Genetic and environmental risk factors for vitiligo and melanoma in Pura Raza Español horses

1 Genetic and environmental risk factors for vitiligo and melanoma in Pura Raza Español 1 horses 2 3 M.J. SÁNCHEZ-GUERRERO, M. SOLɆ, P.J, AZOR, J. SÖLKNER ‡ and M. VALERA 4 Department of Agro–Forestry Sciences, ETSIA, University of Seville, Road Utrera Km 1, 41013 5 Seville, Spain. 6 †Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, P.O. 7 Box 7023, 75007, Uppsala, Sweden. 8 ‡Division of Livestock Sciences, University of Natural Resources and Life Sciences, 1180, Vienna, 9 Austria. 10 11 Keywords: dermatology, heritability; inbreeding; prevalence; Spanish Purebred horse 12 13 *Correspondence e–mail: [email protected] 14 15 Ethical animal research 16 The study was approved by the Ethical Committee of the University of Seville (Spain). A 17 representative of the National Association of Pura Raza Español Horse Breeders (ANCCE) gave 18 explicit informed consent for use of their data in this study. 19 20 Authors’ declaration of interests 21 No competing interests have been declared for any authors. 22 23 Sources of funding 24 The study was supported by grants of the National Association of Pura Raza Español Horse 25 Breeders 26 Acknowledgements 27 The authors wish to thank the National Association of Pura Raza Español Horse Breeders for 28 providing the data used in this study. 29 30 Authorship 31 2 M. Valera and M.J. Sánchez performed the study design. M.J. Sánchez, M. Valera were involved in 32 the execution, as well as preparation of the manuscript. P.J. Azor was involved in the data 33 collection. M. Solé participated in the analysis of data. J. Sölkner contributed to presentation and 34 discussion of data also corrected the language. All the authors read and approved the final version 35 of the manuscript. 36 37 Word count: 4862 38 39 3 Genetic and environmental risk factors for vitiligo and melanoma in Pura Raza Español 40 horses 41 42 Keywords: dermatology, heritability; inbreeding; prevalence; Spanish Purebred horse 43 44 45 Summary 46 47 Background: Vitiligo and melanoma are relatively common disorders in grey Pura Raza Español 48 horses (PRE) and horse breeds with grey-coloured coats. 49 Objectives: To determine the breed prevalence, environmental risks factors and estimate the 50 genetic parameters for vitiligo and melanoma in PRE. 51 Study design: Retrospective cohort study. 52 Methods: We analysed data from a large worldwide population of PRE. The database included the 53 vitiligo and melanoma scores, on either a four- or six-point linear scale, of 11,436 horses. Genetic 54 parameters were estimated using a Bayesian genetic animal model including the four associated 55 environmental risk factors as systematic effects. Inbreeding was used as a covariate, and animal and 56 residual effects were included as random effects. 57 Results: Of the horses included in the study, 2.8% and 20.5% showed some traces of vitiligo 58 around the eyes and mouth, respectively, while 1.6% showed varying degrees of melanoma. Age, 59 coat colour and inbreeding were significantly associated with the three outcomes studied. The 60 estimated heritability for the whole population was 0.09 (standard deviation, s.d. +0.019), 0.44 (s.d. 61 +0.031) and 0.13 (s.d. +0.037), for eye vitiligo score, nostril vitiligo score and melanoma scores 62 respectively. The genetic correlations ranged from 0.42 (s.d. +0.084) between eye and nostril 63 vitiligo score to 0.15 (s.d. +0.096) between nostril vitiligo and melanoma. 64 Main limitations: Vitiligo scores for the perianal regions were not collected. The veterinarian 65 responsible for each assessment was not recorded. 66 4 Conclusions: Vitiligo and melanoma are prevalent in this population and those environmental risk 67 factors and genetics both have an effect on the clinical expression of the diseases. These findings 68 may help to reduce prevalence through breeding programmes. 