In this study, we have conducted comprehensive molecular analysis of a cohort of 265 Polish patients with cornification disorders. Novel pathogenic variants and modifying variants were detected in MeDOC patients. For part of the patients, we also preformed transcriptome analysis and described characteristics of ALOX12B vs. TGM1 deficient epidermis on mRNA and lipid levels.

In the cases analysed by NGS (245) the diagnostic yield of 87% was achieved, a value similar to the diagnostic efficacy obtained by others [12]. A higher detection rate of 89% was obtained in a small subgroup of X-linked ichthyosis patients, selected by an X-linked inheritance pattern, in whom only the STS MLPA test was performed.

The majority of patients were sent from different clinical centres across Poland and their clinical evaluation was limited or not available, making the unbiased, broad NGS-gene panel the method of choice. However, as the clinical data were not fully available, we had to trust the suspicion of MeDOC made by the referring clinician and therefore could not assess with certainty if in the 13% of unsolved cases, the right diagnostic methods were used. For example Nagtzaam et al., reported that in Dutch population in 20% of cases with clinical suspiction of X-linked ichthyosis, the genetic analysis showed variants in the other gene(s) than STS, leading to diagnosis change [13].

Based on the presented results, we can conclude that in the referred population ARCI patients were mostly represented, (126/254, 49%). In 56/126 (45%) cases, the ALOX12B gene was involved, making it the most common cause of ARCI in the tested population, while in 33/126 (26%) and 18/126 (14%) of cases the TGM1 and ALOXE3 genes were causal, respectively.

The overrepresentation of ALOX12B patients was also noticed among Czech ARCI patients, where 18/47 (38%) probands had ALOX12B pathogenic variants [14] and among Middle-Eastern ARCI patients of Muslim origin – 26% [15]. This is in contrast to several other European studies, that indicated the TGM1 variants as the ARCI leading cause, e.g. in Austria, Scandinavia, Galicia [12, 15,16,17]. These discrepancies could at least partially be explained by the high abundancy of recurrent ALOX12B and ALOXE3 variants in the Polish and regional (Czech) population [14]. Nevertheless, in the future, the availability of more population-specific genomic data will enable more precise estimations regarding variant frequency.

A large multicenter study on 224 ALOX12B and ALOXE3 patients of various ethnicities has shown that the p.Tyr521Cys variant accounts for 22% of all ALOX12B alleles [18]. In our group however, this variant was detected in 45% (52 of 115) of ALOX12B alleles (including the one allele found as a secondary finding). In Austrian and Czech cohorts this variant was also frequent and present in 6/14 (42%) and 10/36 (28%) of ALOX12B alleles, respectively [12, 14]. Although the origin of the p.Tyr521Cys variant is not known, our results suggest that a founder effect may have contributed to its high prevalence in Poland.

Of note, another frequent variant in ALOX12B: p.Ala597Glu was found in 11/115 alleles (9.5%) in our cohort, similarly to Czech patients (4/36 alleles – 10%), while in multicentre studies mentioned above, it was found in 12/282 (4%) alleles, confirming its regional higher prevalence [14, 18]. For ALOXE3, the most prevalent variant: p.Arg234Ter was detected in 22/37 (59%) alleles and 9/18 (50%) in Polish and Czech cohorts, respectively. Among various ethnicities this variant was present in 21% of ALOXE3 alleles. Interestingly, the p.Pro630Leu was present in our population only once, while in a multicentred and Czech cohort it was frequent (40% and 28% of ALOXE3 alleles, respectively) [14, 18].

The TGM1 results revealed, beside two well-known recurrent pathogenic variants: p.Arg126His and p.Val379Leu, an additional one: duplication of exons 10–14 (c.(1402 + 1_1401-1)_(2225 + 1_2226-1)dup), found in 9 probands. According to the HGMD database (v. Professional 2023.3), this variant was described only once in the literature before, but was detected using in silico NGS analysis in a single patient and its presence was not confirmed by DNA analysis [17]. We performed DNA quantitative tests in probands and their parents, which confirmed that the duplication was present (manuscript under preparation).

