), demonstrated that: (1) the major form of the natural wax C42H82O2 was indeed based on oleic acid, and (2) MSE could be used for verification of structures of compounds, if needed.

Figure 1

High resolution mass spectrometric analysis of human meibomian lipids using MS and MSE approaches. (a) Experimental mass spectrum of lignoceryl oleate; five isotopomers are shown; area between m/z 622 and 625 is magnified 36x. (b) Theoretical mass spectrum of lignoceryl oleate calculated using the isotope modeling utility of the MasLynx software package. (c) MSE fragmentation spectrum of meibomian lignoceryl oleate; (d) MSE fragmentation spectrum of authentic lignoceryl oleate.

Direct gross inter-group comparison of Asian and Caucasian meibum

Next, Asian and Caucasian meibum samples were compared using RP-UPLC/ESI MS in positive ion mode. For each study group, individual runs were combined into a single averaged output data file (ADF) using the “Combine All Files” routine and “Mean Peak Intensities” setting of the MassLynx software—ADF-Asians and ADF-Caucasians (ADF-A and ADF-C), —and analyzed as described below.

A side-by-side, “fingerprint” style comparison of ESI TIC generated from ADF-A and ADF-C data files produced the first clear evidence of their overall similarity: both types of samples replicated each other to the minute details (Fig. 2a). Importantly, the averaged total lipid content in the samples, estimated from their integrated TIC, was high and quite comparable in both groups, with a total ion current for ADF-A and ADF-C being (1.08 ± 0.02) × 107. This fact facilitated their direct side-by-side comparison. Then, observation high resolution ESI mass spectra for both types of samples were obtained from their ADF files (Fig. 2b,c). Notably, averaged mass spectra of Asian and Caucasian meibum were also found to be exceptionally similar with virtually the same ion patterns and signal intensities in both groups. Note that the portions of spectra from m/z 500 to 1,300 were magnified in post-processing 2x  to compensate for the lower signal intensity of ions in that area compared to the major ion m/z 369.3558.

Figure 2
figure2

Gross inter-group comparison of Asian and Caucasian meibum. (a) RP-UPLC/ESI PIM total ion current chromatograms of Asian (ADF-A, red) and Caucasian (ADF-C, black) meibum. (b) Observation mass spectrum of Asian meibum. (c) Observation mass spectrum of Caucasian meibum.

Most of meibomian lipids were characterized structurally and quantitatively in our previous studies28,29,30,31,33,36,37,42,43,44,45,46 and related publications by independent groups26,39,47,48. Therefore, those experiments will not be described here. However, the structures of major lipid analytes relevant to this manuscript were verified in automatic MSE fragmentation experiments49,50 which were a part of every LC/MS analysis conducted in this study. A list of major lipids that were observed in these experiments (Fig. 2b,c) is presented in Table S2. Those included Chl, CE, WE, TAG, OAHFA, DiAD and Chl-OAHFA, which will be discussed in more detail later in the manuscript. As in our previous publications27,37, only exceedingly minor (regularly below 0.1% of total meibum), and randomly varying, amounts of phospholipids and sphingomyelins were detected in the current study. Because of their low abundance, they will not be discussed in this paper.

Then, the CE fractions of both study groups were compared. In the conditions of ESI analysis, authentic CE produced a complex MS which was dominated by their common fragment (M – FA + H)+ and CE adducts of the (2M + NH4)+ nature. Also noticeable were ions (M + NH4)+, (M + Na)+, (2M + H)+, and (2M + Na)+, while (M + H)+ was the weakest of all, which hampered its use in CE identification and quantitation. Representative spectra of authentic cholesteryl lignocerate are shown in Fig. 3a–c. Other individual CE produced equally complex spectra. The strongest analytical ion of all was (M – FA + H)+. This common analytical ion with a theoretical m/z value of 369.3521 (C27H45; a proton adduct of dehydro-Chl) is formed from any CE due to its spontaneous fragmentation and a loss of a FA residue in the ion source of the mass spectrometer27,28,29,44. When the EIC of an equimolar 50 μM mixture of four authentic CE with C16:1, C18:1, C22:1, and C24:1 FA residues was overlaid with that of human meibum (Fig. 3d), the peaks of standards (red trace) clearly identified their natural counterparts in human meibum (black trace). The latter conclusion was verified by high resolution MS and MSE of individual lipids, as illustrated in Fig. 1 for WE. Note the high level of reproducibility of the analyses in our RP-UPLC/MS experiments19.