69 70 5 Introduction 71 Vitiligo is an acquired, non-contagious disorder, characterized by progressive patchy loss of skin 72 pigmentation. Vitiligo tends to cover hair and mucous membranes (due to a loss of melanocytes in 73 the affected areas) [1]. Despite the fact that the phenotype of vitiligo has been readily recognized 74 for thousands of years, the underlying pathobiology has remained largely unknown [2]. Similarly, 75 melanomas are usually pigmented neoplasm which often present multicentric malignancy in the 76 later stages [3]. They are most commonly found in grey horses and reach a prevalence of around 77 80% in horses of 15 years of age or older [4]. Studies in humans indicate that there is a genetic link 78 between melanoma and vitiligo [5,6]. Moreover, in horses, the dominant phenotype of greying with 79 age is associated with a high incidence of melanoma and vitiligo-like skin depigmentation [7]. 80 Since the Pura Raza Español (PRE) is a mainly grey breed, the animals are susceptible to vitiligo 81 and melanoma. Since 2012, it has been standard practice to record a vitiligo score (VS) and 82 melanoma score (MS) in PRE breed registers [8]. This provides the opportunity to estimate the 83 breed prevalence of these disorders, and to examine potential environmental risk and genetic factors 84 associated with these diseases. We hypothesised is that both VS and MS are prevalent in PRE 85 populations worldwide, and that both environmental risk factors and genetics play a significant role 86 in the development of these conditions. The specific aims of this work were to: (1) measure the 87 breed prevalence of vitiligo and melanoma in a worldwide population of PRE, including non-grey 88 horses, (2) establish the potential links with inbreeding, age, sex, coat colour and geographic area 89 for the development of vitiligo and melanoma in this breed and (3) estimate the heritability and 90 genetic correlations for VS and MS. A better understanding of how these factors contribute to the 91 development of vitiligo and melanoma is crucial to improve the management and breeding selection 92 of mainly grey horse breeds. 93 Material and methods 94 Population 95 6 The database included records of 11,436 animals (4,132 stallions and 7,304 mares) with a mean age 96 of 4.51 (s.d. 1.675) years (3–21 years old). These records were taken during the morphological 97 evaluation (between 2012 and 2016 from 36 countries) that all the PRE have once in their life 98 before being registered in the PRE-studbook [9]. Breeders are free to apply for this evaluation at 99 any time in the horse's life, and so animals of different ages are represented in the dataset. The 100 pedigree information (four complete generations, comprising 35,430 horses )was collected from the 101 official PRE-studbook. 102 Outcome variables: vitiligo and melanoma phenotype 103 The inspections were carried out by 18 previously-trained veterinarians who travel around the world 104 assessing horses for registration in the studbook. The VS was graded on a scale from 0 to 3 based 105 on inspection of the face [7] (Supplementary Table 1). VS was not scored for the perianal and anal 106 regions. Although the appraisers took three different face scores for vitiligo for eyes, mouth and 107 nostrils, the genetic correlation between the nostril and mouth scores was close to 1, so they were 108 included as one single trait (nostril and mouth vitiligo score; NVS) evaluated together, and eye 109 vitiligo score (EVS) separately. The degree of vitiligo was scored in the same test as the MS [3], 110 which was detected by inspection and palpation. For the MS, a modified classification system on a 111 scale from 0 to 5 was used, as described in [10] Supplementary Table 1. 112 Exposure variables 113 The potential environmental risk factors studied were: age, sex, coat colour and geographic area of 114 the stud (the region where the horse was located when it was examined for vitiligo and melanoma). 115 The number of horses examined for each potential environmental risk factor by populations is 116 shown in Supplementary Table 2. An association frequency analysis of all the environmental risk 117 factors (M-L Chi-square) was also carried out (Supplementary Table 3). In addition, the degree of 118 inbreeding, defined as the probability that an individual possesses two identical alleles by descent at 119 a randomly chosen locus [11], was also studied (Supplementary Table 4). 120 Descriptive statistics 121 7 All the analyses were performed after dividing the database into 3 datasets: subpopulation 1 (S1; 122 grey PRE: horses with three generations of ancestors with a grey coat colour (n=1719); 123 subpopulation 2 (S2; grey PRE: horses with at least one generation of non-grey ancestors in the last 124 three generations (n=3325)); subpopulation 3 (S3; non-grey PRE: horses with non-grey coat colour 125 in any of the last three generations of ancestors (n=6392)). 