In 23 cases of 245 analysed by NGS (9%), we revealed additional variants in genes encoding proteins involved in epidermal barrier formation. Most of them were known pathogenic/likely pathogenic ones, however few were novel and few were VUS. There is no consensus in Poland and Europe as to the genetic testing and reporting of single allele incidental findings in autosomal recessive disorders. For example, in Poland carriership of pathogenic/likely pathogenic variants in the genes included in the phenotype-related panels are reported for the purpose of genetic counselling, while in the Netherlands often only causative variants are, and carrier status is not a part of the diagnostic result. Furthermore, there is an open question whether the single allele variants influence the phenotype. This aspect is intriguing in particular, considering that in 10/23 cases, the additional gene was FLG. This gene encodes filaggrin, the multitasking protein involved in epidermal differentiation, barrier formation and moisturising. FLG defects lead to ichthyosis vulgaris (IV), an autosomal semidominant condition with incomplete penetrance (83–96%) and variable expressivity [19, 20]. Biallelic pathogenic variants in FLG are associated with a relatively severe phenotype, however the vast majority of patients have a mild phenotype of IV and/or symptoms of atopy due to heterozygous mutation in FLG [20,21,22,23,24].

It has already been anecdotally observed that FLG variants exaggerate the symptoms of X-linked ichthyosis [25,26,27,28,29]. Moreover, in a Dutch cohort of 109 male patients with clinical suspicion of X-linked ichthyosis (XLI), FLG variants were concomitant with an STS pathogenic variant in 4% of cases [13]. Another report showed a FLG variant in a patient with ARCI caused by PNPLA1 pathogenic variants, however it is impossible to say if and how the phenotype was influenced by filaggrin defect [30]. It seems that FLG variants are not well recognized or described as incidental findings in ichthyosis types other than X-linked ichthyosis. It is safe to assume that the frequency of FLG pathogenic variants in the general European population is around 4–7% [20, 31]. However, in the available reports on ichthyosis cohorts, the presence of FLG pathogenic variants is either not reported or was not investigated.

In the remaining cases, we detected pathogenic variants in the other genes, that can be causative for different types of autosomal recessive ichthyosis (ARCI). Importantly, six of them were nonsense variants, meaning that only 50% of protein would be produced, provided the second allele is normal. As many of the ARCI subtypes share the same pathways/processes in the epidermis such an haploinsufficiency of the other protein might hypothetically have some influence on the phenotype. Such observations have already been made in other genodermatoses, e.g. epidermolysis bullosa [32,33,34]. Therefore further studies are needed to verify if this has clinical implications in ichthyosis.

The transcriptome analysis of 18 patients revealed overexpression of genes encoding proteins involved in immunity, epidermis development, differentiation and signalling. In contrary, downregulation of those involved in cell adhesion and motility, developmental process and differentiation, extracellular structure organization, signalling and communication was shown.

Recent transcriptomic analyses showed the Th17/Th22 mediated immune skewing in different clinical subtypes of ichthyosis [35, 36]. In accordance with their studies, the Th17/Th22 markers were also upregulated in our group. Six of them were among the top 50 Differentially Expressed Genes (DEG) (S100A9, S100A8, S100A7, SERPINB3, SERPINB4, S100A7A), but the overall number of upregulated genes involved in Th17/Th22 immune response included also IL36G, IL36RN, IL26, KLK10, EPN3, PI3, VNN3 and others. Importantly, despite general convergent gene expression profile, some discrepancies regarding individual genes are noticeable. For example, Kim detected IL17A/C, IL22 and IL-23R expression using RNA-seq, that was below the threshold in Malik analysis based on microarrays [36]. In our group, only IL17C and IL22RA1 (Subunit Alpha1 for IL22) were upregulated, although in IL22RA1 logFC was 1.6. This may reflect individual differences, methodological issues and heterogeneity of ichthyoses, as discussed below. Nevertheless, genes involved in cornification and barrier formation were concordantly upregulated in this and previous studies.