Figure 3
figure3

Gross characterization of the cholesteryl ester pools in Asian and Caucasian meibum. (ac) High resolution ESI PIM mass spectra of authentic cholesteryl lignocerate demonstrate formation of various adducts of the compound. (d) Superposition of extracted ion chromatograms of characteristic ion m/z 369.35 obtained for a 50 μM equimolar mixture of four authentic monounsaturated cholesteryl esters (red trace; authentic cholesteryl esters of palmitoleic C16:1, oleic C18:1, erucic C22:1-, and nervonic C24:1 fatty acids), with that of a representative human meibum sample (black trace). (e) Superposition Asian (ADF-A, red trace) and Caucasian (ADF-C, black trace) chromatograms of characteristic ion m/z 369.35 revealed their alikeness.

This approach was used to compare samples collected from humans of different ethnicities. Two ion chromatograms extracted from ADFs of Asian and Caucasian meibum (Fig. 3e) were virtually indistinguishable and demonstrated no significant differences between two ethnicities. Conveniently, the overall presence of CE in the samples, estimated from the integrated EIC of ion m/z 369.3521, was almost identical for both groups: the average total ion current for their (M – FA + H)+ ions was (7.08 ± 0.10) × 106. Only minor (≤ 9%) variations in the ratios of individual peaks were observed, which were well within a routinely observed variability in individual lipids in human meibum (± 15%, regardless of the study group19,29,51,52). Free Chl was virtually undetectable in these experiments in both types of meibum, though it was clearly observable in our earlier and current APCI experiments (see below) as a compound with an RT of about 6.5 ± 0.3 min. Notably, Chl was not detected in the ESI experiments by Lam et al.26 either. It seems that poor ionizability of Chl is one of the limitations of the ESI approach for meibum studies. A more detailed evaluation of meibomian CE was conducted using APCI MS and will be discussed later in the text.

Taken together, we found no evidence of noteworthy differences between CE of Asian and Caucasian meibum in RP-UPLC/ESI MS experiments: the lipids were qualitative and quantitatively the same in two ethnicities.

Then, meibomian WE, listed in Table S2, were evaluated. As an example, ESI MS data for lignoceryl oleate (monoisotopic MW 618.6311) and lignoceryl stearate (MW 620.6467) are illustrated in Fig. 4a–f. As a general rule, WE formed (M + H)+, (M + NH4)+ and (M + Na)+ adducts, whose relative balance was influenced by the experimental conditions. However, the intensity of WE proton adducts was sufficiently high for reliable detection of all analytes, if LC/MS conditions were kept unchanged. One can see that EIC of lignoceryl oleate produced at least two closely eluting chromatographic peaks with identical MS spectra, which indicated that there were at least two isobaric forms of the compound. Importantly, the slower eluting form (Fig. 4a, peak with RT of 22.20 min) co-eluted with authentic lignoceryl oleate (Fig. 4c, peak with RT of 22.16 min), while the major form of meibomian wax eluted faster and had a RT of 21.60 min. However, MS1 and MSE mass spectra of both forms were indistinguishable. The same conclusions were made for lignoceryl stearate (Fig. 4e,f). The first major UPLC peak with RT of 21.60 was produced by lignoceryl oleate [its third isotope peak has a theoretical m/z value of 621.6461 (Fig. 1a,b)], while the second double peak with RT of 24.1/24.60 min was true lignoceryl stearate whose (M+H)+ peak has a theoretical m/z of 621.6545. The reason for the presence of at least two isobaric forms of each WE is the existence in meibum branched WE33. Representative results for six major oleic-acid based meibomian WE—C41H81O2 (proton adduct; theoretical m/z 605.6233, trace 1), C42H83O2 (619.6389, trace 2), C43H85O2 (633.6545, trace 3), C44H87O2 (647.6702, trace 4), C45H88O2 (661.6858, trace 5), and C46H90O2 (675.7015, trace 6) are shown in Fig. 4g,h as EIC. Importantly, the instrument response rose linearly with the increase in the amount of injected sample (Fig. 4i).