126 The Kolmogorov–Smirnov normality test was used to verify whether the outcomes conformed or 127 not to normal distributions. Next, a multivariate Generalized Non-linear Model (GLZ) was used to 128 examine associations between the potential environmental risk factors, the interactions and the 129 outcomes. The implemented GLZ method, is a robust general linear model which applies an ordinal 130 multinomial distribution through a generalized logit function instead of the usual normal 131 distribution. It also uses maximum likelihood methods in the generalized model to test whether a 132 reduced model provides the same fit as a full model. The goodness of fit criteria of the models 133 (Akaike Information Criterion and Bayesian Information Criterion) was also estimated. 134 Additionally the maximum likelihood chi-square test was performed to study relationships among 135 the environmental risk factors (Supplementary Table 3). A simple linear regression analysis was 136 carried out to describe the relationship between the outcomes and the inbreeding (Supplementary 137 Table 4). Inbreeding was computed using Endog v4.8 software [12]. The other analyses were 138 performed using the Statistica package for Windows, v.12.0 [13]. 139 Genetic model 140 Heritability coefficients, genetic correlations and estimated breeding values (EBV) were estimated 141 using restricted maximum likelihood (REML) methodology based on a multiple-trait animal model 142 which was resolved using a Bayesian approach with TM software [14]. The outcomes were used in 143 their linear scale (1 - 4 for EVS and NVS and 1 - 6 for MS, because TM is programmed to take 0 as 144 a missing value). The Bayesian approach allowed us to work with non-normal traits. The equation 145 in matrix notation to solve the mixed model was: 146 y= Xb+Zu +e, with: 147 8 2 2 u e N s s ææ ææ æ æ ææ æ÷ ææ æ÷ ææ æ÷ æ æ ææææ ææ u0 A0 , e0 0I : 148 where y is the vector of observations, X the incidence matrix of systematic effects, Z the incidence 149 matrix of animal genetic effects, b the vector of systematic effects, u the vector of direct animal 150 genetic effects, e the vector of residuals,  u2 the direct genetic variance,  e2 the residual variance, I 151 an identity matrix and A the numerator relationship matrix. 152 The fitted model included only the environmental risk factors which were statistically significant 153 for each outcome (Supplementary Table 2). The model included inbreeding as a covariate and the 154 following systematic effects for each specific trait: age (≤3; 4-6 and >6); sex (male or female); 155 geographic area (Spain, Europe, United States of America (USA) and Latin America), and coat 156 colour (grey, bay, black and chestnut). The Bayesian approach was carried out four times separately 157 for each population in a multivariate analysis with the three outcomes together. 158 Results 159 The breed prevalence of EVS, NVS and MS in the whole population was 2.8% (CI: 2.50%-3.10%), 160 20.5% (CI: 19.76%-21.24%) and 1.6% (CI: 1.37%-1.83%), respectively. In S1, the prevalence was 161 7.2% (CI: 5.98%-8.32%), 49.8% (CI: 47.44%-52.16%) and 4.0% (CI: 3.07%-4.92%); in S2, it was 162 3.6% (CI: 2.96%-4.93%), 31.6% (CI: 30.03%-33.18%) and 2.0% (CI: 1.52%-2.48%); and in S3, it 163 was 1.2% (CI: 0.93%-1.46%), 6.9% (CI: 6.27%-7.52%) and 0.8% (CI: 0.06%-1.01%) for EVS, 164 NVS and MS. 165 The prevalence of vitiligo and melanoma, the number of Pura Raza Español horses affected by each 166 phenotypic score divided by the environmental effects, the p-value (including 2-way interactions), 167 and the goodness of fit criteria (AIC and BIC) of the model are given in Table 1, Supplementary 168 Table 2 and Supplementary Figure 1. Sex was significantly associated with EVS (except in S3) and 169 NVS. Most of the interactions sex with age/geographic area/coat colour were not significant 170 although some were significant (the most important sex-age in S1 for NVS). Age was significantly 171 associated with EVS, NVS and MS for all the coat colours and S1 and S2, and the interaction age-172 9 coat colour was relevant in S1. For non-grey PRE, age was not a significant risk factor for EVS and 173 MS (only the interaction age coat colour was significant for EVS). 