Furthermore, in contrast to Kim et al., who reported downregulation of lipid metabolism genes, we found upregulation of several genes involved in unsaturated fatty acid biosynthetic, oxoacid and lipid metabolic processes. However, when we compared our results with those of Kim, it turned out that of 277 genes upregulated in our group, around 40 were also upregulated in their patients and only a few were assigned as downregulated [36]. Moreover, a set of genes related to lipid biosynthesis was shown to be upregulated in TGM1-ARCI patients, as reported by Zhang et al. [37]

In order to have further insight into this issue, we compared the expression profile narrowing to two patients subgroups: ALOX12B-deficient (n = 5) and TGM1-deficient (n = 7) patients. Those subgroups were the most numerous in our group of patients. The proteins encoded by both genes are involved in production and formation of the corneocyte lipid envelope. Specifically, ALOX12B encodes arachidonate 12-lipoxygenase which is a key enzyme processing arachidonic acid (20:4n–6) during synthesis of barrier lipids, while transglutaminase 1 (TGM1) mediates the cross-linking of proteins in the corneocyte protein envelope and the attachment of the corneocyte lipid envelope [38].

Although differences between the number of upregulated and downregulated genes were shown when ALOX12B-deficient vs. controls and TGM1-deficient vs. controls DEGs sets were compared, we haven’t found DEGs between ALOX12B-deficient and TGM1-deficient patients, emphasizing that the expression pattern in those two groups are highly similar. In contrast, when we compared the fatty acids (FAs) profile in the epidermis taken from lesional skin of ALOX12B-deficient and TGM1-deficient patients, the differences in content of selected groups of FAs were visible. Recently we analysed the FAs profile in normal epidermis, which showed the differences in FA distribution with respect to age and skin location [10]. Therefore, the results of MeDOC patient analyses were narrowed with respect to collection site. We showed that the content of C16:1 in TGM1 was lower than in ALOX12B patients (inactive ALOX12B impairs processing of lipids biosynthesis). Conversely, the number of long chain FAs was higher in TGM1 patients. Though the tendency is visible, the statistical significance was achieved in only certain classes of FAs, probably due to small group sizes.

Previous studies on atopic dermatitis and the Netherton syndrome have already shown that the level of C16-18 adversely corresponds with epidermal barrier function [39, 40]. Importantly, it has been proposed, that the altered ratio of mono – vs. saturated long chain fatty acids affects production of substrates necessary for ceramide synthesis [41]. Moreover, in vitro studies provided data showing that increase in FAs and/or monounsaturated fatty acids (MUFA) content influences lipid organisation and, consequently, barrier permeability [40]. Our results also focus on this aspect showing that in ALOX12B patients, the amount of total even chain fatty acids (ECFA), odd chain fatty acids (OCFA) and saturated fatty acids (SFA) were diminished compared to the TGM1 group, with the exception of MUFA that was increased. Although detailed data on lipid composition in MeDOC skin are limited, it is well known that the ratio of main lipid types, as well as their content and organisation, vary in different ichthyosis types [42]. Whether phenotypic differences between patients with defects in ALOX12B and TGM1 genes are directly related to the different lipid levels and composition is highly intriguing. Unfortunately, due to limited data, we are currently unable to comment on this aspect. Of note, however, we are the first to publish comparative data on FA in epidermis of patients with ARCI. Hence, providing further insights into the complexity of lipid homeostasis in the ARCI epidermis, we also provoke novel (yet unanswered) questions.

We analysed large cohort of patients with different types of MeDOC, the clinical data in most cases were highly limited, though. In majority of cases the referral doctors indicated only “ichthyosis” or “keratoderma” on the referral form. Importantly, the material was referred to our laboratories by over 70 clinicians (geneticists, dermatologists, neonatologists) from over 30 different clinical canters across country. Therefore, in frame of this manuscript it was not possible to provide clinical descriptions. Importantly, MeDOCs are group of rare diseases, therefore the most valuable phenotypic evaluations, enabling the assessment of clinical nuances, would be those performed in specialized reference canters. Nevertheless, few cases of MeDOC patients presented in this article were phenotypically characterised before (as indicated in Additional file 3) [8, 11, 43].

Although the genetic basis of MeDOC are generally well established, the functional consequences of genetic defects in barrier formation genes is less recognised. Currently more tools are available for robust wide-ranging biochemical research, however there are still several limitations of such studies. Accordingly, there were also limitations of our study. Firstly, we encountered difficulties in collecting epidermal samples from healthy volunteers. We struggled to match age, sex and biopsy site in our control group. Therefore, the majority of control skin was mostly obtained during bariatric surgery from normal skin of feminine abdomen without clinical symptoms of MeDOC. Also genotyping was not performed in the control group.