Discussion

This study was performed to investigate the intercellular lipid and ceramide composition in canine skin by three different sampling methods. There are only a few publications focusing mainly on canine skin ceramides, comparing healthy dogs and those with atopic dermatitis.^15-18,39 Of these studies, only one was able to separate 11 ceramide classes by HPTLC and HPLC/MS.¹⁷ Using the HPTLC procedure applied in our study, it was possible to separate all major epidermal lipids, including the well-known eight bands of ceramides (Figure 1). By coupling HPTLC with MS, we were able to verify the identification of the ceramide bands. This was an important step because we used an uncommon procedure for HPTLC. In conclusion, we were able to identify nine ceramide classes (EOS, NS, EOP, NP, EOH, NH, AS, AP and AH) definitely using Rf values and m/z values by coupling HPTLC and ESI-MS results.

Additionally, there were some minor peaks in the bands of CerAS+NH and CerNS that could match CerAdS and CerNdS, respectively, as recently described by Yoon et al.¹⁷

Regarding the quantification of ceramides, we were able to show that the choice of the ceramide standard used for quantification greatly influences the determined ceramide composition, probably due to different degrees of charring. For the CerAP and CerNP standards, for example, we found similar species using MS. However, the corresponding masses of the CerAP species are higher than those of the CerNP species. This could be a reason for the higher charring degree of the CerAP standard compared to the CerNP standard. Thus, owing to the amount of carbon contained in the long-chained CerEOS, CerEOP and CerEOH, the respective standard of CerEOS should be used for quantification of these ceramide classes. The quantification of CerNP with the standard of CerAS and vice versa seems to be possible, because the median discrepancies between the respective measurements were not greater than 10%. For the same reason, the exchange of the standards of CerNS and CerAP for quantification of the respective bands might be possible. However, there are still standards missing for some ceramide classes, so quantification of these remains an uncertain factor when an assessment of the whole ceramide composition is desired.

Conversely, measurements of only these ceramides with a corresponding standard

may lead to incorrect results when information concerning the amount of total ceramides is needed. All of these factors should be considered in the interpretation of results dealing with ceramide composition.

There are numerous further factors influencing the outcome of epidermal lipid studies. In humans, changes of different lipid fractions have been shown to vary with the age of the study participants.^7,40 Regarding canine epidermal lipids, the breed is an additional factor that probably influences the lipid composition.⁴¹ Owing to the exclusion criteria complicating the recruitment of dogs that could participate in this study, we were not able to restrict the age or breed.

Another important factor is the body site of skin sampling. In human stratum corneum, variations in lipid content between different anatomical sites have been described^13,42, but no differences within the ceramide composition were found.⁴⁰ To the best of our knowledge, no investigations of different body sites have yet been published for canine skin lipid composition. We investigated the lipid composition at the inguinal region and at the caudal back for the following reasons. In dogs, the inguinal region is the most common site to sample because clinical signs of atopic dermatitis often manifest here. In contrast, the caudal back is usually not affected in atopic dermatitis and is different from the inguinal region with respect to skin thickness and the density of hair follicles. These anatomical differences might also explain the differences in the lipid compositions between the two body sites, especially regarding the higher percentage values of cholesterol sulphate, cholesteryl esters and triglycerides at the caudal back.^43,44 Within the ceramide profile, our results suggest differences between the body sites for the two ceramide classes of CerAH and CerAP. According to the available literature, this would be the first time that such differences have been observed. However, we were not able to draft statistical statements regarding the significance of these differences owing to the sample size of only five dogs. Hence, such differences might not exist with a study using a larger number of dogs. This issue should be investigated in the future, as well as the comparison of further body regions, especially regarding the local distribution of clinical signs in canine atopic dermatitis.

The comparison of the different sampling methods with respect to the lipid content revealed the striking observation that the highest total lipids were obtained from heat-separated epidermis. The explanation is obvious, in that the deeper epidermal layers will not be harvested by the other two skin sampling methods. For that reason, it did not make sense to compare the absolute values. However, a disadvantage of the relative comparison is the influence of all values on each other. This might explain the lower percentage amounts of ceramides, cholesterol and cholesteryl esters in heat-separated epidermis. The percentage amounts of phospholipids, free fatty acids and triglycerides are much higher in the biopsy samples, possibly due to contaminations by subcutaneous fat or just due to harvesting deeper epidermal layers, as mentioned above. Another source of variability is the difficult sample preparation regarding the heat separation of the epidermis from the dermis, because the biopsies were small and the separation of the epidermis was sometimes a challenge. The low levels of the intraclass correlation coefficients point to the problem of reproducibility.

Sampling skin lipids by the skin scrub method revealed a larger lipid amount per square centimetre of skin than by scraping. Additionally, the percentage amount of phospholipids (including PE) was higher in the samples collected by skin scrub.

These findings suggest that deeper layers than the stratum corneum were harvested by scrubbing the skin with n-hexane and ethanol (2:1 v/v), because the content of phospholipids increases in the deeper layers of the epidermis.⁴⁵ The lower percentage values of total ceramides and cholesterol in the scrapings compared to the skin scrubs might be due to the influence of the higher percentage amount of free fatty acids in the scrapings.

Significant differences between the sampling methods within the ceramide profile were rarely found. These differences may be less pronounced with a larger study population. Nevertheless, our findings regarding the canine ceramide profile are mainly in accordance with previous publications in dogs. Strikingly, we found amounts of CerEOH in the range of 14−20%, whereas Popa et al.⁴¹ could only find traces. Yoon et al.¹⁷ found about 9% of CerEOH. An explanation for such discrepancies may be differing HPTLC procedures, as well as different ceramide

standards used for quantification and different quantitative expressions (percentage of sample content, micrograms per milligram of protein or micrograms per square centimetre), but also the sampling method and the size of the study population might have influenced the results.

Given that the reproducibility of both scraping and skin scrub is comparably high and there are only little differences within lipid and ceramide profiles, both sampling methods seem to be suitable for collecting skin lipids. However, when the investigation of skin lipids in vivo is desired, some difficulties in both procedures may occur. Scraping of lesional skin could lead to contamination by blood, resulting in distorted lipid measurements. In contrast, struggling during skin scrubbing could lead to solvent loss and therefore to a lower lipid yield. This problem can be minimized by the use of sufficiently high-walled metal cylinders. Although the skin scrub method is able to cause some irritation due to the alcohol-containing solvent, it is a less invasive method than skin scraping. Although we have used the skin scrub technique only on dead dogs and although the practicability of using the skin scrub on live dogs remains to be proved, this should be the preferred method.

In conclusion, we demonstrated that several factors have to be considered in the interpretation of results of skin lipid measurements, particularly concerning the ceramides. Apart from the sampling method, the standards used for quantification have a great influence on the lipid composition determined. Additionally, the anatomical site might affect the results of canine skin lipid studies. Furthermore, this study revealed scraping and skin scrubbing as suitable and comparable sampling methods for canine epidermal lipids, whereas the analysis of heat-separated epidermis should not be the method of first choice, owing to poor reproducibility.

Whether the investigation of canine epidermal lipids collected by skin scrub in vivo will be suitable for the detection of variations in skin diseases remains to be proved.