174 Coat colour was significantly associated with all the traits in the whole population (55.54% of the 175 entire current PRE pedigree have a grey coat). In fact, for vitiligo, the vast majority were grey (465 176 of 597 young horses studied with an NVS greater than 1). When only non-grey PRE were 177 considered, coat colour was not significantly associated with MS. Although the interaction age-coat 178 colour was significant for the three traits in the whole population, in 3 year-olds, the coat was not an 179 important factor: only 35 of the 3728 PRE of this age had a MS greater than 1, and of these 35, 18 180 did not have a grey coat (12 were bay, 7 black and 1 chestnut). The geographic area was 181 significantly associated with EVS, NVS and MS. Inbreeding was a significant risk factor in the 182 whole population and particularly for NVS in the grey PRE population. 183 Heritability coefficients and genetic correlations of all traits in the four scenarios considered in the 184 study are shown in Table 1. Heritability ranged from 0.88 (s.d. +0.057) for MS in S1 to 0.07 (s.d. 185 +0.018) for EVS in S2 and 0.07 (s.d. +0.022) for MS in the S3. The genetic correlation ranged from 186 0.09 (s.d. +0.202) for EVS-MS in S3 to 0. 57 (s.d. +0.127) for EVS-MS in S1. The EBV for the 187 three traits studied, classified according to phenotypic class, is represented in Figure 1. The EBV 188 ranged from 0.37 (s.d. +0.091) to -0.09 (s.d. +0.085) for EVS, 1.59 (s.d. +0.305) to -0.64 (s.d. 189 +0.328) for NVS and 0.60 (s.d. +0.189) to -0.07 (s.d. +0.072) for MS. 190 Discussion 191 This study has, for the first time, addressed a breed prevalence of vitiligo and melanoma using a 192 large worldwide population of PRE. Our results showed that mouth and nostril vitiligo are prevalent 193 in PRE: nearly 21% of the PRE studied of any sex, age, coat colour and worldwide area, were 194 affected. Eye vitiligo and Melanoma had lower prevalences. However, the prevalence of all these 195 defects was higher in grey horses with three generations of grey ancestors (S1) and lower in non-196 grey horses with non-grey ancestors in any of the last three generations (S3). There were differences 197 16 322 TABLE 2: Heritability coefficients on diagonal and genetic correlations on off-diagonal with 323 standard deviations among the vitiligo and melanoma scores in the four Pura Raza Español 324 subpopulations studied 325 EVS 1 NVS 2 MS 3 Whole population EVS 0.09 (0.019) 0.42 (0.084) 0.21 (0.156) NVS 0.44 (0.031) 0.15 (0.096) MS 0.13 (0.037) Subpopulation 1 EVS 0.61 (0.111) 0.51 (0.126) 0.57 (0.127) NVS 0.62 (0.072) 0.27 (0.085) MS 0.88 (0.057) Subpopulation 2 EVS 0.07 (0.018) 0.33 (0.164) -0.34 (0.334) NVS 0.47 (0.056) 0.10 (0.187) MS 0.16 (0.058) Subpopulation 3 EVS 0.09 (0.017) 0.18 (0.142) 0.09(0.202) NVS 0.15 (0.030) -0.10 (0.170) MS 0.07 (0.022) 1Eye vitiligo score 326 2 Nostril and mouth vitiligo score 327 3 Melanoma score 328 17 Subpopulation 1 (PRE greys with the three preceding generations with grey coat colour), 2 (PRE 329 greys with at least one of the three preceding generations with non-grey coat colour) and 330 Subpopulation 3 (Nongrey PRE with the three preceding generations with non-grey coat colour) 331 332 18 References 333 1. Taïeb, A. and Picardo, M. (2009) Clinical practice. Vitiligo. N. Engl. J. Med. 360, 160–9. 334 http://www.scopus.com/inward/record.url?eid=2-s2.0-58149383594&partnerID=tZOtx3y1. 335 Accessed November 8, 2015. 336 2. Zhang, Z. and Xiang, L.F. (2014) Genetic susceptibility to vitiligo: Recent progress from 337 genome-wide association studies. Dermatologica Sin. 32, 225–232. 338 3. Moore, J.S., Shaw, C., Shaw, E., Buechner-Maxwell, V., Scarratt, W.K., Crisman, M., Furr, 339 M. and Robertson, J. (2013) Melanoma in horses: Current perspectives. Equine Vet. Educ. 340 25. 341 4. McFadyean, J. (1933) Equine melanomatosis. J. Comp. Pathol. Ther., 46, 186–204. 342 5. Wankowicz-Kalinska, A., Poole, C. Le, Wijngaard, R. Van Den, Storkus, W.J. and Das, 343 P.K. (2003) Melanocyte-specific immune response in melanoma and vitiligo: Two faces of 344 the same coin? Pigment Cell Res. 16. 