Figure 4
figure4

Gross characterization of meibomian wax esters in Asian and Caucasian population. (a) RP-UPLC/ESI PIM analyses revealed that at least two isobaric form of lignoceryl oleate with different retention times co-exist in human meibum; extracted ion chromatogram of ion m/z 619.64 is shown. (b) High resolution mass spectrum of meibomian lignoceryl oleate. (c) RP-UPLC analysis of authentic straight chain lignoceryl oleate produced a single UPLC peak with a retention time that matched that of a slower peak from panel (a). (d) Mass spectrum of authentic lignoceryl oleate. (e) Extracted ion chromatogram of ion m/z 621.66 observed in human meibum produced four UPLC peaks, two of which were produced by (M + 2) isotopomers of lignoceryl oleate (retention times 21.60 min and 22.20 min), while the last two were produced by proton adducts of lignoceryl stearate; the peak 24.60 min co-eluted with authentic straight chain lignoceryl stearate (not shown). (f) Mass spectrum of meibomian lignoceryl oleate matched that of authentic compound. (g,h) Extracted ion chromatograms of six major wax esters of Asian (ADF-A, panel g) and Caucasian (ADF-C, panel h) meibum. (i) Linearity of the instrument response was verified for major analytes, such as lignoceryl oleate (shown). (j) Overall comparison of wax ester pools of Asian and Caucasian meibum revealed their high similarity.

Other major WE were analyzed in a similar fashion: PRA of major analytes of interest were calculated by integrating corresponding EIC of their (M + H)+ adducts. The data on four major classes of WE (saturated and mono-, di-, and tri-unsaturated ones) are summarized in Fig. 4j. Note that Lam et al.26 did not report any saturated WE in Asian samples. However, the latter lipids were identified and quantitated using high temperature GC/MSn33, and were also detectable in all types of samples in the current study. Our current data clearly demonstrated that the overall balance of various types of WE was almost identical in both races.

Finally, we compared the apparent balances of CE and WE in Asian and Caucasian meibum using ADF-A and ADF-C data files. Our starting assumption was that the total ion current for a particular analyte in a sample was a (quasi)linear function of the total ion current of all analytes in the sample. To verify it, the following approach was used. Firstly, the effect of increasing amounts of injected meibum sample on the total ion current measured by the detector was measured by integrating their TIC (Fig. 5a–d). Undeniably, the normalized peak areas and RT of major components were not altered, while absolute TIC rose linearly (r2 > 0.999) with the increase in the amount of the injected sample (Fig. 5e). Note that a typical meibum sample in our study was within the range of TIC peak areas shown in the graph. Secondly, EIC of meibomian wax lignoceryl oleate were generated from those TIC, integrated and corresponding peak areas were plotted as shown in Fig. 5f. Again, a (quasi)linear detector response with r2 = 0.998 was observed. Thirdly, plotting TIC peak areas vs. EIC peak areas demonstrated a perfectly linear relationship between the two (Fig. 5g). These tests were repeated for different combinations of samples and analytes (such as WE, CE, and TAG) with the same outcome, demonstrating that no matrix effects affected the data, and the instrument response was (quasi)linear regardless of the measured parameters. It appeared that TIC could be used as common denominators for comparing RA and PRA of lipids that were measured using their EIC.