345 6. Jin, Y., Riccardi, S.L., Gowan, K., Fain, P.R. and Spritz, R.A. (2010) Fine-mapping of 346 vitiligo susceptibility loci on chromosomes 7 and 9 and interactions with NLRP1 (NALP1). 347 J. Invest. Dermatol. 130. 348 7. Curik, I., Druml, T., Seltenhammer, M., Sundström, E., Pielberg, G.R., Andersson, L. and 349 Sölkner, J. (2013) Complex inheritance of melanoma and pigmentation of coat and skin in 350 Grey horses. PLoS Genet. 9, e1003248. http://www.scopus.com/inward/record.url?eid=2-351 s2.0-84874766852&partnerID=tZOtx3y1. Accessed November 12, 2015. 352 8. ANCCE (2012) Specific Purebred Spanish Horse Stud Book Rules and Regulations (BOE no 353 127, May 28, 2012). 354 http://www.ancce.es/ver_contenido.php?id=TXT_DATOS&menu=1&i=en. 355 9. BOE-A-2017-4522 (2017) Resolución de 6 de abril de 2017, de la Dirección General de 356 Producciones y Mercados Agrarios, por la que se publica la de 22 de marzo de 2017, por la 357 que se aprueba la reglamentación específica del libro genealógico y el programa de mejora 358 de la raza equ. 359 10. Desser, H., Niebauer, G. and Gebhart, W. (1980) Polyamin- und Hisatmingehalt im Blut von 360 pigmentierten, depigmentierten und melanomtragenden Lipizzanerpferden. Zbl Vet Med 27, 361 45–53. 362 11. Meuwissen, T.H.E. and Luo, Z. (1992) Computing inbreeding coefficients in large 363 populations. Genet. Sel. Evol. 24, 305–313. 364 12. Gutiérrez, J.P. and Goyache, F. (2005) A note on ENDOG: A computer program for 365 analysing pedigree information. J. Anim. Breed. Genet. 122. 366 13. StatSoft Inc. (2007) STATISTICA (data analysis software system). www.statsoft.com. 367 14. Legarra, A. (2008) TM Threshold Model. http://acteon.webs.upv.es/. 368 15. Bowling, A. and Ruvinsky, A. (2000) The Genetics of the Horse, BCAB Inter. 369 16. Rodríguez, M., García-Barona, V., Peña, L., Castaño, M. and Rodríguez, A. (1997) Grey 370 horse melanotic condition: A pigmentary disorder. J. Equine Vet. Sci. 17, 677–681. 371 17. Seltenhammer, M.H., Simhofer, H., Scherzer, S., Zechner, P., Curik, I., Sölkner, J., Brandt, 372 S.M., Jansen, B., Pehamberger, H. and Eisenmenger, E. (2003) Equine melanoma in a 373 population of 296 grey Lipizzaner horses. Equine Vet. J. 35, 153–157. 374 http://www.scopus.com/inward/record.url?eid=2-s2.0-0038728292&partnerID=tZOtx3y1. 375 18. Rosengren Pielberg, G., Golovko, A., Sundström, E., Curik, I., Lennartsson, J., 376 Seltenhammer, M.H., Druml, T., Binns, M., Fitzsimmons, C., Lindgren, G., Sandberg, K., 377 Baumung, R., Vetterlein, M., Strömberg, S., Grabherr, M., Wade, C., Lindblad-Toh, K., 378 Pontén, F., Heldin, C.-H., Sölkner, J. and Andersson, L. (2008) A cis-acting regulatory 379 mutation causes premature hair graying and susceptibility to melanoma in the horse. Nat. 380 Genet. 40. 381 19. Hofmanová, B., Vostrý, L., Majzlík, I. and Vostrá-Vydrová, H. (2015) Characterization of 382 greying, melanoma, and vitiligo quantitative inheritance in Old Kladruber horses. Czech J. 383 19 Anim. Sci. 60. 384 20. Teixeira, R.B.C., Rendahl, A.K., Anderson, S.M., Mickelson, J.R., Sigler, D., Buchanan, 385 B.R., Coleman, R.J. and Mccue, M.E. (2013) Coat color genotypes and risk and severity of 386 melanoma in gray quarter horses. J. Vet. Intern. Med. 27. 387 21. Fleury, C., Bérard, F., Leblond, A., Faure, C., Ganem, N. and Thomas, L. (2000) The study 388 of cutaneous melanomas in Camargue-type gray-skinned horses (2): Epidemiological 389 survey. Pigment Cell Res. 13. 390 22. Valentine, B.A., Calderwood Mays, M.B. and Cheramie, H.S. (2014) Anaplastic malignant 391 melanoma of the tail in non-grey horses. Equine Vet. Educ. 26. 392 23. Weyden, L. Van Der, Patton, E.E., Wood, G.A., Foote, A.K., Brenn, T., Arends, M.J. and 393 Adams, D.J. (2016) Cross-species models of human melanoma. J. Pathol. 238, 152–165. 394 24. Smith, S.H., Goldschmidt, M.H. and McManus, P.M. (2002) A comparative review of 395 melanocytic neoplasms. Vet. Pathol. 39. 396 25. Hofmanová, B., Vostrá-vydrová, H., Majzlík, I. and Vostrý, L. (2016) The effect of 397 inbreeding on melanoma and vitiligo occurrence in Old Kladruber grey horses. 179–182. 398 26. Foley, G.L., Valentine, B.A. and Kincaid, A.L. (1991) Congenital and Acquired 399 Melanocytomas (Benign Melanomas) in Eighteen Young Horses. Vet. Pathol. 28. 400 27. Valentine, B.A. (1995) Equine melanocytic tumors: a retrospective study of 53 horses (1988 401 to 1991). J. Vet. Intern. Med. 9. 402 403 404