Figure 5
figure5

Total ion current chromatograms (TIC) obtained in identical conditions can be used for normalization of meibum samples. (a–d) TIC produced by sequential injections of 0.2 μL, 0.5 μL, 0.7 μL, and 1.0 μL of a representative sample human meibum differed only in the signal-to-noise ratio. (e) TIC peak areas rose linearly with the amount of the injected sample. (f) Peak areas of lignoceryl oleate, measured as extracted ion chromatograms (EIC) of the ion m/z 619.64, rose linearly with the amount of the injected sample. (g) A linear correlation between TIC and EIC justified the use of TIC as common denominators for comparing different samples.

Using this approach, RA of CE in Asian and Caucasian samples, measured as the ratio of a sum of all peak areas in EIC of ion m/z 369.3517 (CE1, CE2, … CEn) from Fig. 3e, and a total ion current measured as a sum of all peaks in corresponding TIC (such as in Fig. 2a), according to Eq. (1). For Asian meibum, RA of CE was 6.6%, while for Caucasian—an almost identical 6.5%.

Similarly, RA of major WE were calculated. As WE had no common analytical ion, their individual m/z values of their (M + H)+ adducts were summed instead. Their total RA for Asian and Caucasian meibum were 6.7% and 6.9%, correspondingly.

Thus, the almost identical distribution of CE and WE in both types of meibum led us to a conclusion that the pools of CE and WE in Asian and Caucasian meibomian gland secretions were essentially the same. Minor differences were well within the limits of natural variability of meibomian lipids reported in our previous studies19,29,51,52, and experimental errors, and are unlikely to be of physiological significance.

Intra-group variability of Asian and Caucasian meibum lipid profiles

The experiments discussed in the preceding section provided strong evidence of the overall biochemical similarity of Asian and Caucasian meibum. However, the very nature of such an integrative approach that averaged the data for all samples and analytes for each group made it difficult to visualize and estimate possible intra-group differences in the Asian and the Caucasian populations. To gather information on the degree of intra-group variability of meibomian lipidomes, we conducted targeted (or “supervised”) lipidomic analysis of individual meibum samples using RP-UPLC/APCI-MS and ESI–MS.

Elution profiles of major individual CE, WE, Chl-OAHFA, TAG, DiAD and OAHFA were obtained for every study sample. The choice of compounds for evaluation was based on our previous publications on the topic.

As an example, EIC of a meibomian C24:1-CE with an experimental m/z 735.6955, its mass spectrum and a chromatogram and a mass spectrum of authentic cholesteryl nervonate are shown in Fig. 6a–d. The experimental spectra matched a theoretical MS spectrum of its (M + 1) to (M + 4) isotopomers (Fig. 6e). Using m/z values from Table S2, EIC of a range of individual CE were obtained, integrated, and compared using their PRA (Fig. 6f,g). Evidently, the profiles of individual meibomian CE were not affected by the ethnicity of the subjects, and the standard deviations were small. Also identical in both ethnicities was the ratio of Chl to total CE: 0.0123 in Asians and 0.0127 in Caucasians.

Figure 6
figure6

Inter- and intra-group variability of individual cholesteryl esters in Asian and Caucasian population. (a) Extracted APCI PIM ion chromatogram (EIC) of meibomian cholesteryl nervonate. (b) EIC of authentic cholesteryl nervonate. (c) High resolution mass spectrum of the meibomian ester. (d) High resolution spectrum of authentic cholesteryl nervonate. (e) Theoretical mass spectrum of cholesteryl nervonate. (f) Distribution of molecular species of saturated cholesteryl esters in Asian and Caucasian meibum (normalized data). (g) Unsaturated cholesteryl esters of Asian and Caucasian meibum.

Then, major saturated, mono- and di-unsaturated WE were evaluated (Fig. 7). All tested WE were expressed at nearly the same levels in both study groups. Interestingly, saturated WE produced a bell-shaped compound profile with a clear maximum at C42H82O2 and C43H84O2, while the compound profile of mono-unsaturated WE had a reproducible valley at C43H84O2 and two maxima at C42H82O2 and C44H86O2. The makeup of di-unsaturated WE was even more complex, clearly favoring compounds with an even number of carbons C42H80O2, C44H84O2, C46H88O2, and C48H92O2.

Figure 7
figure7

Inter- and intra-group variability of individual wax esters in Asian and Caucasian population. (a) Saturated wax esters (normalized). (b) Mono-unsaturated wax esters. (c) Di-unsaturated wax esters.

Next, the distribution of molecular species of major diesters—extremely long chain Chl-OAHFA and DiAD—was investigated (Figs. 8 and 9). For both ethnicities, the pools of Chl-OAHFA were dominated by di-unsaturated compounds with even numbers of carbon atoms in their OAHFA moieties (m/z 1,100 and 1,128). As with other tested classes of lipids, the expression levels of Chl-OAHFA were race-independent, and so were the levels of DiAD. Notably, all major DiADs were of di-, tri-, and tetra-unsaturated nature, some of which are shown in Table S2 and Fig. 9b.

Figure 8
figure8

Inter- and intra-group variability of individual cholesteryl esters of (O)-acylated ω-hydroxy fatty acids (Chl-OAHFA) in Asian and Caucasian population. (a) Distribution of molecular species of Chl-OAHFA (normalized). (b) Molecular structure of the major Chl-OAHFA in human meibum.

Figure 9
figure9

Inter- and intra-group variability of individual cholesteryl esters of diacylated α,ω-diols (DiAD) in Asian and Caucasian population. (a) Distribution of molecular species of DiAD (normalized). (b) Molecular structure of one of the major DiAD in human meibum.

TAG—a diverse group of nonpolar lipids that mainly fulfill the role of energy and carbon storage—were investigated and found to be identical in both races (Fig. 10). Unlike CE, WE, Chl-OAHFA, and DiAD, all TAG had almost exclusively C14 to barely C22 fatty acids in all positions: No extremely long chain FA were detected in any of the tested TAG species. The main TAG in all samples was triolein, which accounted for ≥ 40% of the pool.

Figure 10
figure10

Inter- and intra-group variability of individual cholesteryl esters of triacylglycerols (TAG) in Asian and Caucasian population. Triolein (m/z 885.79, detected as a proton adduct) was the main TAG in every sample of human meibum representing close to 40% of the TAG pool.

A combined graph for tested nonpolar lipids is shown in Fig. 11a. Asian samples were marginally enriched with CE, with a RA of (36 ± 4)% compared to (31 ± 4)% for Caucasians. Though statistically significant (p ≤ 0.05), the measured difference was an order of magnitude smaller than one would expect from earlier studies26.

Figure 11
figure11

Targeted and untargeted lipidomic analyses of Asian and Caucasian meibum. (a) Inter- and intra-group variability of major nonpolar lipid classes in Asian and Caucasian population. CE and WE dominated the meibomian lipidomes in both races and were present at similar levels. Note that apparent abundances are proportional to, but do not equate, molecular ratios. (b) A PCA biplot of study samples. Scores: Asian samples—blue dots; Caucasian samples—green dots. Loadings: red dots.

Close similarities between Asian and Caucasian meibum were confirmed using the unsupervised Principal Component Analysis (PCA) of the raw RP-UPLC/APCI-MS data files. The model required at least 11 scores to explain 95% of the variances (Supplemental Fig. S1). Thus, only {PC1 vs. PC2}and {PC1 vs. PC2 vs. PC3}graphs are presented in the paper. The PCA biplot for the two types of samples (Fig. 11b) demonstrated two highly overlapping clusters of scores (i.e. samples). A minor separation between Asian and Caucasian samples, and a tighter clustering of the Asian samples in the same two right quadrants, were attributed to the effect of Chl-containing compounds (m/z values of 369.3519, 1,128.0660, 1,009.9868 and others) and long chain mono- and di-unsaturated WE (m/z 647.6704, 659.6681, 661.6853, 673.6856, 689.7140, 699.7002, 701.7167, 715.7219, and 729.7484), while shorter chain WE (m/z 563.5758, 577.5900, 589.5915, 591.6074, 605.6231, 615.6073, 619.6392, 631.6379 and 633.6547) grouped in the left two quadrants. Their chemical assignments are shown in Table S2. However, when three principal components (PC1, PC2 and PC3) are plotted in a 3D graph (Fig. 12), the similarity between Asian and Caucasian meibum became even more evident, with just a few outliers falling beyond the Hotelling T2 ellipse. Note that intra-group sample-to-sample variability for both groups were of the same order of magnitude as the inter-group differences. Thus, supervised approaches are generally considered a better option for highly similar samples53.

Figure 12
figure12

A  3D {PC1 vs. PC2 vs. PC3} scores plot for Asian and Caucasian meibum samples. Note a high degree of overlapping of the samples of two kinds within the Hotelling T2 (0.95) range. Samples that are significantly different from the rest of the pool are located outside the sphere.

Finally, major polar lipids of meibum—OAHFA—were compared using negative ion mode MS. Alongside OAHFA, another interesting compound—cholesteryl sulfate (Chl-S, C27H46O4S, theoretical m/z of anion 465.3038)—was also monitored. All OAHFA detected in meibum of both ethnicities (Fig. 13a,b and Table S2) belong to the family of extremely long chain lipids with their mono-unsaturated and di-unsaturated ω-hydroxy FA moieties ranging from C26, at least, C36 (Fig. 13c). The major acylating FA were of C16 and C18 nature with one and two double bonds, while a much smaller percentage of OAHFA had tri-unsaturated FA moieties. As with other classes of lipids, there were no differences detected between Asian and Caucasian samples in the molecular distribution of various OAHFA species. To determine if the overall amounts of OAHFA were different or similar in meibum of two races, the OAHFA’s RA were calculated using Eq. (1). Their values were found to be almost identical for Asians and Caucasians—40.2% and 43.2%, respectively. However, the standard deviations for both races were rather high (10.8% for Asians and 16.2% for Caucasians), placing OAHFA amongst the most variable groups of meibomian lipids. Similar results were obtained for Chl-S—31.4 ± 16.2% for Asians and 30.1 ± 12.7% (mean ± SD) for Caucasians. To verify these conclusions, standard box plots for both ethnicities were generated (Fig. 12d,e). The Mann–Whitney Rank Sum test confirmed that there were no ethnicity-associated differences in the RA of Chl-S: their median values were 25.1% and 26.0% for Asians and Caucasians, respectively (Mann–Whitney U statistic = 319; p = 0.911). The same conclusion was made with regard to OAHFA: their median values were 39.9% for Asians and 38.9% for Caucasians (U = 310; p = 0.78). Note that just a handful of samples fell in the 10th and 90th percentile and could be considered outliers: the vast majority of the samples were in the 25th and 75th percentiles. Thus, irrespective of the implemented analytical techniques, no noticeable differences between Asian and Caucasian OAHFA and Chl-S were observed.

Figure 13
figure13

Inter- and intra-group variability of polar lipids in Asian and Caucasian population. (a) Distribution of molecular species of mono-unsaturated (O)-acylated ω-hydroxy fatty acids (OAHFA) (normalized). (b) Distribution of molecular species of di-unsaturated OAHFA. (c) Molecular structure of the major OAHFA in human meibum. (d) The box plot for the overall presence of cholesteryl sulfate in human meibum. (e) The box plot for the overall presence of OAHFA in human meibum. The Mann–Whitney Rank Sum tests confirmed that there were no ethnicity-associated differences in the distribution of Chl-S and OAHFA between races.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *